# Humanoid supply chain

URL: https://nader.io/projects/robotics-analysis/

What a robot is actually made of

Three tiers of it: the component makers who sell to every assembler, the critical-minerals names they get confused with, and the magnet layer in between. Priced, fact-checked, and tested to see which of them actually move together.

Compiled **20 Nov 2026** Tracked **18 names, 3 tiers**

Doc **RW·3T** Sheets **09** Venues **Tokyo · Xetra · US** Refresh **daily 07:00 UTC**

01 The parts, and who makes them

A humanoid is mostly gearboxes, motors and magnets. Read the table downwards and you go from the joint that moves to the dirt it is made of — each row naming the part, what it does, and who actually supplies it.

Six tier-one suppliers, located on the joints they sell into. Everything above the band is assembly; the band is the input.

| Part | What it does | Who makes it | Listing |
| --- | --- | --- | --- |
| JointStrain wave gear | Zero-backlash gearbox for precision joints. Twenty to forty per humanoid. | Harmonic Drive Systems; Schaeffler from 2027 | 6324.T · SHA0.DE |
| JointCycloidal RV reducer | Heavier gearing for legs and load-bearing axes. | Nabtesco | 6268.T |
| MotorServo motor & drive | Turns the joint and holds it against load. | Yaskawa Electric | 6506.T |
| MotorIntegrated actuator | Motor, gearbox and controller folded into one unit. | Nidec | 6594.T |
| JointActuator module & bearings | Complete joint assemblies, motors and electronics built in-house. | Schaeffler | SHA0.DE |
| PerceptionEdge vision SoC | On-robot sight, with no cloud round-trip. | Ambarella | AMBA |
| MaterialSintered NdFeB magnet | The rotor magnet inside every motor above. | TDK · Aichi Steel · MP Materials | 6762.T · 5482.T · MP |
| MaterialRare earth feedstock | Neodymium and praseodymium for the magnet; dysprosium and terbium to keep it working hot. | MP Materials · USA Rare Earth · Lynas | MP · USAR · LYSDY |The last two rows are shaded because they are inputs rather than parts — the magnet and the metal that goes into it. Nobody assembles a robot from them directly, which is the distinction tiers two and three exist to make.

**Latest — 20 Nov 2026.** Tokyo and Sydney use Friday 20 November closes. Xetra, New York and US OTC use Thursday 19 November. Friday is still forming in the Western venues, so the 103 completed weekly returns still end on 13 November, with the robotics basket at 0.3576 against REMX.

**Tokyo component shares fell alongside the broader market.** Harmonic Drive closed at ¥3,722 (−10.14%) on 1.26 times its prior 20-session average volume; Nabtesco at ¥3,105 (−8.03%) on 1.65 times. Yaskawa fell 8.49% and Nidec 4.97%. The Nikkei 225 fell 5.06% to 56,436.65; TOPIX trackers 1306.T and 1308.T fell 3.98% and 3.85%. TDK fell 6.91% and Aichi Steel 2.04%. Yahoo’s Tokyo daily closes are null, but its completed-session quote metadata and [dated venue histories](<https://stockanalysis.com/quote/tyo/6324/history/>) confirm the auction prices. Null bars are not unchanged sessions.

**Ambarella rose against a falling semiconductor benchmark.** Its 19 November close was $58.93 (+7.85%) on 2.01 times prior average volume, while the PHLX Semiconductor Index fell 2.41%. The issuer’s next earnings release remains [24 November after the US close](<https://investor.ambarella.com/node/17431/pdf>). Perpetua fell 8.05% to $15.47, alongside GDX (−3.77%) and GLD (−1.76%). REMX fell 2.14% to $44.32 while MP gained 1.97% and USA Rare Earth 4.73%. These are concurrent returns, with no causal attribution.

**The venue dates matter.** Lynas settled at A$12.44 (−2.74%) in Sydney on Friday after its ADR gained 1.32% to $8.46 in Thursday’s US session. Schaeffler’s Thursday Xetra daily bar gives €5.19 (−1.52%). Its 18 November bar has backfilled at the previously verified €5.27, closing the last logged German gap. The next scheduled issuer events remain Ambarella on 24 November and the Lynas AGM on 25 November. Nidec’s statutory half-year filing deadline remains 16 December.

**Latest — 19 Nov 2026.** Tokyo and Sydney use Thursday 19 November closes. Xetra, New York and US OTC use Wednesday 18 November. The 103 completed weekly returns still end on 13 November, with the robotics basket at 0.3576 against REMX.

**Harmonic Drive quantified its combined AI and humanoid category.** The [19 November presentation](<https://contents.xj-storage.jp/xcontents/AS06879/72ea9203/5839/4596/a735/21a7aa1f8a6b/140120261118582977.pdf>) (slides 20, 25 and 29–30) reports ¥1.2bn of first-half sales and approximately ¥1.5bn of orders in the combined category, with orders up 30.6% year on year. These are not humanoid-only figures. The initial full-year category sales plan is ¥2.3bn. Management targets gearhead introduction in March 2027 and actuator introduction in December 2027, and envisages full-scale mass production in 2028–2030 for its large-volume customer strategy. Those are company targets, not delivered volumes.

**Compass Diversified amended its debt covenants.** Its [18 November 8-K](<https://ir.compassdiversified.com/node/14346/html>) reports supplemental indentures executed on 16 November after receiving the required noteholder consents. They change Lugano’s subsidiary treatment and several debt and payment provisions, and add approximately $10.1m of payment-in-kind principal. The note maturities and coupon rates are unchanged. This is a parent financing development for Arnold exposure, rather than a magnet-order disclosure.

**The market moves were mixed.** Harmonic Drive rose 0.63% to ¥4,142 while Nabtesco fell 1.40% to ¥3,376. The Nikkei 225 rose 0.71% to 59,444.21. Lynas settled at A$12.79 (+3.40%) after its ADR gained 2.96% to $8.35 in the prior US session. REMX fell 1.44% to $45.29, MP gained 3.13% and USA Rare Earth fell 1.13%. Perpetua fell 1.95% while GDX fell 5.97% and GLD fell 3.20%. These are concurrent returns, with no causal attribution. Schaeffler closed at €5.27 (−0.75%) in Xetra’s auction, corroborated by [S&P Global’s venue history](<https://stockanalysis.com/quote/etr/SHA0/history/>).

**Latest — 18 Nov 2026.** Tokyo and Sydney use Wednesday 18 November closes. Xetra, New York and US OTC use Tuesday 17 November. The 103 completed weekly returns still end on 13 November, with the robotics basket at 0.3576 against REMX.

**Perpetua signed a non-binding tungsten memorandum.** Its [17 November release](<https://perpetuaresources.com/perpetua-resources-announces-non-binding-mou-with-a-major-american-aerospace-defense-company-to-evaluate-domestic-tungsten-opportunities/>) describes potential non-equity funding of up to $55.5m at the discretion of an unnamed US aerospace and defence company, with a possible right to acquire up to 50% of future tungsten production. Further work depends on definitive agreements and funding. A tungsten development would be separate from the approved Stibnite gold-antimony mine plan and would require additional review and permits. Perpetua closed 0.88% higher at $17.16 on 1.53 times its prior 20-session average volume.

**Harmonic Drive and Nabtesco fell for a third session.** Harmonic Drive closed at ¥4,116 (−2.53%), with supplementary results materials due 19 November. Nabtesco closed at ¥3,424 (−0.35%) on 1.89 times its prior 20-session average volume. The Nikkei 225 rose 0.28% to 59,024.28. Lynas rose 2.83% to A$12.37 in Sydney after its US ADR closed down 7.63% at $8.11 in the previous US session.

**REMX fell 2.36% to $45.95 in the US.** USA Rare Earth fell 6.94%, MP Materials 3.41% and Ambarella 2.71%. The PHLX Semiconductor Index rose 0.34%. Schaeffler fell 2.93% to €5.31 in Xetra’s closing auction, confirmed by the [17:35:01 CET quote](<https://es.marketscreener.com/cotizacion/accion/SCHAEFFLER-AG-176430175/cotizaciones/>) and [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>).

**Latest — 17 Nov 2026.** Tokyo and Sydney use Tuesday 17 November closes. Xetra, New York and US OTC use Monday 16 November. The latest completed correlation week ends on 13 November, with the robotics basket at 0.3576 against REMX.

**Harmonic Drive and Nabtesco fell for a second session.** Harmonic Drive closed at ¥4,223 (−4.80%). Nabtesco closed at ¥3,436 (−3.86%) on 2.08 times its prior 20-session average volume. The Nikkei 225 fell 1.34% to 58,860.43. Lynas fell 8.24% to A$12.03 in Sydney, while its ADR fell 1.90% to $8.78 in the prior US session.

**In the US on Monday, REMX fell 5.92% to $47.06.** USA Rare Earth fell 12.62%, MP Materials 6.89% and Novonix’s ADS 8.57%. Ambarella fell 8.47% to $55.03 alongside a 5.28% decline in the PHLX Semiconductor Index. Schaeffler closed at €5.47 (−1.26%) in Xetra’s auction, corroborated by the [17:35:08 CET quote](<https://de.marketscreener.com/kurs/aktie/SCHAEFFLER-AG-176430175/?iSubCodezb=190544066>) and the venue’s [daily history](<https://www.investing.com/equities/schaeffler-historical-data>).

**Agility’s amended registration statement repeats its rollout timetable.** The [16 November filing](<https://investors.agilityrobotics.com/node/7716/html>) keeps the target of first-half 2027 early access for Digit 5. It describes approximately $300m of conditional multi-year Digit 5 orders tied to 1,000 robots under a three-year service contract as of May 2026. The contract depends on meeting specified milestones or product requirements. The filing also covers potential resale of up to 20,102,500 existing PIPE shares by selling holders. Agility says it would receive no proceeds from those resales. A [separate amendment](<https://investors.agilityrobotics.com/sec-filings/sec-filing/s-1a/0001213900-26-120832>) covers potential resales of approximately 160.9m shares, including shares issuable on warrant exercise, subject to applicable lockups. It also covers issuance of up to 4,212,500 shares on warrant exercise.

**Latest — 16 Nov 2026.** Tokyo and Sydney use Monday 16 November closes. Xetra, New York and US OTC use Friday 13 November. The completed weekly correlation window still ends on 13 November and the basket remains at 0.3576 against REMX.

**Nabtesco fell 13.86% to ¥3,574** in the first session after its [third-quarter filing](<https://www.nabtesco.com/cms/wp-content/uploads/Summary_of_Consolidated_Financial_Statements_for_FY2026_Q3_e.pdf>), on 2.82 times its prior 20-session average volume. Harmonic Drive fell 7.10% and Yaskawa fell 6.07%. The Nikkei 225 closed at 59,662.02 (−2.42%) and the TOPIX trackers 1306.T and 1308.T fell 1.11% and 1.12%, respectively. Nabtesco’s full-year guidance remained in place in Friday’s release.

**Nidec’s half-year securities report is now due 16 December.** Nidec [applied for an extension](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1116-01/261116-01en.pdf>) and [announced approval](<https://www.release.tdnet.info/inbs/140120261116581515.pdf>) via TDnet at 17:00 JST. Its separate [quarterly results summary](<https://www.release.tdnet.info/inbs/140120261116581518.pdf>) was also delayed, with a release date still to be announced. Nidec had closed at ¥1,707 (−4.53%) before the approval notice.

**Sydney closed lower.** Lynas settled at A$13.11 (−0.61%) and Novonix at A$0.075 (−6.25%). The US lines in the price table still use Friday’s closes.

**Latest — 15 Nov 2026.** The weekend check confirmed Friday 13 November closes across Tokyo, Sydney, Xetra, New York and US OTC. Schaeffler’s daily bar backfilled at the previously verified auction close. The 103-return sample still ends on 13 November, with the robotics basket at 0.3576 against REMX.

**Nabtesco will time further Hamamatsu expansion to demand.** In the [13 November results Q&A](<https://www.nabtesco.com/cms/wp-content/uploads/Results_Briefing_QA_for_FY2026_Q3_e.pdf>), management said Tsu still has spare capacity and reaffirmed a ¥200m 2026 sales target for the small RVmini reducer. Customer adoption timing remains open. Nabtesco’s first Tokyo session after the results is 16 November, the same day as Nidec’s statutory filing deadline.

**GXO and Apptronik put numbers on a potential rollout.** Their [12 November letter-of-intent announcement](<https://investors.gxo.com/news-releases/news-release-details/gxo-and-apptronik-expand-partnership-scale-humanoid-deployment>) describes an initial North American deployment in late 2027 and a possible rollout of up to 1,000 Apollo robots over four years. The plan depends on meeting performance, safety and operational targets in the expanded trials. These are conditions to watch for the assembler volume thesis.

**REalloys outlined a US magnet production target.** The [13 November SEC filing](<https://www.sec.gov/Archives/edgar/data/1567900/000118518526005348/realloys8k111126.htm>) furnished an [earnings release](<https://realloys.com/realloys-press-release/realloys-reports-third-quarter-2026-results/>) outlining a design target of approximately 3,000 tonnes per year of sintered NdFeB magnets by 2030. Site, timing and capital requirements remain open, and the company expects to need significant additional financing. Its [quarterly report](<https://www.sec.gov/Archives/edgar/data/1567900/000162828026073547/aloy-20260930.htm>) says it is no longer pursuing the potential JS Link transaction.

**Latest — 14 Nov 2026.** The tables use Friday 13 November closes across all venues. Nidec flagged a filing delay on Saturday, Perpetua moved its financing target into 2027, and the weekly correlation window advanced through Friday.

**Nidec’s next statutory deadline is Monday.** Nidec says filing by the **16 November** statutory deadline will be difficult and is [considering applying for an extension to 16 December](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1114-01/261114-01en.pdf>). The half-year report would use the prior-period adjustments already reflected in the September filing. The audit disclaimer and amended prior-year reports are on a separate track. On Friday the company [cancelled the record date for the proposed extraordinary general meeting](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1113-01/261113-01en.pdf>) and [declared a zero interim dividend](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1113-02/261113-02en.pdf>). The Tokyo shares closed unchanged at ¥1,788 before the weekend notice. [JPX](<https://www.jpx.co.jp/english/listing/measures/alert/index.html>) still lists Nidec as a Security on Special Alert.

**Perpetua moved the Stibnite financing timetable.** The [13 November 10-Q](<https://www.sec.gov/Archives/edgar/data/1526243/000110465926127248/ppta-20260930x10q.htm>) puts expected completion of EXIM documentation and conditions in **Q1 2027**, after the earlier Q4 2026 target. The final investment and construction decision is also expected in Q1. The $2.9bn facility has EXIM board approval and remains subject to closing conditions. The company reported $443.4m in unrestricted cash at 30 September and expects about $198m of Q4 spending. Perpetua scheduled a [project webinar for 17 November at 11:00 EST](<https://perpetuaresources.com/perpetua-resources-announces-third-quarter-2026-financial-results-and-provides-project-update/>). The shares closed at $17.78 (−0.22%) while GDX moved −0.24% and GLD +0.16%.

**MP gained and REMX fell for a second US session.** MP closed at $38.76 (+3.50%), REMX at $50.02 (−1.69%) and USA Rare Earth at $8.16 (+1.12%). Ambarella closed at $60.12 (−1.54%); the PHLX Semiconductor Index fell 3.43%. Nabtesco’s next Tokyo session after its results is 16 November.

**All six suppliers rose over the week.** The weekly equal-weight basket returned +7.92% while REMX returned −2.68%. Advancing the 103-return window moved the coefficient from 0.3675 to **0.3576**. Schaeffler settled at €5.54 (−0.72%) on Friday, verified against [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>) while Yahoo’s daily bar is incomplete.

**Latest — 13 Nov 2026.** Nabtesco published its delayed third-quarter results after the Tokyo close and kept its full-year forecast. The price tables use 13 November in Tokyo and Sydney and 12 November on Xetra and in the US.

**Nabtesco’s Q3 precision-reducer orders rose 18% year on year and were flat quarter on quarter.** The [16:00 JST filing](<https://www.release.tdnet.info/inbs/140120261113578060.pdf>) includes KPMG AZSA’s quarterly review. Nine-month sales rose 14.4% to ¥250.813bn and operating profit rose 50.9% to ¥23.352bn. Full-year guidance stays at ¥344bn of sales and ¥32.6bn of operating profit. The [briefing](<https://www.nabtesco.com/cms/wp-content/uploads/Results_Briefing_Material_for_FY2026_Q3e.pdf>) reports 85% utilisation at Tsu, 95% at Changzhou and an ongoing production ramp at Hamamatsu. Management cited industrial demand from Chinese and South Korean automakers and gave no separate humanoid order figure. Nabtesco closed at ¥4,149 (−0.12%) before the release as the Nikkei 225 gained +0.53% to 61,142.35.

**MP Materials rose while the rare-earth ETF fell.** On 12 November MP closed at $37.45 (+1.96%), REMX at $50.88 (−4.23%) and USA Rare Earth at $8.07 (−2.36%). The [12 November USAR 10-Q](<https://www.sec.gov/Archives/edgar/data/1970622/000197062226000072/usar-20260930.htm>) confirms that November Serra Verde carbonate sales started under a 15-year take-or-pay agreement covering Phase 1 production. Lynas’s ADR closed at $9.03 (−4.34%) on 4.36 times its prior 20-session average volume, followed by A$13.19 (−1.12%) in [Sydney](<https://markets.ft.markitdigital.com/data/equities/tearsheet/summary?s=LYC%3AASX>) on 13 November. Perpetua closed at $17.82 (−5.91%) as GDX fell 3.95% and GLD fell 2.82%.

**Schaeffler closed at €5.58 (−0.89%) on 12 November.** [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>) and Yahoo’s post-auction quote agree. Yahoo’s 11 November daily bar backfilled at €5.63. The DAX rose +0.41%. The weekly basket-to-REMX figure remains 0.3675 through 6 November.

**Agility Robotics began Nasdaq trading on 5 November as AGLT.** The Digit assembler closed its [business combination with Churchill Capital Corp XI](<https://www.sec.gov/Archives/edgar/data/2074973/000121390026116898/ea031120401ex99-2.htm>) on 4 November. Its [10 November S-1](<https://www.sec.gov/Archives/edgar/data/2074973/000121390026118025/ea0311220-s1_agility.htm>) targets early access to Digit 5 in the first half of 2027. That is a customer-demand milestone to track alongside suppliers’ disclosures.

**Latest — 12 Nov 2026.** Harmonic Drive fell 4.17% in the first Tokyo session after it raised guidance. US rare-earth stocks rebounded on 11 November while gold miners fell again. The tables use 12 November closes in Tokyo and Sydney and 11 November closes on Xetra and in the US.

**Harmonic Drive closed at ¥4,785 (−4.17%) on 1.99 times its prior 20-session average volume.** The Nikkei 225 fell 0.70% to 60,819.36; the TOPIX trackers 1306.T and 1308.T moved +0.10% and −0.12%, respectively. Harmonic Drive’s [11 November filing](<https://www.release.tdnet.info/inbs/140120261106567186.pdf>) gives no separate humanoid order figure. Nabtesco’s delayed quarterly results are [scheduled for 13 November](<https://www.nabtesco.com/cms/wp-content/uploads/News-release-20261014e.pdf>); Harmonic Drive’s supplement is due 19 November and Ambarella’s call is 24 November.

**Lynas moved in opposite directions across two sessions.** Lynas closed at **A$13.34 (−4.17%)** in Sydney on 12 November after its ADR had closed at $9.44 (+2.61%) in the US on 11 November. REMX closed at $53.13 (+2.49%), USA Rare Earth at $8.27 (+4.36%) and MP Materials at $36.73 (+0.74%). Perpetua fell to $18.94 (−2.02%) as GDX moved −1.48% and GLD moved −1.13%. The PHLX Semiconductor Index rose +5.66% while Ambarella closed at $62.49 (+3.53%). [Sydney close](<https://markets.ft.markitdigital.com/data/equities/tearsheet/summary?s=LYC%3AASX>).

**Schaeffler closed at €5.63 (+2.18%) on 11 November.** Yahoo’s new daily gap was resolved for price using auction-time data and [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>). The 10 November daily bar backfilled at €5.51. The DAX rose +1.20%. The weekly sample still ends on 6 November; the REMX-to-basket correlation remains **0.3675**.

**Serra Verde reported first offtake sales in November after October shipments.** [USA Rare Earth’s separate 9 November release](<https://investors.usare.com/news-releases/news-release-details/usa-rare-earth-announces-start-commercial-scale-shipments-and>) distinguishes mixed rare-earth carbonate shipments and sales from Stillwater’s September block magnet sales. Serra Verde still targets an initial 4,000 tonnes per year of total rare-earth oxide in Q2 2027.

**Latest — 11 Nov 2026.** Harmonic Drive raised full-year profit guidance after the Tokyo close and reported a 70.7% increase in first-half orders. The filing gives no separate humanoid order figure. The market tables use **11 November** closes in Tokyo and Sydney and **10 November** closes on Xetra and in the US.

**Harmonic Drive reported ¥50.386bn of first-half orders, up 70.7% year on year.** It cited stronger semiconductor equipment orders and broke out order intake by region and product. First-half revenue rose 29.8% to ¥36.124bn, and operating profit rose from ¥465m to ¥4.205bn. The company lifted FY3/27 revenue guidance from ¥74.5bn to **¥78bn**, operating profit from ¥8.5bn to **¥10.2bn** and the annual dividend forecast from ¥20 to **¥28**. It plans to post supplementary results material on 19 November. [Half-year results](<https://www.release.tdnet.info/inbs/140120261106567186.pdf>) · [Guidance and dividend notice](<https://www.release.tdnet.info/inbs/140120261106567176.pdf>).

**Schaeffler repeated its approximately €400m humanoid order figure.** A company release dated 10 November and datelined Yokohama repeats the order figure and the two Greater China customer wins from its 20 October Taicang factory announcement. The 3 November nine-month statement described an expanding order book without giving an amount. The factory releases provide no line throughput or realised revenue figure. [20 October group release](<https://www.schaeffler.com/en/media/press-releases/press-releases-detail.jsp?id=88222851>) · [10 November release](<https://www.schaeffler.co.in/en/news-media/press_releases/press_releases_detail.jsp?id=88226241>).

**Tokyo and Sydney closed higher on 11 November.** The Nikkei 225 rose **1.31% to 61,248.59**. The two Nikkei ETFs rose 1.37% and 1.35%, and the two TOPIX trackers rose 0.83% and 0.76%. Harmonic Drive closed at **¥4,993, +3.03%** before its filing. TDK led the six tracked Tokyo listings at **¥2,365.5, +4.81%**. Lynas closed at **A$13.92, +1.61%** in Sydney after its US ADR had closed at $9.20, −0.76%. [Sydney close](<https://markets.ft.markitdigital.com/data/equities/tearsheet/summary?s=LYC%3AASX>).

**Rare earths and gold equities fell on 10 November.** REMX closed at **$51.84, −3.50%**, MP Materials at $36.46, −3.42% and USA Rare Earth at **$7.92, −9.59%** on **1.63 times** its prior 20-session average volume in the first session after its Q3 release. Perpetua closed at **$19.33, −8.65%** while GDX fell 5.43% and the GLD and IAU bullion funds fell 2.68% and 2.62%, respectively. The S&P 500, Nasdaq Composite and PHLX Semiconductor Index fell 0.38%, 0.74% and 0.35%.

**Schaeffler’s 9 November daily bar backfilled at €5.56.** The 10 November daily OHLC was null. The 17:35:25 CET quote, 17:30 five-minute bar and S&P Global’s Xetra history agree on **€5.51, −0.90%**. The DAX rose 0.31% that session. The weekly correlation sample still ends on 6 November, at **0.3675** over 103 completed weekly returns. [Xetra daily history](<https://stockanalysis.com/quote/etr/SHA0/history/>).

**Latest — 10 Nov 2026.** China extended the pause on six October 2025 export-control notices through **10 January 2027**. USA Rare Earth disclosed its first commercial Stillwater magnet sales, and MP Materials reported that it had completed GM magnet qualification. The market tables use **10 November** closes in Tokyo and **9 November** closes on Xetra and in New York.

**China published the extension on 9 November.** Joint Announcement No. 58 of 2026 continues the suspension of the October measures from **11 November through 10 January 2027**. The six notices include rare-earth items and technology, along with some battery and graphite items. The separate April 2025 licensing regime remains in force. [Joint announcement](<https://www.mofcom.gov.cn/zwgk/zcfb/art/2026/art_91466027f465467985942a639a9485f4.html>).

**USA Rare Earth sold sintered NdFeB block magnets in September.** The **9 November 16:05 ET** release identifies the customer only as a high-performance materials manufacturer and gives no magnet-only revenue figure. It also reports that Serra Verde began commercial-scale heavy mixed rare-earth carbonate shipments under its offtake in October. The operational targets put Stillwater’s second 600 tpa line and Serra Verde’s 4,000 tpa TREO run rate in **Q2 2027**. USAR closed **$8.76, +6.50%**, before the release. [USA Rare Earth Q3 results](<https://investors.usare.com/news-releases/news-release-details/usa-rare-earth-reports-third-quarter-2026-financial-results>).

**MP Materials’ 5 November results separate qualification from sales.** It completed GM qualification and expects saleable finished-magnet deliveries and associated revenue in Q4. Q3 Magnetics revenue was **$18.6m** from magnetic precursors, down 15% year over year. MP closed **$37.75, +6.61%** on 9 November. [Form 10-Q](<https://www.sec.gov/Archives/edgar/data/1801368/000180136826000057/mp-20260930.htm>).

**Other filings since the last compilation.** Compass Diversified reported Arnold Q3 sales of **$47.7m** and adjusted EBITDA of **$8.9m**, versus $37.7m and $4.1m a year earlier. Nidec renewed **¥600bn** of bank commitments on 5 November, with a March 2027 equity covenant. Perpetua reported new Stibnite drill results on 4 November, including gold and tungsten at SB597. TDK’s separate [30 October release](<https://www.tdk.com/en/news_center/press/20261030_01.html>) also confirms approximately ¥100bn for TMR sensors in Niigata, previously carried here as press-sourced. Aichi Steel disclosed [flexible bonded magnet samples](<https://www.aichi-steel.co.jp/ENGLISH/news/mt-assets/docs/20261106_news_.pdf>) on 6 November. [Compass results](<https://ir.compassdiversified.com/news-releases/news-release-details/compass-diversified-reports-third-quarter-2026-financial-results>) · [Nidec notice](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1105-01/261105-01en.pdf>) · [Perpetua release](<https://perpetuaresources.com/perpetua-announces-new-gold-discovery-at-stibnite-gold-project-continued-high-grades-from-clark-tunnel-fault-zone/>).

**Tokyo recovered for two sessions after the 6 November selloff.** The Nikkei 225 fell from **61,077.90** on 5 November to **57,434.77** on 6 November and recovered to **60,459.58** on 10 November, **+2.87%** that day. Both Nikkei ETFs rose about 3%, and the two TOPIX ETFs rose 2.09% and 2.37%. Harmonic Drive led the six tracked Tokyo lines at **¥4,846, +10.21%** on **0.96 times** prior twenty-session average volume. It remained below its 4 November close of ¥4,931.

**The 9 November New York rebound was broad.** The S&P 500 gained **2.35%**, the Nasdaq Composite **2.74%** and the PHLX Semiconductor Index **6.85%**. REMX rose **+4.51%** to **$53.72**. The Lynas ADR gained **+10.36%** to **$9.27**. Sydney settled the following session at **A$13.70, +4.18%**. Perpetua gained **+8.35%** to **$21.16** while GLD and IAU rose 1.88% and 1.83%.

**Schaeffler’s 3 November close backfilled at €5.66.** The 9 November Xetra daily OHLC was null. Yahoo’s 17:36:55 CET metadata and S&P Global’s Xetra history agree on **€5.56, +1.09%**. The 17:25 continuous print was €5.58. [Xetra daily history](<https://stockanalysis.com/quote/etr/SHA0/history/>).

**The correlation window advanced one week.** The week ended 6 November put the robotics basket down **11.27%** and REMX up **2.82%**. The rolling coefficient moved from **0.3918 to 0.3675** over 103 completed weekly returns. Both the daily-close and native weekly routes agree to six decimal places.

**Latest — 4 Nov 2026.** **Schaeffler put a €480m restructuring bill on the table on 3 November and named humanoid robotics a growth area in the same release. The market took the bill: −5.19% on Xetra, the widest fall in tier one. Tokyo then opened on 4 November, traded 4.8% lower at its worst, and closed down only 1.42%, with Harmonic Drive printing −12.3% intraday before finishing −5.25% on nearly three times its average volume.**

**The programme is the company’s own document, and it is larger than the order book it sits beside.** The release, datelined Herzogenaurach **3 November 2026** and headed “Schaeffler announces further performance improvement program”, plans a net reduction of **3,460 positions** in the Europe region, **2,660** of them in Germany, and the closure of **five sites**, four of them German: **Dortmund** (about 430 employees), **Morbach** (160), **Berlin** (270), **Schwalbach** (385, with many expected to move to Frankfurt and about 190 positions affected) and **Herstal** in Belgium (15). One-time expenses are put at approximately **€480m**, most of it booked as restructuring provisions in **2026**, against savings of approximately **€300m per year by 2030**. Implementation is scheduled primarily for **2027 and 2028**, eased by a voluntary temporary pay reduction at all levels starting with the board. The reasons given are the European automotive downturn, weak combustion demand, a slow electric ramp and competition.

**The robotics sentence is in the same release, and it is the only part of it this card is built on.** COO Dr Jochen Schröder names **humanoid robotics**, alongside defence, as a promising future growth area building on the group’s existing core-business technology. That is a named growth area and not a number, which matters because the number went missing the same morning: the **13 October** preliminary release put humanoid orders on hand at approximately **€400m**, and the validated nine-month statement published hours before this programme did not restate it. One aggregator has already divided the two, printing that the job-cut bill is **1.2 times** the €400m order book. The arithmetic is right, and the piece says plainly that the comparison is its own and not a link the company draws, so it is not an audit row. It is worth recording for what it shows: the only quantified humanoid figure on this card is now three weeks old, absent from the audited statement, and already circulating as the denominator of a ratio nobody at the company computed. The chip on the card is dated accordingly.

**A thirty-ninth German hole, and the thirty-eighth closed to the cent.** The **2 November** bar has backfilled at exactly **€5.97**, the figure published here last run, so the tally reads **thirty-nine holes, thirty-nine reconstructions, thirty-eight confirmed to the cent**. The new one is **3 November**, and it has the familiar shape: a null close with **1,564,612** shares intact, a quote-level day range of **€6.04** to **€5.62**, and the Frankfurt line carrying the identical hole, which again places the failure upstream and removes the second German daily route in the same stroke. Two routes agree on **€5.66**: regularMarketPrice stamped **17:37:33 CET** against the 17:30 auction, and a five-minute bar stamped exactly **17:30:00** at the same figure. The trap inside the trap pulled the usual way. The last continuous tick, at **17:25**, carries **€5.63**, so taking it rather than the auction would have published a **−5.70%** session instead of the **−5.19%** one that happened. previousClose was no help this run, still carrying 2 November’s €5.97 because Xetra had not reopened, which is why the rule names the auction print first and the reopened-venue route second.

**Tokyo’s 4 November is the largest intraday reversal this page has recorded, and the published figures for it disagree with each other.** The **Nikkei 225** opened **62,973.16**, fell to **60,030.10**, which is **4.84%** below Monday’s close and the first trade under 61,000 in about three months, then closed **62,188.63, down 1.42%**, recovering **3.59%** off its own low. The morning session closed **60,835.36, down 3.56%**, and that is the figure the wires carried: correctly labelled in the body of the pieces checked here, headlined without the label. Read the full day instead, because the index gave back three fifths of what the morning said it had lost. The driver named was the profitability of AI investment, led down by Kioxia, SoftBank Group and Tokyo Electron, which makes it a Tokyo story rather than a magnet one: the Philadelphia Semiconductor Index had closed **2.63% higher** the night before.

**Harmonic Drive is where the reversal cost the most and gave the most back.** It opened **¥5,114**, never traded above **¥5,142**, printed a low of **¥4,565**, which is **12.28%** below Monday’s ¥5,204 and two and a half times the index at its own low, and closed **¥4,931, down 5.25%**, **8.02%** above the low, on **3,302,900** shares against a twenty-session average of 1,237,300: **2.67 times**, the heaviest relative turnover on the page. That is the **lowest close since 23 October**, and it undoes the whole of the four-session repair recorded on this card, which had taken the line to ¥5,530 on 28 October. It now sits **10.8%** below that and **8.5%** below the ¥5,390 it printed the day before the 22 October rout, so the recovery this row was the only name on the page able to claim is gone. Nothing was filed. **Seven days to the order figure**, second-quarter FY3/27 results being dated 11 November 2026, and the order book has now been silent since 7 August.

**Two rows went to lows the reversal did not repair, and the only Tokyo gain is in the magnet layer.** **Yaskawa −1.99% to ¥3,839** is its **lowest close since 21 November 2025**, within a point of a twelve-month low, and it widens the drawdown to **51.5%**, the second-widest on the page behind Schaeffler. **TDK −2.54% to ¥2,451** on 1.18 times average volume is its **lowest close since 15 April 2026**, five sessions after a 23.7% upgrade to full-year operating profit: the tape has taken the whole of that upgrade and more, which restates this row’s standing problem rather than contradicting it, because the release that lifted the forecast still does not size magnets. **Nabtesco −2.71% to ¥4,167** came on **1.78 times** average volume, and **Nidec −1.70% to ¥1,732** on **0.52 times**, the lightest on the page, with nothing filed since 30 October. **Aichi Steel +1.32% to ¥2,375** is the only Tokyo gain and a third consecutive one, recovering a seventh of the 29–30 October fall that cost it the shallowest-drawdown title.

**The largest moves on the page are in the twelve-month column again, and they belong to the calendar.** The Western names’ 1y base stepped three sessions, from 31 October onto **3 November 2025**, and the six Tokyo lines’ from 31 October onto **4 November 2025**. Every tier-two cell improved on a session that fell. **Lynas** goes **−15.2% to −7.8%**, 7.4 points, on a base down 9.7% from $9.89 to **$8.93** against a −1.91% session. **USA Rare Earth** goes **−54.5% to −48.5%** on a base down 13.1%, $19.45 to **$16.91**. **MP** goes **−39.2% to −34.4%** on a base down 8.6%, and **Perpetua −32.8% to −29.1%** on one down 5.7%. Tokyo’s bases mostly went the other way, so **Yaskawa**’s cell falls **−7.7% to −12.2%** and **Nidec**’s **−6.4% to −12.0%** on bases up 3.0% and 4.5%, each of which is larger than its own session. The four-week column moved the same way for Tokyo: the base rolled from 2 onto **5 October**, the top of the Harmonic Drive run, so that cell reads **−26.0%** against −20.8%, and **TDK**’s **−28.9%** against −21.4% on a base up 7.8%. Read every column against its own base date.

**The only gain worth the name in the US session is in the magnet layer, and the answer to it is due tonight.** **Compass Diversified +3.00% to $12.02** on **0.76 times** average volume, 469,000 shares against 616,185, its **highest close since 17 August**, which puts the Arnold row **150.4%** higher year to date, **94.5%** over twelve months and **9.3%** below its own high, the shallowest drawdown on the page. Something was filed, and it is governance rather than magnets: an **8-K dated 3 November**, period 1 November, reporting board-approved amendments to the trust and operating agreements effective **1 November 2026**. The substantive changes raise the board quorum **from 35% to a majority of directors in office**, require the board to interview director nominees, expand the disclosure demanded of nominees and proposing shareholders, and add compliance with Rule 14a-19. For a holding company still carrying the Lugano restatement, a quorum floor moving from roughly a third to a majority is a real tightening and worth recording; it says nothing about Arnold. **Third-quarter results land after the close on 4 November** with the call at 5:00pm ET, which is where the divestiture question this row has asked since August gets its next answer.

**Tier two fell again on a session the US market rose, and for once the bullion leg explains the one name that requires it.** The **S&P 500 rose 0.36%**, the **Nasdaq Composite 0.83%** and the semiconductor index **2.63%**, while **REMX fell 1.05% to $49.73**, its **lowest close since 1 August 2025**, on 0.46 times average volume, and four of the five minerals names fell with it: **Lynas −1.91%**, **USA Rare Earth −1.64% to $8.71**, a seventeen-month low, **MP −1.36%** and **Perpetua −0.53%**. Perpetua’s card requires gold checked before a move is read as an antimony or a Stibnite event, and gold is nearly all of this one: the gold trust fell **0.66%** and the miners **1.96%**, so a 0.53% fall in a gold-and-antimony developer is the smaller number rather than the larger. Nothing was filed and nothing was published on antimony. **Novonix +4.19% to $2.24** is the largest US move on the page and the least meaningful: **8,700** shares against a twenty-session average of 19,320, or 0.45 times an average that is itself tiny. **MP’s third quarter lands after the close on 5 November.**

**One bad claim added to the audit, and it is the oldest species in it: an intraday figure filed as a day.** The AFP report on the Schaeffler programme, carried 3 November, says “Schaeffler shares fell a little over 2% in Frankfurt”, with no session label and no time. Xetra closed that session **5.19%** lower at €5.66 and the Frankfurt line **5.47%** lower, so the figure is under half the move on either German venue. It does match the opening minutes: the first five-minute bar of the session carries **€5.82**, 2.5% below Monday’s close. This is the open written up as the day. **The correlation window did not move and could not.** Week **45** opened on Monday 2 November and does not close until Sunday 8 November; it now holds two Tokyo closes and two Western ones and is still dropped as in progress, so the sample runs week **2024-45 to 2026-44** ending 30 October for a third consecutive run, the basket holds at 0.3918 against REMX, and every figure in both charts and the prose around them is carried unchanged.

**Latest — 3 Nov 2026.** **Schaeffler’s validated nine-month statement landed at 07:45 CET, fifteen minutes before this run, and the one humanoid order book on this page came back as a sentence with no number in it. Tokyo was shut for Culture Day, so all three venues share a single as-of date of 2 November.**

**The figures confirm the 13 October preliminaries, and the board reconfirmed the guidance with them on 26 October.** Nine-month revenue **€17,481m** against €17,672m, **1.1%** lower reported and **0.4%** lower currency-adjusted; **EBIT before special items €794m**, up **6.5%**, a margin of **4.5%** against 4.2%; free cash flow before M&A **−€40m** for the nine months against +€47m a year earlier, with the third quarter alone **+€260m**; net income attributable to shareholders **€11m** against −€244m after **€201m** of special items, which is **€0.01** a share against −€0.26. Net financial debt is **€5,296m** at 30 September, **2.3 times** EBITDA before special items. Guidance stands at revenue **€22.5–24.5bn**, margin **3.5–5.5%** and free cash flow **€100–300m**. E-Mobility carries the damage: revenue €3,871m, up 4.6% currency-adjusted, against EBIT before special items of **−€571m** on a **−14.8%** margin, with write-downs and provisions for lower European-carmaker volumes, mostly in China, totalling about **€240m** of which **€220m** was booked in the quarter as a special item. Greater China revenue is **7.9%** lower currency-adjusted. Bearings & Industrial Solutions, the division a robot actuator would sit in, did **€4,795m** at an **8.9%** margin.

**What the statement does not do is restate the order book, and that is the figure this page was waiting on.** The 13 October preliminary release put humanoid-robotics orders on hand at approximately **€400m**, against €350m at the half year, after two new Greater China customers in the quarter took the humanoid customer count to five. The validated statement gives the same subject one line and no figure: “**Especially in humanoid robotics, we won new customers and continued to expand our order book during the reporting period.**” No euro amount, no customer count, and still no actuator or robotics segment anywhere in the divisional table. So the only humanoid-specific order book on this page stays a three-week-old preliminary number that the audited statement declines to repeat, and the fourth dated test to pass has produced no countable unit either. Four tests, four guidance answers, no humanoid units.

**Xetra and New York both rolled two sessions, from 30 October to 2 November, and Tokyo could not roll at all.** 3 November is Culture Day, the Tokyo Stock Exchange did not open, and a 07:00 UTC run therefore finds no new Japanese close: the six Tokyo rows carry 2 November unchanged, the two Western venues catch up to the same date, and the method note’s as-of line reads one date across all three, this time by holiday rather than by weekend. **Schaeffler closed €5.97, up 2.75%** into its own statement, on **0.61 times** average volume, 633,374 shares against 1,037,632, between €5.75 and €6.04, its highest close since 29 October. The drawdown comes in **1.3 points to 50.2%**, which hands the deepest-in-tier-one place back to Yaskawa at **50.5%**.

**The German feed swallowed a thirty-eighth, and the thirty-seventh backfilled to the cent.** The **29 October** bar has filled in at exactly **€6.03**, the figure published here from four routes, so the tally is thirty-eight holes, thirty-eight reconstructions, thirty-seven confirmed to the cent, with 2 November the only one open. That bar carries open, high, low and its 633,374 shares with a **null close**, and the Frankfurt line SHA0.F has the identical hole, which again places the failure upstream of the Xetra ticker. Three routes agree on **€5.97**: regularMarketPrice stamped **17:35:06 CET** against the 17:30 auction, the five-minute bar stamped 17:30, and the hourly series ending on the same figure. One disagrees by a cent and is recorded rather than smoothed: the last continuous tick at **17:25 is €5.98**, and the 17:30 five-minute bar carries **zero volume**, so it is a continuous close rather than an auction print. The cent sits inside the bar’s own range either way and changes no cell on the page.

**The worst session on the page belongs to USA Rare Earth, and the only thing dated to it is a calendar notice.** **−7.95% to $8.855**, its **lowest close since 29 May 2025**, seventeen months, on **0.73 times** average volume, 10,101,900 shares against 13,756,665, opening $9.78 and settling three cents above its **$8.825** low. The drawdown widens to **73.0%** below the $32.80 high, second only to Novonix in tier two, and the four-week cell to **−34.8%**, the worst on the tier outright. The company’s own release that day, datelined Stillwater 2 November, sets third-quarter results for **after the close on Monday 9 November** with the call at **4:30pm ET**. That is a date rather than a disclosure: nothing about the Stillwater run-rate, Round Top or the 600 mtpa target is in it, and an eight per cent fall on below-average turnover has no company document behind it at all.

**Nor does the tape explain it, because the complex that usually drags this tier was bid and the tier fell anyway.** The gold trust rose **1.65%**, the gold miners **2.67%** and silver **2.17%**, while the equity market was only mildly lower, the **S&P 500 0.51%** and the **Nasdaq Composite 1.07%**, with the real damage in the chip index at **−3.37%**. Four of the five tier-two names fell into that: **USA Rare Earth 7.95%**, **Novonix 4.44% to $2.15** on 1.32 times average volume, **MP 3.08% to $38.35** on 0.81 times, and **Perpetua 3.00% to $16.15** on 1.20 times, against **Lynas +0.60% to $8.39**. And the fund that holds them rose: **REMX +0.54% to $50.26** on **0.57 times** average volume. A fund adding half a point while four of its five tracked producers lose three to eight is the widest version of the disagreement this section keeps recording, and no rare-earth disclosure, assessment or policy item is dated to the session on either side of it.

**Perpetua’s two legs point opposite ways for once, which settles the attribution rather than complicating it.** Its card requires bullion checked before a move is read as an antimony or a Stibnite event, and bullion went the other way: gold up 1.65% and the miners up 2.67% against a 3.00% fall in the stock. So this one belongs to its tier and not to its metal. Nothing was filed, no antimony assessment is dated to the session, and the **$2.9bn** EXIM facility remains board-approved from 21 May and **undrawn**, with the company still guiding financial close to the fourth quarter, which has five full trading weeks left in it.

**The best session on the page is in the magnet layer and belongs to the holding company rather than to a magnet.** **Compass Diversified +4.66% to $11.67** on **1.21 times** average volume, 752,600 shares against 621,730, matching its **4 September** close to the cent and so its highest since, which leaves the Arnold row **143.1%** higher year to date and **11.9%** below its own 52-week high while most of the names around it set lows. Nothing is dated to the session. Its third quarter is tomorrow, **4 November**, and the two questions on that card are unchanged: what Arnold’s adjusted EBITDA does against the roughly **50%** year-on-year gain reported for the second quarter, and whether a divestiture programme that has never named Arnold starts to. The row is still an indirect holding in a diversified group, not a magnet position.

**Ambarella fell less than its own index, which is the first session in a fortnight the bid report does not explain.** **−2.30% to $61.67** on **0.79 times** average volume, 1,203,600 shares against 1,518,300, between **$61.18** and $65.12, its **lowest close since 14 October**, leaving it **36.2%** below the $96.69 high and **12.9%** lower year to date. The chip index fell **3.37%** the same session, half again as far. Three sessions after the exclusivity report that put 3.27m shares through the tape at 2.30 times average, turnover is back below average and the exclusivity is still a report rather than a document. A semiconductor name trading with the semiconductor tape, and under it, is what this row looked like before July.

**The correlation window did not move and could not.** Week **45** opened on Monday **2 November** and runs to Sunday **8 November**; it holds all three venues’ 2 November closes and nothing else, so it is dropped as in progress, the sample still runs week **2024-45 to 2026-44** ending 30 October, and every figure on both charts is unchanged with basket-to-REMX at 0.3918. The cross-check was rerun anyway, against native weekly bars bucketed by exchange-local ISO week with each timestamp converted using the offset in force on its own date, and it agrees to **four decimal places** on all fourteen names, on both axes and on the basket: a **fifth** consecutive clean run. None of the eleven 52-week highs moved either, so every change in the drawdown column this run is price.

**The four-week columns came in a point, and two thirds of it is the calendar.** REMX’s cell goes **−22.2% to −20.5%** and the basket’s **−14.6% to −13.9%**, so the gap narrows from **7.6 to 6.6 points**. Decomposed: REMX improved 1.7 points, of which **1.3** is its base rolling 30 September’s **$64.26** onto 2 October’s **$63.21** and **0.4** the price; the basket improved 0.7 points, **0.6** base and **0.1** price. Net, **0.7** of the point is a date change and **0.3** is a price. Only two bases could move at all, because Tokyo had no new close, and inside the basket they pull opposite ways: Schaeffler’s base rolled €6.39 onto **€6.23** while its price rose, taking its own four-week cell from −9.1% to **−4.2%**, nearly five points and on its own most of the basket’s improvement, against Ambarella’s base rolling $69.51 onto **$68.65** while its price fell, worsening its cell from −9.2% to **−10.2%**. Read the bases before reading convergence.

**One bad claim added to the audit, and it is an aggregator labelling a date it cannot have got from the company as verified.** bullstory.io carries USA Rare Earth’s third quarter as **5 November 2026** under the heading “Upcoming verified earnings release” and the subheading “Verified data”. The company’s own release of **2 November** sets it at **9 November**, four days later. A second tracker carried **13 November** and a third **5 November** on the same morning, and one of them, stockanalysis.com, has since corrected to 9 November. Guessing a date is ordinary; printing the guess under the word verified is the fault, and it is the species this audit keeps catching: a figure true of nothing in particular, published as though it were sourced.

**Latest — 2 Nov 2026.** **Monday clears the two rows later sessions had already overtaken, and the session it clears them with is the worst the minerals tier has had on this page: Friday’s AI-trade crack took REMX to a fifteen-month low and four of the five tier-two names down between 6 and 9%. Tokyo, trading after all of it, barely flinched.**

**The driver was not rare earths, and the index tape says so plainly.** The **Nasdaq 100 closed 28,384.85, down 3.46%**, **QQQ 3.47% to $690.49**, the **S&P 500 1.82%** and the **Dow 1.09%**, with **Intel −28.58%** on its guidance cut. High beta fell further, and tier two is high beta: **USA Rare Earth −8.56% to $9.62** on 1.17 times average volume off an intraday low of $9.19, **Lynas −7.85% to $8.34** on 0.76 times, **Perpetua −7.81% to $16.65** on 1.33 times, **Novonix −6.25% to $2.25** on 2.78 times (49,200 shares against a twenty-session average of 17,680), **REMX −5.12% to $49.99** on 1.10 times, and **MP Materials −4.97% to $39.57** on 1.02 times, the mildest of the five. The benchmark figure is the one worth keeping: **$49.99 is REMX’s lowest close since 1 August 2025**, fifteen months, and it leaves the fund **55.2%** below its $111.55 high. In tier one, **Schaeffler −3.65% to €5.81** and **Ambarella −2.26% to $63.12**, the session this page reported last run without pricing. **CODI −1.15% to $11.15.** None of the six carries a filing: checked directly, MP, Perpetua and CODI have lodged nothing with the SEC since 28 October, and Ambarella’s only 30 October filing is the third-party Schedule 13G/A this page already logged.

**Perpetua’s card requires both legs checked before a move is attributed to it, and this time the bullion leg is nearly all of it.** Gold futures fell **3.29% to $3,898.90** and GLD **3.47%**; the miners went further, **GDX −5.21%**. Nothing was published on antimony and nothing was filed by the company. A 7.81% fall in a gold-antimony developer on a day the metal fell 3.3% and its sector proxy 5.2% is the gold leg plus beta, not a Stibnite event, and the EXIM financial close this page has watched since May remains board-approved and undrawn with no closing announcement.

**Tokyo traded after every bit of that and took almost none of it.** The **Nikkei 225 closed 63,081.86, down 0.33%**, and the two TOPIX trackers this page checks fell **0.82%** (1306.T) and **0.34%** (1348.T), so there was no Japanese second leg. **Harmonic Drive rose 0.79% to ¥5,204** on **1.35 times** average volume, 1,602,500 shares against 1,188,580, between ¥5,088 and ¥5,263, its first gain after two falls. **Aichi Steel +0.26% to ¥2,344** on 1.89 times, its first gain after three consecutive falls and the only other riser. **Nabtesco −0.65% to ¥4,283** on 1.32 times, a third consecutive fall. **Yaskawa −1.88% to ¥3,917** on 0.77 times: a third consecutive fall and its **lowest close since 26 November 2025**, which puts it **50.5%** below its high and makes it the deepest drawdown in tier one. **Nidec −2.17% to ¥1,762** on **0.30 times** average volume, 4,344,300 shares against 14,288,120, a second consecutive fall on turnover its own October sessions make look thinner than it is. **TDK −2.22% to ¥2,515**, a third consecutive fall and its **lowest close since 15 April 2026**, having opened ¥2,611, traded as low as **¥2,436.50**, 5.3% down, and recovered more than half of it by the auction.

**One company filed today, and it is the one whose own bid has been in the market all month.** Nabtesco disclosed on **2 November** that its buyback is **complete**: **3,306,600 shares for ¥14,999,799,800** between 3 August and **30 October**, against a 31 July board authorisation of up to 4 million shares and up to **¥15bn**. It spent **99.999%** of the cash cap and took **82.7%** of the share cap, finishing a month before the 30 November deadline, and October alone accounts for **1,445,500 shares for ¥6,422,216,300**, an average of **¥4,443**. All of it is to be **cancelled on 15 December 2026**, which retires about **2.8%** of the shares outstanding. The mechanical reading is the one to keep: 2 November is the first session since 3 August with no bid from the company itself underneath the stock, and it closed 0.65% lower on 1.32 times average volume.

**And the number this page watches on that name now has a date, which moved.** Nabtesco’s third quarter for the year ending December 2026 was due **30 October**; a notice of **14 October** put it at **13 November 2026**. The reason is the governance thread rather than the business: having filed amendments to its internal control reports for prior years stating material weaknesses that should have been disclosed, the company must attach an **independent auditor’s review report** to the quarter, and has allowed the auditor the time to issue one. So 13 November is both the first utilisation read since the RV capacity doubling and the first Nabtesco quarter to arrive with a review opinion attached. Harmonic Drive’s half year is the other near date: the aggregators carry **11 November**, while the company’s own calendar publishes only the October-to-December window, so this page treats the day as reported rather than disclosed until a notice says otherwise.

**Novonix filed its September quarter on 30 October, and both halves of it land on this card’s watch line.** The Appendix 4C puts cash at **US$33.652m** at 30 September against a net operating outflow of **US$9.765m** for the quarter, which the form’s own item 8.5 turns into **3.4 quarters of funding available**. That figure counts operating cash only. Add the quarter’s **US$9.169m** of investing outflow, which is the Riverside build and not optional, and the same arithmetic gives **1.8 quarters**; take out the second **US$7m** Yorkville payment made in October, after the quarter end, and the December quarter opens nearer **US$26.7m**. Facilities are **US$58.1m and fully drawn**, with **nothing unused**: a US$30.1m PNC real estate loan and US$28m of Yorkville debentures outstanding at 30 September, since reduced to US$21m. Customer receipts for the quarter were **US$582,000**. The quarterly carries no going-concern statement of its own; the half-year’s material uncertainty stands behind it unamended.

**The second half is a contract, and it is a downgrade this page was not carrying at all.** The November 2024 **binding offtake** with PowerCo, a minimum **32,000 tonnes over five years from 2027**, has been replaced by an **Amended and Restated Supply Framework Agreement**. In the company’s own words the parties have “**removed the previous volume, pricing, and delivery commitments**” and will “negotiate in good faith the terms of a new offtake agreement (including volume and price)” once qualification and certification are done; the St Thomas, Ontario plant the tonnage was for has had **production deferred from 2027 to 2029**. Riverside took **ISO 9001:2015** in October, which the filing is careful to say “does not itself constitute customer product qualification”; the initial C-sample still has specification parameters outstanding; and mass-production sales stay guided to the **second half of 2027**. A company with under two quarters of funding on the fuller measure has swapped its only binding volume commitment for an agreement to agree, and that is a larger change to this card than any price on it.

**The correlation window moved for the first time in four runs, and the figure barely did.** ISO week **44** closed on Sunday **1 November**, so the sample runs week **2024-45 to 2026-44** and ends at the **30 October** close in every venue, which is the rare case of one date across all three. Basket-to-REMX goes 0.3949 to 0.3918. The individual figures move no further: Lynas **0.64**, MP 0.57, Perpetua 0.53, Schaeffler 0.47, USA Rare Earth 0.38, Novonix 0.37, Nidec 0.35, Yaskawa 0.26, Nabtesco **0.24**, Ambarella 0.20, Harmonic Drive 0.19, with only Lynas and Nabtesco changing in the second decimal and each by one point. The cross-check against native weekly bars bucketed by exchange-local ISO week, each timestamp converted with the offset in force on its own date, agrees to **four decimal places** on all fourteen names, on both axes and on the basket: a fourth consecutive clean run, and the first computed on a window that actually moved. A week that was violent in price and inert in correlation is now the fourth this page has recorded, which is the argument of this whole section arriving as data rather than as a claim.

**One of the eleven 52-week highs moved, and it is Novonix for a third consecutive run.** Its high falls **$16.40 to $16.15**, down 1.5%, as the window rolled two days with the request date. Its drawdown widens to **86.1%**, of which **0.9 points is the price** and **0.2 the base pulling back the other way**, and it stays the deepest on the page by fifteen points over USA Rare Earth’s 70.7%. The other ten are unchanged, so every other change in that column is price, and ten of the eleven widened. **Harmonic Drive is the single exception**, coming in 0.5 points to **43.1%**, because it is the only one of the eleven that rose.

**Two bad claims added to the audit, both about Friday, and both the same fault wearing different clothes: a figure true of some other moment, published as the day’s.** The first is the aggregator habit this page has now caught three times. A 24/7 Wall St piece stamped **2:08pm ET on 30 October** has USA Rare Earth “at $9.52, down 10%” and MP Materials “at $39.21, down 6%”; by the bell USA Rare Earth closed **$9.62, down 8.56%** and MP **$39.57, down 4.97%**, so both the headline percentage and both prices overstate the session, and neither price is a close, an open or a low. The second is new in kind and worse in degree: a tech-insider.org piece dated 30 October reports the **Dow at “52,342.56, down just 0.25%”** and the **Nasdaq Composite at “26,184.72, down 1.38%”** as Friday’s closes. Both are **29 October** to the decimal, index level and percentage together. Friday’s actual closes were the **Dow 51,770.31, down 1.09%** and the **Composite 25,437.00, down 2.86%**, so the piece understates the session by 0.8 and 1.5 points on a day whose whole interest is its size. Its Intel figure, **−28.8%**, is genuinely Friday’s (−28.58%), and its Amazon figure, “4.4% during Friday’s session”, is Thursday’s (−4.25% against Friday’s −1.41%). A mixed-date article is harder to catch than a stale one, because one true figure makes the rest read as checked.

02 Tier one — the component layer

| Company | Layer | Last | 1d | 1m | YTD | 1y | vs 52w high | 1y trend |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Harmonic Drive Systems6324.T · ADR HSYDF | Strain wave gears | ¥3,722 | −10.1% | −30.9% | −1.5% | +30.4% | −59.3% |  |
| Nabtesco6268.T · ADR NCTKY | Cycloidal RV reducers | ¥3,105 | −8.0% | −28.7% | −17.2% | −5.1% | −49.2% |  |
| Yaskawa Electric6506.T · ADR YASKY | Servo motors & controllers | ¥3,180 | −8.5% | −24.4% | −33.1% | −17.9% | −59.8% |  |
| Nidec6594.T · ADR NJDCY (thin OTC) | Motors & integrated actuators | ¥1,644 | −5.0% | −6.7% | −22.9% | −16.8% | −44.1% |  |
| SchaefflerSHA0.DE · ADR SFFLY | Actuator modules & bearings | €5.19 | −1.5% | −11.9% | −37.9% | −19.2% | −56.7% |  |
| AmbarellaAMBA · Nasdaq | Edge AI vision silicon | $58.93 | +7.9% | −6.8% | −16.8% | −32.6% | −38.8% |  |Prices in local listing currency at each venue’s last completed close, dated in the method note. Percentage moves are computed from that same local line, so they are clean of currency translation. ADR tickers are given for access, not for pricing — see the method note.

Joint 20–40 units per humanoid

### Harmonic Drive Systems

6324.T

**19 November presentation.** **Harmonic Drive quantified its combined AI and humanoid category.** The [19 November presentation](<https://contents.xj-storage.jp/xcontents/AS06879/72ea9203/5839/4596/a735/21a7aa1f8a6b/140120261118582977.pdf>) (slides 20, 25 and 29–30) reports ¥1.2bn of first-half sales and approximately ¥1.5bn of orders in the combined category, with orders up 30.6% year on year. These are not humanoid-only figures. The initial full-year category sales plan is ¥2.3bn. Management targets gearhead introduction in March 2027 and actuator introduction in December 2027, and envisages full-scale mass production in 2028–2030 for its large-volume customer strategy. Those are company targets, not delivered volumes.

**20 November.** Harmonic Drive closed at ¥3,722 (−10.14%) on 1.26 times its prior 20-session average volume. Yahoo’s daily close is null; its 15:30 JST quote metadata supplies the completed auction close.

**19 November.** Harmonic Drive closed at ¥4,142 (+0.63%) on 0.57 times its prior 20-session average volume.

**18 November.** Harmonic Drive closed at ¥4,116 (−2.53%) on 0.90 times its prior 20-session average volume.

**17 November.** Harmonic Drive closed at ¥4,223 (−4.80%) on 0.86 times its prior 20-session average volume.

**16 November.** Harmonic Drive closed at ¥4,436 (−7.10%) on 0.66 times its prior 20-session average volume. Supplementary materials are due 19 November.

**13 November.** Harmonic Drive closed at ¥4,775 (−0.21%) on 1.11 times its prior 20-session average volume. Its 11 November results showed first-half orders up 70.7%. Supplementary materials are due 19 November.

The closest thing to a monopoly on the list. Strain wave gears are the compact, zero-backlash gearboxes inside precision joints, and there is no substitute at volume. The ~85% share claim is consistent with how the market is generally described, and the humanoid order book is real but still small: roughly **¥2.5bn of humanoid-related orders** guided for FY3/26, which management has suggested could double or triple in FY3/27.

The catch is that today's profit does not come from humanoids. [FY3/26 results](<https://finance-frontend-pc-dist.west.edge.storage-yahoo.jp/disclosure/20260513/20260513529667.pdf>) show **¥59.6bn revenue (+7.0%)** and ¥2.567bn in operating profit, up from ¥6m in FY3/25. The recovery is now visible, though: April–June revenue of **¥16.68bn (+23.6%)** and operating profit of ¥1.84bn against ¥122m a year earlier, on consolidated orders of ¥24.13bn (+55.7%) — roughly ¥3bn of which was a one-off North American surgical-robot order. On 7 August the company lifted FY3/27 guidance to **¥74.5bn revenue and ¥8.5bn operating profit**, from ¥68bn and ¥6.2bn. The 11 November revision supersedes that forecast.

Share **~85%** FY3/26 rev **¥59.6bn** H1 orders **+70.7%** FY3/27e op **¥10.2bn**

**21 October: +2.47% to ¥5,390 on 1.32 times average volume, the largest rise in tier one and a second consecutive session above its own average.** The Nikkei rose 0.85% the same session, so this is roughly three times the market, and it takes the line back off the ¥5,260 it closed at the day before. The twelve-month cell gained only a point against that, +64.4% to **+65.6%**, because the year-ago base rolled again, ¥3,200 to **¥3,255**: price put four points in and the calendar took nearly three straight back out. Separately, the competitive position now has a date on it. Schaeffler’s **13 August 2026** release says it has completed validation testing of _formed_ strain wave gearboxes for humanoid robots and plans mass manufacture from **2027**, claiming manufacturing cost more than **25%** lower and material consumption more than **75%** lower by forming rather than cutting. Those are company claims and are not verified here, and 2027 is not now. But it is the first dated second source for the exact product this card is built on, it comes from inside tier one, and the competition is on cost rather than capability. The parts table now names both.

**Watch**Humanoid-specific orders. The November results report consolidated orders by region and product. The company plans to post supplementary materials on 19 November. The September tape ran eleven sessions from the 4 September close of ¥5,730: +5.5%, +4.2%, −4.5%, +2.1%, +2.7%, −3.2%, +0.7%, +0.3%, +2.1%, 0.0%, then **+3.9% to ¥6,200** on the Bank of Japan's rate decision. Tokyo then shut for three sessions, and the twelfth came on the reopen: **+2.4% to ¥6,350** on **24 September**, its highest close since **17 August** and 10.8% above where the run started. The thirteenth ended it, and quietly: **−0.8% to ¥6,300** on **25 September**, on a session the Nikkei rose **1.30%** — so the stock gave up two points of relative ground without a word from the company, having gained two or three the day before on the same silence. The fourteenth took it back and then some: **+1.6% to ¥6,400** on **28 September**, its highest close since **17 August**, on a day the index fell 0.73% and the name two rows below it closed limit-down. None of the run came from the order book. Two figures in this row have moved a long way on the calendar rather than on the tape, and both need reading carefully. The month reads **+12.2%** against +0.2% a fortnight ago, because the four-week base has rolled onto 28 August's **¥5,750**, from the ¥5,960 of 18 August. And the twelve-month column _fell_ from +116.9% to +95.9% on 18 September, because 18 September 2025's **+15.0%** day dropped out of the window — then _jumped_ to **+123.0%** on 24 September, twenty-seven points on a 2.4% session, because a three-day holiday rolls the base six calendar days in one step, onto 24 September 2025's **¥2,847**, ten per cent below the ¥3,165 it replaced. A fifteenth session added **0.8% to ¥6,450** on **29 September**, a fresh high since 17 August on a day the index fell 0.60%, so a point and a half of relative gain on the same silence. The sixteenth gave a little of it back: **−0.5% to ¥6,420** on **30 September**, on a session the Nikkei rose **1.94%** — two and a half points of relative ground surrendered, again with nothing published. The seventeenth was the largest of the lot and the first to clear the August range: **+3.0% to ¥6,610** on **1 October**, its **highest close since 13 August**, on the session the Nikkei rose 3.30% — so most of it was index rather than name, which has been true of this row for a fortnight. The eighteenth gave a little back, **−0.6% to ¥6,570** on **2 October**, against an index down 0.94%, and the nineteenth took it to a new high for the run: **+1.4% to ¥6,660** on **5 October**, its **highest close since 13 August**, on an index up 2.40%, so again most of it is index rather than name. Nineteen sessions, ¥5,730 to ¥6,660, **16.2%**, and not one of them an order disclosure. **Then the run ended, and three sessions later so did most of what it had earned.** The twentieth session broke it, **−0.9% to ¥6,600** on **6 October**; **7 October** added back **+0.45% to ¥6,630**; and **8 October** took **−7.09% to ¥6,160**, the worst session this page has recorded on the name, on **1.20m shares** against 580,400 — a little over twice the volume, opening ¥6,510, never trading above ¥6,520, and settling ¥30 off a low of ¥6,130. That is the **lowest close since 17 September** and it undoes roughly half of a nineteen-session, 16.2% climb in one go. The symmetry is the point: not one of the nineteen up sessions carried an order disclosure, and neither does this one. A row that rose on index beta and sentiment has now fallen on the same thing, five times as fast as the index did. The columns moved further than the price. The four-week cell reads **+3.0%** against +13.9% last run, and the twelve-month cell **+108.8%** against +137.9% — a **29.1-point** fall, the largest this page has recorded in that column, of which the session is 16.9 points and the year-ago base stepping one session, up 5.9% onto 8 October 2025's **¥2,950**, is the other 12.2. At a gain of this size a base step is heavily levered; read it against its own date rather than as news. The drawdown is the figure no base can flatter, and it went the other way with the price, widening five and a half points to **32.7%** below the ¥9,150 high. **Friday recovered a little of it and Tuesday took much more than that back.** **9 October** added **+0.97% to ¥6,220**; then Tokyo shut for Sports Day, and the reopen on **13 October** fell **4.66% to ¥5,930** on **1.55m shares** against 791,000, close to twice the volume, its **lowest close since 15 September**. Against a Nikkei down 1.82% that is two and a half times the index, and it carries no disclosure either: the order book has still said nothing since 7 August. The drawdown widens to **35.2%**, the four-week cell comes in to **+2.4%** from +7.4%, and the year-to-date to **+56.9%**. Twenty-three sessions on from the ¥5,730 close of 4 September, a run that peaked at **¥6,660** on 5 October has surrendered **¥730** of its ¥930 gain, leaving the name **3.5%** above where the run began, and not one session in either direction carried an order number. **Two sessions have now taken out the rest of it.** 14 October eased **1.52% to ¥5,840**, and **15 October fell 11.64% to ¥5,160** in the CPI selloff, on **1.71m shares** against a twenty-session average of 935,580: its **worst session since 28 July** and its **lowest close since 1 May 2026**, and at roughly **2.4 times** the index the hardest-hit name in Tokyo that day. That puts it **9.9% below** the ¥5,730 the run started from, so a nineteen-session, 16.2% climb and everything after it has been given back in full. The drawdown widened **7.4** points to 43.6%, the widest this page has recorded on the name. **16 October took part of it back, and on lighter volume than the fall:** **+3.68% to ¥5,350** on **778,100 shares**, **0.85 times** the twenty-session average, which brought the drawdown in to 41.5%. **19 October gave a little of it back, −0.93% to ¥5,300** on **1.21 times** average volume, after trading as low as ¥5,010. That leaves the drawdown **42.1%**, the year to date **+40.2%** and the four-week cell **−14.5%** on a base that has rolled to 18 September's **¥6,200** from 16 September's ¥5,970, which is most of the four-point move in that cell rather than anything the price did. The twelve-month cell reads **+75.2%** against a base of ¥3,025, itself rolled onto 17 October 2025. A bounce on four fifths of average turnover is not the same event as a fall on nearly twice it. Still nothing from the order book since 7 August. **26 October was the largest session on the page and it has a date behind it rather than a cause.** **+3.98% to ¥5,090** on **1.16 times** average volume, its **highest close since 21 October**, on a day the Nikkei fell **0.80%** to its lowest close since 30 July: nearly five points of relative gain, with nothing filed. The drawdown narrows **2.1** points to **44.4%** and the year to date reads **+34.7%**. The twelve-month cell is the instructive one and it did almost nothing — **+63.60% to +63.67%** — because the year-ago base rose 3.94% onto 24 October 2025’s **¥3,110** while the price rose 3.98%, so the two cancelled to seven basis points on the biggest move of the run. The four-week cell improved to **−19.2%** as Tokyo’s base finally cleared the September holidays onto 25 September’s ¥6,300. **The wait now has an end date:** second-quarter FY3/27 results are scheduled for **11 November 2026**, which is where the quarterly humanoid order figure this line has asked for since August either appears or does not. **27 October added a third consecutive gain and the largest of the three.** **+4.91% to ¥5,340** on **1.09 times** average volume, its **highest close since 21 October**, against a Nikkei up 1.73%: a little over three points of relative gain, with nothing filed. Three sessions on from the rout this row sits **9.54%** above its 22 October close, which does not quite undo it: a 9.55% fall needs a 10.56% rise, and the line is still **0.93%** below the ¥5,390 it printed the day before. It is nonetheless the only name on the page to have made back nearly all of the worst Tokyo session this page has recorded, and the magnet layer two tiers below has made back none of it. The drawdown comes in **2.8** points to **41.6%** and the year to date reads **+41.3%**. The twelve-month cell is the one that moved: **+63.7% to +74.8%**, eleven points, because this time the year-ago base fell while the price rose rather than cancelling it exactly as it did last run. Fifteen days to the order figure. **28 October made it four, and this row has now fully undone the rout.** **+3.56% to ¥5,530** on **0.69 times** average volume, its **highest close since 14 October**, against a Nikkei up 1.08%: nearly two and a half points of relative gain on the lightest turnover of the four-session run, with nothing filed. The line sits **13.44%** above its 22 October close and, more to the point, **2.60%** above the ¥5,390 it printed the day before the rout, so the 9.55% fall is repaid in full and the recovery is no longer partial. It is the only name on the page that can say that. The drawdown comes in **2.0** points to **39.6%**, under forty for the first time since 15 October, and the year to date reads **+46.3%**. The twelve-month cell gains a further **10.0** points to **+84.8%**, of which 6.2 is the session and 3.8 a year-ago base eased ¥3,055 to ¥2,992. Four sessions, ¥4,875 to ¥5,530, and not one of them an order disclosure. **Fourteen days to the order figure.** **29 October ended the run, and the repair holds anyway.** **−0.90% to ¥5,480** on **2.01 times** average volume, 2,154,200 shares against 1,073,585, the heaviest relative turnover in tier one, against a Nikkei up **1.10%** to a fifth consecutive gain: so two points of relative ground given back on twice the usual volume, with nothing filed. The line still sits **1.67%** above the ¥5,390 it printed the day before the 22 October rout and **12.41%** above the rout’s own close, so the recovery recorded above is intact rather than retraced. The drawdown widens half a point to **40.1%**, back over forty, the year to date reads **+45.0%** and the twelve-month cell eases to **+84.5%** on a year-ago base stepping ¥2,992 to **¥2,971**, so nearly all of that four-tenths is the session rather than the calendar for once. **Thirteen days to the order figure**, and the order book has now been silent since 7 August. **4 November erased the repair in a single session, and the intraday print was worse than the close by seven points.** **−5.25% to ¥4,931** on **3,302,900** shares against a twenty-session average of 1,237,300, **2.67 times** and the heaviest relative turnover on the page, after opening ¥5,114 and trading as low as **¥4,565**, which is **12.28%** below Monday’s close and two and a half times the Nikkei at its own low of −4.84%. The close is **8.02%** above that low and the **lowest since 23 October**. It puts the line **10.8%** below the ¥5,530 of 28 October and **8.5%** below the ¥5,390 printed the day before the 22 October rout, so the only full recovery on this page is no longer one. The drawdown widens to **46.1%**, the four-week cell to **−26.0%** on a base rolled onto 5 October’s ¥6,660, the top of the run, and the twelve-month cell eases to **+69.3%**. As of 4 November the half-year filing was seven days away.

Joint Legs, load-bearing axes

### Nabtesco

6268.T

**20 November.** Nabtesco closed at ¥3,105 (−8.03%) on 1.65 times its prior 20-session average volume. Yahoo’s daily close is null; its 15:30 JST quote metadata supplies the completed auction close.

**19 November.** Nabtesco closed at ¥3,376 (−1.40%) on 1.36 times its prior 20-session average volume.

**18 November.** Nabtesco closed at ¥3,424 (−0.35%) on 1.89 times its prior 20-session average volume.

**17 November.** Nabtesco closed at ¥3,436 (−3.86%) on 2.08 times its prior 20-session average volume.

**16 November.** Nabtesco closed at ¥3,574 (−13.86%) on 2.82 times its prior 20-session average volume in the first session after its results. The Nikkei 225 fell 2.42%.

**13 November.** Nabtesco closed at ¥4,149 (−0.12%) before its 16:00 JST [quarterly filing](<https://www.nabtesco.com/cms/wp-content/uploads/Summary_of_Consolidated_Financial_Statements_for_FY2026_Q3_e.pdf>). The company held its full-year sales and operating profit guidance at ¥344bn and ¥32.6bn, respectively.

**13 November results Q&A.** [Management](<https://www.nabtesco.com/cms/wp-content/uploads/Results_Briefing_QA_for_FY2026_Q3_e.pdf>) said Tsu has spare capacity and that further expansion at Hamamatsu will depend on demand. It kept the 2026 RVmini sales target at ¥200m. Customer adoption timing remains open.

The heavy-duty counterpart: cycloidal RV reducers for medium-to-large joints, with roughly 60% global share. Of the six, this is the one where the cited financials check out cleanly — FY2025 delivered **¥307.9bn revenue (+9.8%)** and **¥20.7bn operating profit (+60.3%)**, with FY2026 guided to ¥344bn and ¥32.6bn.

The [Q3 briefing](<https://www.nabtesco.com/cms/wp-content/uploads/Results_Briefing_Material_for_FY2026_Q3e.pdf>) lists annual precision-reducer capacity of 1.15m units across Tsu, Changzhou and Hamamatsu. Component Solutions recorded ¥72.949bn of nine-month orders (+23.6%), ¥69.780bn of sales (+21.3%) and ¥7.425bn of operating profit (+90.6%).

Share **~60%** FY25 rev **¥307.9bn** Op profit **+60.3%** FY26e op **+55.6%**

**2 November: the buyback is finished, and the disclosure landed the same morning.** Nabtesco told the exchange that it had completed the programme resolved on 31 July 2026: **3,306,600 shares for ¥14,999,799,800**, bought on market between 3 August and **30 October** against caps of 4 million shares and **¥15bn**. That is **99.999%** of the cash cap and **82.7%** of the share cap, finished a month inside the 30 November window, with October alone taking **1,445,500 shares for ¥6,422,216,300** at an average of **¥4,443**. All of it is to be **cancelled on 15 December 2026**, retiring about **2.8%** of the shares outstanding, so this is a genuine reduction in the count rather than treasury parking. The mechanical consequence is dated and worth holding: **2 November is the first session since 3 August with no bid from the company itself underneath the stock**, and it closed **−0.65% at ¥4,283** on 1.32 times average volume, a third consecutive fall. A company that has just spent ¥15bn of its own cash in a quarter it will not report until 13 November is making a statement about capital allocation, not about utilisation; the utilisation number is still the one this card is waiting for.

**21 October: +1.49% to ¥4,357 on 0.80 times average volume, recovering two fifths of the previous session’s fall.** Nothing is dated to it. The pattern this card keeps recording holds in both directions: the largest fall on the page one session, a partial bounce the next, each on roughly four fifths of normal turnover. The volatility this line is known for is still there; the participation is not.

**Watch**Watch whether Hamamatsu’s production ramp raises output and whether orders resume sequential growth. The [13 November briefing](<https://www.nabtesco.com/cms/wp-content/uploads/Results_Briefing_Material_for_FY2026_Q3e.pdf>) shows Tsu holding at 85% utilisation and Changzhou moving from 100% to 95%. Nabtesco says deliveries of RVmini began in the second half as planned.

**Earlier trading.** It has been the most volatile name on the tier, and has now gone quiet: −5.7% on 2 September, then +1.6%, +1.4% and +3.6%, then −6.1% on 8 September and +4.5% on 9 September — and since then −0.7%, −1.3%, −0.3% and +0.2% — four sessions inside 1.5% after a fortnight of violent ones — before **+2.6% to ¥4,551 on 16 September**, its best session since 9 September, then +0.7% to ¥4,581 and +0.8% to ¥4,619 on 18 September — three gains running — and, after the three-day holiday, a fourth: **+3.0% to ¥4,759 on 24 September**, its best session since 9 September. A fifth followed, **+0.7% to ¥4,792** on **25 September**, then the streak broke at −0.2% and resumed at +0.3%, and then the largest session of the sequence arrived on the last day of the quarter: **+2.6% to ¥4,919** on **30 September**, its highest close since **17 August** and its best session since 24 September, on an index up 1.94%, so for once the move is market rather than name, and the name still led it by seven tenths. Eight of the last ten sessions higher is the longest run this name has put together in the period this page has tracked, and none of it is company news. Two more sessions have followed and both went with Tokyo rather than ahead of it: **−1.0% to ¥4,926** on **2 October** and **+1.1% to ¥4,979** on **5 October**, the latter its **highest close since 12 August**. Then the CPI session took it with everything else: **−7.98% to ¥4,441** on **15 October**, on **1.73m shares** against a twenty-session average of 895,355, its **lowest close since 15 September**. **16 October recovered a little of it, +1.06% to ¥4,488** on 0.88 times average volume, and **19 October gave that back, −0.67% to ¥4,458** on 1.22 times. It now sits **27.1%** below its 52-week high, still the tightest on the tier, and the four-week cell reads **−3.5%** on a base that has rolled again, onto 18 September's **¥4,619** from 16 September's ¥4,551. Read that the same way as Harmonic's: the base moved, and nothing the company said moved any of these numbers. The same holds on **26 October**, more sharply: the price rose **0.53% to ¥4,140** on 1.25 times average volume and the four-week cell _fell_ 2.8 points to **−13.6%**, because the base finally rolled off 18 September onto 25 September’s **¥4,792**. The twelve-month cell lost 2.3 points to **+6.9%** the same way. A rising session that worsens two columns is the clearest case this card has for reading both against their own dates. Nabtesco still carries the tightest drawdown in tier one, now **29.9%** after **27 October added 3.50% to ¥4,285** on 1.13 times average volume, its highest close since 21 October and roughly twice the index. The twelve-month cell gained 2.3 points to **+9.2%** and the four-week cell improved to −10.6%, both on bases that moved with it rather than against it for once. Nothing filed. **28 October: +3.15% to ¥4,420 on 1.06 times average volume, a third consecutive gain and its highest close since 19 October.** Nothing is dated to it. The row is now **6.10%** above its 22 October close and **1.45%** above the ¥4,357 it closed at the day before the rout, so this line has repaid the fall too, a session after Harmonic Drive did. The drawdown narrows **2.2** points to **27.7%**, the tightest in tier one bar Ambarella, and the year to date reads **+17.9%**. For once the turnover came with the price: three gains on 1.13, then 1.06 times average, which is participation rather than an absent seller. **4 November gave the participation argument back from the other side.** **−2.71% to ¥4,167** on **1.78 times** average volume, 1,859,600 shares against 1,042,985, after opening ¥4,315 and trading as low as ¥3,993: the second-heaviest relative turnover in tier one and the **lowest close since 26 October**. The drawdown widens to **31.8%** and the year to date reads **+11.2%**, while the twelve-month cell _improved_ to **+13.0%** on a year-ago base down 4.8% to ¥3,687, which is the calendar outrunning a down session again. Nothing was filed since the **2 November** notice that the ¥15bn buyback is complete, so the bid that supported the last three gains has been gone for two sessions and this is the first full one without it.

Actuator Motor + gear + electronics

### Schaeffler

SHA0.DE

**19 November.** Schaeffler closed at €5.19 (−1.52%) on 0.85 times its prior 20-session average volume.

**18 November.** Schaeffler closed at €5.27 (−0.75%) on 0.81 times its prior 20-session average volume. The auction close is corroborated by S&P Global; Yahoo’s daily bar is pending backfill.

**17 November.** Schaeffler closed at €5.31 (−2.93%) in Xetra’s closing auction. [S&P Global’s venue history](<https://stockanalysis.com/quote/etr/SHA0/history/>) confirms the price and 764,238 shares. Yahoo’s daily close is pending backfill.

**16 November.** Schaeffler closed at €5.47 (−1.26%) in Xetra’s closing auction. [Xetra history](<https://www.investing.com/equities/schaeffler-historical-data>) and the backfilled Yahoo daily bar confirm the auction close.

**13 November.** Xetra closed at €5.54 (−0.72%). The [Yahoo daily bar](<https://finance.yahoo.com/quote/SHA0.DE/history/>) has backfilled at the same price as the post-auction quote and [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>).

The most concrete commercial traction on the list. Schaeffler signed a strategic partnership with Swiss **Hexagon Robotics** in April 2026 for strain wave and planetary gear actuators, and a separate technology and supply partnership with UK-based **Humanoid**. It is also a customer: it plans to deploy at least a thousand Hexagon robots across its own plants within seven years.

**And on 13 October that traction finally got a number.** With its preliminary nine-month figures Schaeffler disclosed humanoid-robotics **orders on hand of approximately €400m**, against **€350m** at the end of the first half, after acquiring **two further customers in the Greater China region** during the third quarter and taking its humanoid customer count to **five**. That is the first quarterly, countable humanoid figure any tier-one name on this page has published. The qualification is the one this card has always carried, and the same release supplies it: the group turned **€17.5bn** of nine-month revenue, against €17.7bn a year earlier, into a **4.5%** EBIT margin before special items, with the third quarter alone back down at **4.2%**, nine-month free cash flow before M&A at **negative €40m** against positive €47m, and about **€240m** of one-off E-Mobility special charges, roughly €220m of it in the quarter. Guidance for 2026 and the 2028 mid-term targets were both confirmed. So the best humanoid disclosure on this page is also a rounding item against the revenue line it sits inside, which is this card in a single release. The validated interim statement follows on **3 November**.

The in-house manufacturing point is fair — motors, power electronics and encoders are built internally. The offset is that Schaeffler remains predominantly an automotive supplier, so a humanoid win is diluted by a large, cyclical, structurally pressured base business. At the 13 November close the stock was down **33.7%** YTD and **16.0%** over twelve months. It was **53.8%** below its 52-week high, the widest current drawdown in tier one. The fortnight of going nowhere ended on Friday. Four sessions had sat inside nine cents — €7.34 on 7 September, €7.43 on 8 September, €7.40 on 9 September and €7.34 on 10 September — and then 11 September broke the range, −2.0% to €7.19, the lowest close since 2 September. Monday went further, −3.3% to €6.95, and Tuesday further again: −2.4% to **€6.78**, the lowest close since **20 March** — nearly six months undone in three sessions. Wednesday halted it, +1.2% to €6.86, the only tier-one name to rise that session, and Thursday reversed harder, **+3.1% to €7.07** — its best session since 2 September. That did not hold either: −2.0% to €6.93 on 18 September, then +0.9% to €6.99 on 21 September and **+2.0% to €7.13** on 22 September, its best close in a week — before giving it back, −2.2% to €6.97 on 23 September, in the same dollar-led metals session that took the minerals tier down. Then **24 September** broke the whole range: **−6.2% to €6.54**, its worst session since **31 July** and its lowest close since **26 November 2025**. The sector context does not cover it — the DAX fell 0.57% with autos the worst-performing corner, and the largest carmakers in that move fell 2.2–2.4% — and there was no company disclosure on the day. The shape argues against a single headline: no gap at the open, a slow grind from €6.84 to €6.75 by mid-morning, then an acceleration through the last ninety minutes on 1.33m shares, more than three times Monday's volume. What there was, and what this page missed for two days, is a sell-side catalyst one session old: **Jefferies cut its target to €8.70 from €9.45 on 23 September**, keeping a Buy, with analyst Vanessa Jeffriess writing that Schaeffler's reduction of its own 2028 targets had damaged credibility — while calling the non-automotive robotics business the most attractive investment story in the sector. That note is the documented cause of the **−2.2%** on 23 September, which this page had loosely hung on a dollar-led metals session a German auto-parts supplier has no business being in. It does not by itself carry a 6.2% second-day move, and no disclosure does, so the caution stands — but “a cause it cannot source” was too strong: the two sessions together are **−8.3%**, €7.13 to €6.54, and the first of them has a name on it. It never joined the mid-August drawdown, so for a fortnight it had nothing to give back; it has given back far more than that since, and the month now reads **15.0%** lower where three weeks earlier this was the one tier-one name essentially flat on that measure. Two quiet sessions have extended it rather than ended it: **−2.6% to €6.46** on 28 September and **−0.9% to €6.40** on **29 September**, the **lowest close since 21 November 2025**, with nothing published on either day. Note what it is not: a chip name. Schaeffler fell on the same session as the semiconductor selloff but for its own reasons, and at 0.61 to the robotics basket it is one of the two loosest fits on the tier, alongside Ambarella at 0.54 — the two names here that are really something else. Several of these closes were reconstructed from the closing auction print when the German daily feed swallowed the session, and the ledger keeps closing behind itself: the **7, 16, 21, 22 and 23 September** bars have all filled at exactly €7.34, €6.86, €6.99, €7.13 and €6.97, the last of them a fifteenth reconstruction confirmed to the cent. **24 September** was the new hole at the time, and because it carried a 6.2% session rather than a quiet one it was closed on four routes rather than two: the 17:35 auction bar at **€6.54** on 484,271 shares, previousClose and chartPreviousClose both reading the same, and an independent German quote service carrying the session at 6.540. **The newest is 14 October, and on that session this was the only green figure anywhere on the page:** **+1.54% to €5.94**, while Tokyo fell six to twelve per cent the next morning and the US minerals tier fell three to ten the same afternoon. Its bar is null again, and three routes agree on the close: regularMarketPrice stamped **17:35:17 CEST**, previousClose on a five-minute request, and a **17:35** auction bar on **1,221,280 shares** against 13 to 16,488 in each of the six bars before it. **Two sessions have followed and both lost their bar as well.** **15 October** closed **3.20% lower at €5.75**, its lowest close since **1 October 2025**, recovered on three routes at the time and since backfilled by the vendor at exactly that figure. **16 October** then went the other way, **+1.39% to €5.83**, which lifts it off that low and brings the drawdown in to **51.3%**, still the widest in tier one. Its bar is null too, and this is the first hole in the series lost with Xetra already closed rather than open: the volume field is populated at **877,225** shares while the close is empty, and two routes carry €5.83, regularMarketPrice stamped **17:35:21 CEST** and a 17:30 five-minute bar on zero volume, so the continuous close rather than the auction. See the method note.

**On 20 October the order book got a factory behind it, and the number attached to it is the old one.** Schaeffler opened what it calls a humanoid-robotics **lighthouse factory in Taicang**, China: about **7,500 m²** holding **six humanoid robots across five assembly stations** that build the **linear and rotary actuators** going into the next generation of humanoids, plus a training and data centre drawing tasks from the group’s roughly 100 plants and a digital twin of the line. The company describes it as the start of mass production of those components, which is the first physical capacity any tier-one name on this page has put behind a humanoid order book. The caveat is in the same release: it restates humanoid order intake at **€400m** after **two new Greater China customers in Q3**, which are the figures the **13 October** nine-month preliminaries already carried and this card has carried since. So the order book did not move on 20 October; the capacity did. Several outlets ran the €400m as fresh Q3 news that day, ten days after it was first disclosed.

On the window ending 30 October 2026, Schaeffler’s thin SFFLY ADR correlated **0.29** to the robotics basket and the Xetra line correlated **0.61**. The ADR had previously measured as low as 0.16. In the window through 13 November the Xetra line correlates **0.61** to the basket and **0.46** to REMX, the highest REMX coefficient in tier one.

Partners **Hexagon, Humanoid** Humanoid order intake **~€400m, repeated 10 Nov** YTD **−37.9%** Corr robotics **0.61**

**3 November: the largest number the company published this year is a restructuring bill, and the humanoid line beside it is a sentence rather than a figure.** The release, datelined Herzogenaurach and headed “Schaeffler announces further performance improvement program”, plans a net reduction of **3,460 positions** in Europe, **2,660** in Germany, and the closure of **five sites**, four German: Dortmund (about 430 employees), Morbach (160), Berlin (270), Schwalbach (385, with about 190 positions affected and many expected to move to Frankfurt) and Herstal in Belgium (15). One-time expenses are approximately **€480m**, mostly provisioned in 2026, against approximately **€300m of annual savings by 2030**, with implementation scheduled for **2027 and 2028** and a voluntary temporary pay reduction at all levels starting with the board. COO Dr Jochen Schröder names **humanoid robotics**, with defence, as a promising future growth area built on existing core-business technology. **Read the two documents of 3 November together and the direction is clear enough:** the audited nine-month statement declined to restate the approximately **€400m** humanoid order book the 13 October preliminary release had given, while the same morning’s programme quantifies the European automotive problem at €480m of charges and 3,460 jobs. The growth area is named and unsized; the shrinkage is sized. The stock closed **−5.19% at €5.66** on **1.54 times** average volume, its **lowest close since 23 October**, which takes the drawdown to **52.8%**, the widest on this page outside the minerals tier, and the year to date to **−32.3%**.

**A product disclosure this card had not carried, and it belongs here rather than in the order book.** On **13 August 2026** Schaeffler said it had completed validation testing of _formed strain wave gearboxes_ for humanoid robots and laid the foundation for mass manufacture from **2027**, starting in Germany and rolling out to other regions, with manufacturing cost more than **25%** lower and material consumption more than **75%** lower than cut gearboxes. The figures are the company’s and are not verified here. The significance is structural rather than financial: a strain wave gear is Harmonic Drive’s product and the first row of the parts table, and this is the one tier-one name on this page entering another’s core product with a cost claim instead of a partnership. It also sharpens what the €400m order book is for, because a book of orders and a validated low-cost process are different assets and this group now has both. Against that, the dilution problem on this card is unchanged: the same group turned €17.5bn of nine-month revenue into a 4.5% margin. **20 October: +2.08% to €5.89**, the best session anywhere in tier one on this run, which pulls the widest drawdown in tier one back in to **50.8%**. The close had to be rebuilt rather than read: the 20 October daily bar is null in its close _and_ its volume, the thirtieth hole in this series, and two routes agree on €5.89. Turnover comes from the quote-level field instead, 885,599 shares against a twenty-session average of 1,060,542, or 0.84 times normal. See the method note.

**23 October: −2.88% to €5.39, the lowest close since 26 August 2025 and a fourteen-month low.** The drawdown widens to **55.0%**, the widest this page has recorded on the name and still the deepest in tier one. The bar is null again, the **thirty-third** hole in this series, and three routes carry the figure: the 17:30 Xetra auction bar, regularMarketPrice stamped **17:35:12 CEST**, and the Frankfurt and Stuttgart lines at €5.41 and €5.40. No volume survives, so there is no turnover comparison for the session. Nothing is dated to the company: the lighthouse factory opened on 20 October, the €400m order book and five customers date from the 13 October preliminaries, and the next thing that could move this card is the validated interim statement on **3 November**. Note how much of the twelve-month cell is calendar rather than tape, because it is extreme this run: it fell **9.3** points to −17.3%, of which **6.6** is the one-year base rolling onto 23 October 2025 and 2.7 the session.

**Watch**The humanoid order book: approximately **€400m** and five customers, reported with the preliminary nine-month release. The question has narrowed rather than closed, because an order book is not revenue and actuators are still not a reported segment, so the robotics _margin_ stays invisible inside an auto-parts P&L. The **3 November** reading has now happened and it subtracted the figure rather than refreshing it. **26 October was the largest session anywhere on this page and the feed refused to print it.** **+5.57% to €5.69**, its **highest close since 21 October**, recovering most of the fourteen-month low of 23 October, on **814,257** shares or **0.75 times** the twenty-session average. The daily bar returns a null close while reporting that volume, so the figure here is read off the hourly series, which ends on a 17:30 bar at exactly €5.69 and agrees with meta.regularMarketPrice on three requests: the **thirty-fourth** such reconstruction in this line and the only one currently open. The drawdown narrows **2.5** points to **52.5%**, still the deepest in tier one, and the four-week cell improves to −14.2%. A 5.6% session one week before a validated nine-month statement is a market positioning for 3 November, not a disclosure. **27 October recovered to the 21 October close exactly, and the feed swallowed the session again one run after the thirty-fourth hole closed.** It closed **€5.82, up 2.28%**, matching 21 October to the cent, on **733,134** shares, **0.67 times** the twenty-session average. The 26 October bar, published here last run at **€5.69**, has backfilled at exactly that figure; the 27 October bar is null in its place with its volume reported, and the close published here is read off the hourly series, which runs to a **17:30** bar at **€5.82**, agreeing with meta.regularMarketPrice stamped **17:35:18 CET**. That is the **thirty-fifth** hole: thirty-five reconstructions, thirty-four confirmed to the cent, one open. The drawdown comes in **1.1** points to **51.4%**, still the deepest in tier one by nearly five points. The €400m humanoid order book from the **13 October** preliminary release is the only humanoid-attributed order figure on this page, and the validated statement arrived on **3 November at 07:45 CET** without restating it. **2 November: €5.97, up 2.75%** into it on **0.61 times** average volume, 633,374 shares against 1,037,632, between €5.75 and €6.04, the highest close since 29 October, which brings the drawdown in **1.3 points to 50.2%** and hands the deepest-in-tier-one place back to Yaskawa at **50.5%**. **It confirms every group figure and declines to put a number on the one line this card exists for.** Nine-month revenue **€17,481m**, 1.1% lower reported and 0.4% currency-adjusted; **EBIT before special items €794m**, up 6.5%, a **4.5%** margin against 4.2%; free cash flow before M&A **−€40m** for the nine months with the quarter alone **+€260m**; net income attributable to shareholders **€11m** against −€244m after €201m of special items, or **€0.01** a share; net financial debt **€5,296m** at **2.3 times** EBITDA before special items. The board reconfirmed 2026 guidance on 26 October at revenue **€22.5–24.5bn**, margin **3.5–5.5%** and free cash flow **€100–300m**, and the 2028 mid-term targets stand. E-Mobility is where the quarter went: revenue €3,871m, up 4.6% currency-adjusted, against EBIT before special items of **−€571m** on a −14.8% margin, with write-downs and provisions for lower European-carmaker volumes, mostly in China, totalling about **€240m** of which **€220m** was booked in the third quarter. Greater China revenue is **7.9%** lower currency-adjusted. Bearings & Industrial Solutions, the division an actuator would report into, did **€4,795m** at **8.9%**, and nothing inside it is broken out. On humanoids the statement says only: “**Especially in humanoid robotics, we won new customers and continued to expand our order book during the reporting period.**” No euro figure, no customer count, no segment. So the €400m and the five customers stay a preliminary disclosure of 13 October that the audited statement does not repeat, and this card’s question is unchanged rather than answered: an order book is not revenue, and an order book that appears in a preliminary release and not in the validated one is a weaker fact than it looked three weeks ago.

Motor 6-axis integrated motion

### Nidec

6594.T

**20 November.** Nidec closed at ¥1,644 (−4.97%) on 1.39 times its prior 20-session average volume. Yahoo’s daily close is null; its 15:30 JST quote metadata supplies the completed auction close.

**19 November.** Nidec closed at ¥1,730 (+1.76%) on 0.64 times its prior 20-session average volume.

**18 November.** Nidec closed at ¥1,700 (+1.92%) on 0.52 times its prior 20-session average volume.

**17 November.** Nidec closed at ¥1,668 (−2.28%) on 0.58 times its prior 20-session average volume.

**16 November.** Nidec closed at ¥1,707 (−4.53%) on 1.45 times its prior 20-session average volume. After the close, it [announced approval](<https://www.release.tdnet.info/inbs/140120261116581515.pdf>) of its extension to 16 December for the half-year securities report.

**14 November.** On 14 November Nidec said it was considering [seeking an extension to 16 December](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1114-01/261114-01en.pdf>) for its half-year report. It [cancelled the extraordinary meeting’s record date](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1113-01/261113-01en.pdf>) and [confirmed a zero interim dividend](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1113-02/261113-02en.pdf>) on Friday. Tokyo had closed unchanged at ¥1,788.

**5 November.** Nidec renewed two ¥300bn unsecured commitment lines through 5 November 2027. The covenant requires March 2027 consolidated equity of at least 90% of March 2026 equity. The filing addresses liquidity. JPX still lists the company under [Securities on Special Alert](<https://www.jpx.co.jp/english/listing/measures/alert/index.html>). [Issuer notice](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1105-01/261105-01en.pdf>).

The world's largest electric motor maker, building integrated humanoid motion units that fold motor, gearbox and controller together — FLEXWAVE strain-wave reducers, KINEX cycloidal gearboxes and planetary systems shown at Automate 2026. The scale argument is real: roughly **¥2.6 trillion** in annual revenue (about $17bn, not $2.6tn as the post's garbled phrasing implies).

This is the one name carrying a governance problem rather than just an execution problem, and on **4 September 2026** one half of it was finally published. Nidec disclosed the **Investigation Committee's report** on the quality strand, received on 2 September and released once confidentiality review was complete, together with its recurrence-prevention measures. The committee, established in May 2026, found a series of quality-related failures: **4M change-rule violations, falsification and fabrication of testing and inspection results, work outside approved inspection and manufacturing conditions, the use and shipment of substandard and unauthorised items**, and breaches of record preparation and management rules, arising from an “unhealthy state” of excessive cost-reduction pressure, inefficient risk monitoring and inadequate quality-assurance resourcing. The decisive line for holders is financial: Nidec's evaluation of the individual matters found **no material impact on its consolidated financial statements**. Remediation gives quality managers authority to suspend shipments, revises reporting lines, adds permanent complaint boxes, puts every executive and employee through quality-compliance training and starts continuous quality audits; disciplinary action is promised but nobody is named. The stock rose 5.7% on the day and a further 7.7% on the Monday after it, to ¥2,812, before losing 3.2% to ¥2,722 and a further 4.8% to ¥2,591 — giving the whole post-report rally back by Wednesday — then taking most of it straight back with +5.6% to ¥2,737 on 10 September, the largest single gain anywhere on this page, and easing 1.7% to ¥2,691 on 11 September, a further 1.3% to ¥2,655 on 14 September, then 0.5% to ¥2,642, +0.2% to ¥2,646 on 16 September, +2.9% to ¥2,724 on 17 September, and +4.8% to ¥2,856 on 18 September on the Bank of Japan's hike — the largest single gain anywhere on this page in this period, and its highest close since **4 June**. The report landed on 4 September with the stock up 5.7% that session; of the **ten** sessions since, seven moved more than 1.3% in one direction or the other — +7.7%, −3.2%, −4.8%, +5.6%, −1.7%, −1.3%, −0.5%, +0.2%, +2.9% and +4.8% — with a three-session lull in the middle that lasted exactly as long as it took the market to find a new reason. That reading rested on an empty disclosure record, and the record is not empty. On **10 September** — the same session the stock rose 5.6%, the largest single gain anywhere on this page — Nidec disclosed that a shareholder had begun a **derivative action** against two former directors the day before. This page's line that Nidec had disclosed nothing at all since 4 September was accurate when written and stopped being accurate on 10 September; it was repeated for fifteen days after that, and the claim audit now carries it. The narrower statement that survives is that nothing has been disclosed on the _filing_ strand since 4 September, and that the moves since have come from interpretation rather than information — including on the day the suit became public, which the tape did not mark at all. Everything else is still open — though it is worth being precise about what "open" means, because this page previously was not. The separate third-party committee on the accounting irregularities has _finished_: appointed in September 2025, it delivered a mid-term report on 27 February 2026 and its **final report on 17 April 2026**. What outlasts the inquiry is its consequence. Nidec is still correcting accounts back to the year ended March 2022, has not closed FY3/26, and has repeatedly postponed results — including the April–June quarter, delayed again on **5 Aug 2026**. A tariff strand runs alongside: additional customs duties from declaration errors were estimated in March 2026 at **US$69.7m**, about ¥11.14bn, including interest. The quality strand covers changes to materials, processes and designs made without customer approval, mishandled inspection data and mislabelled production locations, and on 5 August an individual shareholder formally requested that the company pursue liability against current and former directors. A second, larger request followed: on **20 August** multiple shareholders jointly demanded the same action under Article 847 of the Companies Act, disclosed the next day. Those were demands that the company sue, and they concern the quality and accounting conduct. What was filed on **9 September** is a different instrument and the first claim in this saga that carries a number: an individual shareholder sued two _former_ directors — founder **Shigenobu Nagamori** and **Hiroshi Kobe** — in the **Kyoto District Court** (case 1749 of 2026) for **¥28,730,558,300** plus delinquent charges under **Article 462, paragraph 1**, the capital-maintenance provision rather than the duty-of-care one. The claim is that three specific capital returns exceeded the distributable amount: the treasury-share buybacks of **September 2022** and **February–March 2023**, and the **December 2022** midterm dividend. That is the restatement arriving as a liability instead of as an accounting entry, and it is the mechanism worth understanding — correct the accounts for those years downward and the distributable amount shrinks retroactively, which is what makes capital already paid out unlawful after the fact. Nidec says the action, being against individuals, will not affect its business performance. That is accurate about the P&L and beside the point about governance: the suit does not need to succeed to establish that the restatement has consequences with numbers attached, and ¥28.7bn is the first one. Nidec has an Executive Responsibility Investigation Committee, running since 13 March 2026, to decide whether directors, auditors and executive officers bear legal liability. The US ADR has barely traded since February, which is why the Tokyo line is the only usable quote. On the tape, **16 October** took it lower again, **−1.04% to ¥1,800**, a fresh low beneath Thursday’s on **0.72 times** average volume and on a session the Nikkei rose 1.39% and every other Tokyo name on this page closed higher. That is the second consecutive new low made on below-average turnover, which is the same reading this card has carried: not a new seller arriving, the bid going away. **19 October made it a third, and on a third of normal volume:** **−0.61% to ¥1,789** on **5,439,500 shares** against a twenty-session average of **16,858,410**, which is **0.32 times** normal and the thinnest session of this whole episode, after trading as low as **¥1,749**, a new 52-week low. It is the session on which the Committee’s conclusion naming nine people was already public, so the reading has to change: this was not an absent bid on no news but an absent bid on the most consequential document this name has produced. It leaves the name **39.2%** off its high, **16.1%** lower year to date and **28.3%** over twelve months.

FY3/26 revenue **¥2.71tn** FY3/26 net loss **¥564.6bn** Quality report **published 4 Sep** Impairment **\>¥632bn** Audit opinion **disclaimed** Internal-control filing **submitted 30 Oct** Off high **−44.1%** UBS target **¥2,400**

**20 October: the Report posted, on the day Nidec said it would, and it carries a number.** “Nidec Discloses the Investigation Report Issued by the Executive Responsibility Investigation Committee”, dated **20 October** in Kyoto, says the redaction is complete and attaches the Report. Two things in it are new. The **appropriation of surplus was excluded** from the Committee’s investigation, because Nidec had not filed amended securities reports for past years and the distributable amount that would have been examined is therefore uncertain; the company says it will examine legal action on that separately, after the restatements are done. And only a **Japanese digest** is out, with the English version promised and not yet published, which is why no English-language wire carries the release. Jiji, Bloomberg and Nikkei, reading the Japanese text, put the damage at about **¥27.6bn** — third-party investigation costs, audit costs and the exchange penalty — with the apportionment among the nine left to a court, because contribution is too hard to split. Nidec itself is still only “deliberating legal actions such as lodging a claim for damages”, so the board resolution an aggregator reported on 16 October still has not happened. **21 October: ¥1,762, up 0.23%**, on 8,954,900 shares against a twenty-session average of 17,266,920, or **0.52 times normal**, with an intraday low of **¥1,748**, the lowest print on this page’s record. The founder found liable for ¥27.6bn moved the close four yen on half the usual turnover, and the ¥1,758 of the session before stands as the lowest close since **12 February 2016**. Seven days remain to the 28 October anniversary. **23 October: the English digest posted, and the tape set a ten-year low.** “Nidec Discloses an English Version of the Digest of the Executive Responsibility Investigation Committee’s Investigation Report” went out through Business Wire at **01:04 EDT on 23 October**, which closes the gap recorded above: the substance is the 20 October release and the **¥27.6bn** in it, unchanged, but it is now readable in English and the wires can carry it. The stock closed **¥1,705, down 0.58%**, on **11,845,700** shares against a twenty-session average of **17,628,095**, or **0.67 times** normal, with a low of **¥1,637**. There is no close at or below ¥1,705 anywhere in the ten years of daily history the feed carries, back to **24 October 2016**.

**Watch**Nidec’s half-year securities report is [now due 16 December](<https://www.release.tdnet.info/inbs/140120261116581515.pdf>). The company plans to incorporate the prior-period adjustments reflected in the September filing. The amended prior-year reports and work on the audit disclaimer are on a separate track. The filing of the 30 October internal-control submission is part of the separate JPX review. The quality strand is closed; the accounting strand is filed but unassured. Nidec obtained approval on 30 June to extend the filing deadline for its annual securities report for the year ended March 2026 from 30 June to **30 September**, and it met that extended date; what it did not obtain was an audit opinion. The April–June quarter behind it is still missing. Until both are current, position sizing here remains a governance decision rather than a robotics one. A third strand is now live alongside them: the **9 September** derivative action over the 2022–23 buybacks and dividend, which turns on what the restated distributable amount actually is — so the filing due on 30 September is also the document that sizes that claim. On the closed strand, the verified count is worth carrying: the 4 September report found **844 cases** of quality misconduct over about a decade — mostly changes to materials and processes without customer approval, alongside falsified tests and country-of-origin mislabelling — and Nidec states no individual matter has a material impact on its financial statements. Press reporting that attaches a ¥250bn write-down to that report is conflating it with the accounting strand; the claim audit now carries this. A fourth strand opened on **28 September** and it is the loudest: _Diamond Online_ reported that morning that the retroactive impairment will be roughly **¥1tn** in the year ended March 2026, and that an emergency board on **25 September** resolved to dismiss president **Mitsuya Kishida**. Nidec confirmed that executive changes and large-scale impairment processing are under consideration and that nothing has been decided, which leaves three impairment figures in circulation from three sources — the third-party committee’s **~¥250bn** in its March report, the only one in a published document; _Nikkei_’s several hundred billion on 15 September; and Diamond’s ¥1tn. The filing due on 30 September is what adjudicates between them, which is why the deadline now carries more than a compliance question. This page also has a standing observation to retire. Through this whole run Nidec showed the _tightest_ drawdown on the page by a wide margin — 3.4% off a 52-week high of ¥2,941 as recently as 25 September — and this page kept noting that a stock trading that close to its high is not the same thing as a stock that has recovered, on a company whose accounts for the year ended March 2026 do not yet exist. One limit-down session settled it: **−17.6% to ¥2,340**, **20.4%** off the high, and the twelve-month column through zero from +6.7% to **−11.3%**. Its twelve-month return, meanwhile, has changed sign twice inside a week, and neither move came from the company. Nidec rose 4% on **12 September 2025**, and that session rolling into the comparison cut the twelve-month return from +7.5% to +2.1% on a 1.3% price move. This page then forecast the next crossing to the day — on **16 September** the base became 16 September 2025’s ¥2,671, and a +0.2% session took the twelve-month number **below zero, to −0.9%**. One day later the base rolled again, to 17 September 2025’s ¥2,691, and a +2.9% price move put it **back above zero at +1.2%**. Two sign changes, two sessions, one of them driven purely by which day fell out of the window. The next session was the clean control case: the base did _not_ move — ¥2,691 on both 17 and 18 September 2025 — so a +4.8% price move took the figure to **+6.1%** and nothing else did. Same column, three consecutive readings, two mechanisms. The holiday then supplied a fourth: the base rolled six calendar days at once, off ¥2,691 onto 24 September 2025's **¥2,642**, so a _falling_ session left the twelve-month figure _higher_, at **+6.8%**. It is the cleanest illustration on this page of a return figure that moves because the calendar did — and a warning about reading any twelve-month column on a date when the base is volatile. The **28 September** reading is the control case for the whole sequence: the base moved by a single session, from ¥2,662.5 to ¥2,636.5, and the column fell eighteen points anyway. When the price does something like this, the calendar stops mattering. **29 September** extended it on ordinary trading rather than a limit — −3.8% to **¥2,250**, a fall of ¥90 against a ¥500 limit, and the **lowest close since 15 April 2026** — taking the drawdown to **23.5%** and the twelve-month cell to **−12.1%**. Then the disclosure record caught up with the news. At an extraordinary board meeting on **29 September** Nidec accepted the resignation of president and chief executive **Mitsuya Kishida** and appointed **Michio Kaida**, First Senior Vice President and vice-chair of its reform committee, in his place, saying Kishida “had, on certain occasions, made statements or engaged in conduct in relation to financial reporting that could not necessarily be regarded as appropriate.” **30 September** took the stock **+4.7% to ¥2,355**, its largest gain of this whole episode, on that plus a _Nikkei_ report of final-stage talks to sell **Nidec Components** to **Carlyle** for more than **¥100bn**, which on that day was a press report and not a disclosure. It became one the next morning: on **1 October** Nidec’s board resolved to transfer all 66,302,120 shares of Nidec Components to **TCG2603 Co., Ltd.**, indirectly owned by a Carlyle-related fund, for **¥102,980m**, transferring **1 December 2026**. The unit earned ¥5,465m of operating profit on ¥32,092m of sales in the year to March 2026, and its end markets include collaborative robots, so the disposal takes out the most robotics-adjacent sensing business inside this name. The drawdown comes back to **19.9%** and the twelve-month cell to **−10.6%**. The document has now arrived, and it settles the impairment question while opening a larger one. Filed on **30 September**, the FY3/26 report puts revenue at **¥2,708.7bn**, an operating _loss_ of **¥518.9bn** and a record net loss of **¥564.6bn**, with impairments of more than **¥632bn**, against a restated prior year of ¥128.2bn operating profit and ¥84.6bn net profit. So the real figure sits between the two that were circulating: far above the roughly **¥250bn** Nidec itself had under impairment review, and well below the **¥1tn** _Diamond Online_ reported. Neither source was right, which is the retrospective case for having declined to print either as fact. The larger question is the audit. **PwC Japan** issued a **disclaimer of opinion** on the consolidated financial statements and, separately, on the internal control report, on the grounds that it could not obtain sufficient appropriate audit evidence, and that is the second consecutive year the securities report has carried one. The accounts are therefore filed and still not assured: every figure in this paragraph is the company's own, unvalidated by its auditor. That is a different condition from the one this page has spent a month describing, and it is not obviously the better one, because “no accounts yet” at least had a resolution date. The date that matters now is the **end of October**, and it is an anniversary rather than a discretion: the exchange designated Nidec a **Security on Special Alert** effective **28 October 2025**, and its rules require the written confirmation of internal management system one year on, with delisting if the controls are judged inadequate. New president **Michio Kaida** has said publicly that he intends to keep the listing. Position sizing here was already a governance decision rather than a robotics one. It is now a listing decision as well, and that is a question about whether the instrument continues to exist rather than about what it is worth. The tape has started pricing that question. On **9 October** Nidec fell **6.07% to ¥2,010** on **21.3m shares** against 11.3m, gapping to ¥2,000 at the open, trading as low as **¥1,941** and closing at its **weakest level since 31 March 2026**, with its year-to-date figure crossing zero to **−5.7%**. Nothing was filed, which is the point: nineteen days from the anniversary, with a disclaimed opinion on the record and internal control over financial reporting assessed **ineffective** as of 31 March 2026, new lows on doubled volume are the market shortening the odds on a deadline rather than reacting to news. Two things said at the **1 October** press conference frame what has to change by then. Kaida said the company is still seeking PwC’s sign-off and aims to complete the work needed to obtain opinions by the **end of October**, the same end of October the written confirmation falls due. And PwC Japan’s own chief executive, **Masataka Kubota**, attended it and said audit evidence obtained from officers and employees who had made false representations could not be trusted. An auditor saying that in public three weeks before the deadline is the single most informative sentence available on this name, and it is not about motors. **And on 7 October a second question arrived from the register.** **Oasis Management**, which holds about **8%**, publicly asked the board to begin a **preliminary review of privatization options**: outside directors to retain their own financial advisers and solicit preliminary, non-binding proposals from strategic buyers and private equity in Japan and overseas, with the disclaimer of opinion and the risk of an involuntary delisting given as the reason. Oasis is explicit that this is a feasibility check and not a decision to sell, and no board response is on the record. It cuts in an awkward direction for a watchlist built on disclosure: a take-private would resolve the listing risk by removing the listing, and with it the quarterly numbers that make this name readable at all. This page missed the item on the day and the claim audit carries the row; it is recorded here because the position on Nidec is now two deadlines rather than one. **And the tape has now gone somewhere it has not been in twelve years.** On **13 October**, Tokyo’s first session after Sports Day, Nidec fell **6.62% to ¥1,877** on **23.96m shares** against 21.28m, taking out the 9 October low without pausing on it. That is the **lowest close in the ten years of daily history available**, with nothing at or below it since the series begins in October 2016 and the last monthly close at that level in **2014**, on a split-adjusted basis. The drawdown widens to **36.2%**, third-widest in tier one behind Schaeffler and Yaskawa, and the year-to-date cell to **−12.0%**. Nothing was filed against the session, and the Nikkei fell 1.82%, so the name shed three and a half times the index. **Two items from the preceding days do belong on this card, and this page reported their absence rather than their content.** **UBS** downgraded Nidec to **Neutral from Buy** on **8 October**, analysts **Shingo Hirata** and **Kohei Takahashi** cutting the target to **¥2,400** from ¥2,800 and putting their operating-profit forecasts **16%** and **22%** below consensus for this year and next; _Bloomberg_ reported the stock down as much as **9.3%** on it. Separately, **R&I** cut the issuer rating one notch to **A+** and placed it on monitor with negative implications, citing the impairment losses. Neither is a filing, which is the distinction this page drew at the time and still holds; what it wrongly concluded from the absence of a filing was that the sessions were not reacting to news. They were, and the claim audit carries that row. With **nine days** to the 28 October anniversary, the sell side and a rating agency have now both moved, and the price is making **ten-year** lows ahead of a deadline that is about whether the instrument continues to exist. **15 October** took it lower again, **−6.24% to ¥1,819**, the **lowest close since 11 July 2016**, and the turnover is the part worth recording: **15.10m shares against a twenty-session average of 15.80m**, which is **0.96 times** normal. A ten-year low made on below-average volume is an absent bid rather than a new seller. **16 October** repeated it, −1.04% to **¥1,800** on **0.72 times** average volume, taking the drawdown to **38.8%**. **And the filing drought ended on that session, which this page missed at the time.** Dated 16 October in Kyoto, Nidec disclosed that it had received the investigation report of its **Executive Responsibility Investigation Committee**, the body established on **13 March 2026** to determine whether current and former directors, auditors and vice presidents “bear legal liability, including whether they breached their duties in the execution of their responsibilities” for the improper accounting. Nidec will publish the Report itself only after redaction for trade secrets and privacy, will “deliberate legal actions such as lodging a claim for damages”, and convened both its Board and its Audit and Supervisory Committee the same day to hear the Committee out and disclose the actions it would take for the time being. None of those three follow-ons has appeared yet, so what exists is a receipt rather than a finding: no contents, no names, no claim. This matters on a card whose question is whether the instrument survives October, because it is the first document to put the directors’ own liability on the record rather than the company’s accounts, and because a damages claim against former management is the one development that cuts _for_ the equity. The release went out at **13:23 JST**, inside the session, and the tape did not move for it; the claim audit carries the row on this page having read the session as newsless. **Two days from the exchange deadline, the tape has gone quiet rather than decided.** **26 October: +0.18% to ¥1,708** on **6.17m shares against a twenty-session average of 8.80m**, **0.70 times** and the lightest turnover in tier one, two sessions after the **lowest close since 31 March 2026**. Nothing has been filed. The twelve-month cell _improved_ 2.6 points to **−30.6%** on a 0.18% session, because the year-ago base fell ¥2,550.5 to **¥2,460** as 24 October 2025 replaced the 23rd — so the best-looking column this name has had in a month is entirely a bad day dropping out of the window, and should not be read as recovery. Two strands now circulate on the deadline that this page records as single-sourced rather than as fact: plans settled this month to discipline about **20** executives and employees, and an exchange decision expected at **year-end** rather than on 28 October itself. Neither appears in a Nidec release or a TSE notice. If the second is right, the anniversary is a filing date and not a verdict. **On 27 October, one day out, the release list still ends at the 23 October English digest.** Nothing is dated 24 to 27 October: no audit opinion, no amended securities reports for the restated years, no internal control confirmation, no damages claim. The stock rose **1.93% to ¥1,741** into that on **0.48 times** average volume, the lightest turnover on this page and the second consecutive run in which this row has had it. The twelve-month cell went the other way on the rising session, −30.6% to **−32.3%**, because the year-ago base rolled up ¥2,460 to ¥2,570.5 and took 3.0 points out against the 1.3 the price put in. Last run that mechanism flattered this row; this run it reverses, and the company did nothing in either direction. **28 October is the anniversary of the designation and the market finally traded it.** **+3.56% to ¥1,803** on **1.55 times** average volume, **26.13m** shares, the heaviest relative turnover in Tokyo and a reversal from 0.48 times the session before, which was the lightest on the page. Its **highest close since 15 October**, and **5.13%** above its 22 October close. The drawdown narrows **2.1** points to **38.7%**. **One disclosure landed, and it is not the one the listing turns on.** On **27 October** the company said it had **decided to file** a damages suit against nine former officers: six former directors (Nagamori, Kobe, Sato, Yoshimoto, Seki, Kishida) and two former corporate auditors (Murakami, Ochiai) under Article 423(1) of the Companies Act, plus a former Vice President who was CFO from May 2022 to March 2026, with the action delegated to three of the Investigation Committee’s own attorneys. No suit is filed, no court is named and no amount is set; the release is headlined “Nidec Files Suit” regardless, which the claim audit now carries. Checked after the 28 October close, the release list ends there: no audit opinion, no amended securities reports, no internal control confirmation. A confirmation letter to the exchange need not be a press release, so silence is not evidence either way. **One cell on this row needs reading with its date in hand.** The twelve-month column improves **19.4** points, −32.3% to **−12.9%**, the largest column move this page has recorded, and 17.0 of it is the base: **28 October 2025**, the session the designation took effect and the shares fell **19.45%** in a day, has just rolled into the denominator. A cell improving nineteen points on an unresolved listing question is a calendar artefact and nothing else. **The silence broke the same day, and it is the most informative thing Nidec has published on this question.** “Notice Concerning the Status of Efforts to Resolve the Disclaimer of Opinion”, released **28 October in Kyoto**, states that the company _aimed_ to file the restated reports for FY3/26, FY3/25 and prior periods, the quarterly reports and the amended results **by 28 October 2026** and has **not** done so: its response to the auditor’s requests on identifying and evaluating the officers and employees involved is incomplete and the auditor’s procedures are unfinished. The confirmation letter, though, is not late. Rule 503(2) of the Listing Regulations requires the document “promptly after one (1) year has elapsed” rather than on the day, and Nidec says it **plans to submit the Written Confirmation of Internal Management System on 30 October 2026**, after which the exchange examines it **“with the aim of lifting the Security on Special Alert designation”**. That is the company’s wording and it is the first time the stated direction on this trigger has been lifting rather than delisting. It also reports the internal management system, including a review based on the Executive Responsibility Investigation Committee’s report, “has been completed and has begun operation”. **The hold-up is people, and the release sizes it.** As of 30 September, **77 individuals across about 50 locations** were under discussion with the auditor as questionable signatories of the Management Representation Letter or candidates for reassignment: **five executive officers in Japan, eight overseas, 23 employees in Japan, 41 overseas**. The plan was to finish and have the auditor done **by 27 October**; it was not. A further **132 candidates** were set for email screening and follow-up, with **120** already inside existing investigations. On timing: “the conditions for completing additional procedures and specific audit completion dates have not been confirmed”, with Nidec’s own procedures **expected to complete by end-November 2026** and the release stating plainly that this “does not indicate the time of completion of the audit… or the time of submission… of the restated report”. Two firm facts: audit contracts concluded **23 October** for the restated FY3/26 report and **24 October** for FY3/27, and Q1 and Q2 FY3/27 to be closed in parallel on a schedule still “under consideration”. **29 October: +3.44% to ¥1,865** on **0.62 times** average volume, its **highest close since 14 October**, a second consecutive gain near three and a half per cent on under two thirds of normal turnover, four sessions after the ten-year low of 23 October. A separate release, “Notice Regarding Organizational and Personnel Changes”, is dated **29 October**; it is not yet readable in English and nothing here rests on it, but personnel remedial measures being the outstanding audit item makes it the next document to read.

Motor Servo + drive + control

### Yaskawa Electric

6506.T

**20 November.** Yaskawa closed at ¥3,180 (−8.49%) on 1.24 times its prior 20-session average volume. Yahoo’s daily close is null; its 15:30 JST quote metadata supplies the completed auction close.

**19 November.** Yaskawa closed at ¥3,475 (−0.86%) on 0.58 times its prior 20-session average volume.

**18 November.** Yaskawa closed at ¥3,505 (+0.95%) on 0.88 times its prior 20-session average volume.

**17 November.** Yaskawa closed at ¥3,472 (−2.50%) on 0.94 times its prior 20-session average volume.

**16 November.** Yaskawa closed at ¥3,561 (−6.07%) on 0.75 times its prior 20-session average volume. October guidance still sets the second-half operating profit test.

**13 November.** Yaskawa closed at ¥3,791 (−0.86%) on 0.68 times its prior 20-session average volume. October guidance still sets the second-half operating profit test.

One of very few firms that can already supply precision motion at industrial scale, now pointed at humanoids: it acquired **100% of Tokyo Robotics**, whose whole-body torque sensing and impedance control fill a genuine capability gap, and ships MOTOMAN Next with NVIDIA GPUs for autonomous control.

But the profit claim in the source post is backwards. For the fiscal year ended February 2026, revenue edged up while **operating profit fell more than 5%**, with pre-tax and net profit down harder. The growth is in the _forecast_, and on 9 October the forecast itself was revised: FY2/27 guidance now reads **¥600bn revenue** (+10.7%) and **¥47.5bn net profit** (+34.8%), with operating profit cut to **¥57.5bn** from ¥60bn. Up just **4.8%** over twelve months after the 19 October session and **44.8% below its high**, second only to Schaeffler — and that twelve-month cell gained 2.7 points on a session the price fell five yen, because its year-ago base rolled from 16 onto 17 October 2025, it is the clearest example here of a stock priced on guidance rather than results. Its year-to-date return turned negative on 1 September and kept going: a 4.4% fall on 8 September wiped out three modest up sessions, a 1.6% loss on 10 September took it below its April low, a further 3.0% on 11 September took it to ¥4,301, then two yen lower on 14 September and three yen lower again on 15 September — three consecutive closes at the weakest since 3 April, and the manner of them was the point: not a capitulation but a slow bleed of two and three yen a session while the rest of the tier drifted higher. That run ended on 16 September, +1.5% to ¥4,360, its first gain in five sessions — then lasted exactly one day, −1.4% to ¥4,299 on 17 September, before the rate decision took it +2.0% to ¥4,383 on 18 September, and the reopen after the three-day holiday added **+2.8% to ¥4,507** on **24 September**, its highest close since **7 September**. Then it gave a little back, **−0.6% to ¥4,480** on **25 September** — and on that session the Nikkei rose 1.30%, so this was the widest relative loss of the six Tokyo names here, nearly two points behind the index. Then it turned: −0.6% to ¥4,453 on 28 September, **+1.7% to ¥4,528** on 29 September, the best of the six that day, and **+2.3% to ¥4,633** on **30 September**, its highest close since **7 September**, on an index up 1.94%. Three more have extended it: **+3.2% to ¥4,779** on 1 October, **+0.6% to ¥4,806** on 2 October, and **+2.6% to ¥4,931** on **5 October**, where the close and the session high are the same figure, its **highest close since 26 August**. Six consecutive gains took the month from 9.5% lower to 5.8% higher, and then the tier-wide de-rating of **8 October** removed a chunk of it, **−3.17% to ¥4,789**. It went into its results at **¥4,783**, after a **−0.13%** session in which it opened at its low of ¥4,703 and recovered all day on 4.37m shares.

**The first-half numbers landed after the close on 9 October, and they cut the one line that measures the business.** Revenue for the six months to 31 August was **¥285,598m, up 9.8%**, with operating profit **¥22,058m, down 5.5%**, profit before tax ¥25,963m (+3.0%), profit attributable to owners ¥18,988m (+4.1%) and basic EPS ¥73.21. The full-year forecast was formally revised for the first time since 10 April, and in opposite directions: revenue from **¥580,000m to ¥600,000m**, operating profit from **¥60,000m to ¥57,500m**, profit before tax to ¥65,500m, net profit to ¥47,500m, EPS to ¥183.11. The company names both causes: a weaker yen on revenue, and the new ERP system’s effect on production and shipments through the half. It moved its own currency assumptions for September to February with it, the dollar from ¥145.0 to ¥155.0 and the euro from ¥170.0 to ¥180.0. The dividend forecast is unchanged at **¥72**.

**Two things in that filing are better news than the revision, and one of them is a photograph.** Second-quarter **orders** were **¥185.4bn, up 38% year on year and 13% on the quarter**, which the company calls a **record-high quarterly order intake**, and the segment split is where the humanoid thesis gets tested rather than asserted. **Motion Control** orders, the servo and drive line this page actually cares about, rose **56%**; **Robotics** rose **33%**; System Engineering fell 6%. Yaskawa attributes the growth explicitly to **data centre and semiconductor-related investment**, and the same split runs through the half: Motion Control revenue **¥137,286m (+21.7%)** on operating profit **¥15,973m (+32.8%)**, against Robotics revenue **¥118,404m (−0.7%)** on operating profit **¥5,974m, down 43.3%**, the fall driven by the ERP system’s effect on Japanese sales and by European restructuring charges. So the segment named after robots nearly halved its profit while the segment selling into the AI build-out grew a third. Anyone holding this name for humanoids is currently being paid by data centres. **The second item is on the second-half measures slide**, under Other Topics: an **“Actuator for humanoid robots (under development)”**, beside a commitment to accelerate data accumulation through field validation and robot design development _toward the commercialization of humanoid robots_. That is a component maker building the integrated actuator itself rather than selling parts into someone else’s, which is the move Nidec and Harmonic Drive have both already made, and it is the first time Yaskawa has put it in a results pack. It carries no grade, no date, no customer and no volume, so it is a statement of direction and not a number. It is still the most on-thesis disclosure this name has produced.

**The market has now voted, and mildly.** Tokyo was shut for Sports Day, so the first session to price any of it was **Tuesday 13 October**, and the stock closed **+1.19% at ¥4,840** on **11.72m shares** against 4.37m, a **2.7-fold** volume increase and the heaviest turnover this page has recorded on the name. On a day the Nikkei fell **1.82%** and the two gearbox rows beside it fell 4.7% and 6.6%, a gain of any size is a clear relative verdict: three points better than the index, and the only tier-one name in Tokyo to rise meaningfully. _Jefferies_ analyst **Sho Fukuhara** upgraded it to **Hold from Underperform** with a **¥5,300** target, per Investing.com, on the order beat and management’s commentary. Read the rating and not that article’s arithmetic: it is the subject of the claim audit’s newest row, having reported the operating-profit revision at ten times its actual size and the session at an intraday price. The drawdown comes in to **38.9%**.

**The arithmetic the revision leaves behind is the part to hold on to.** Operating profit is the only figure revised downwards, and it still carries a full-year guide of **+21.5%**. With ¥22,058m banked, that implies **¥35,442m** in the second half against ¥23,973m a year earlier, or **+47.8%**; on net profit, ¥28,512m against ¥16,993m, or **+67.8%**. A stock priced on a forecast has just moved more of that forecast into the half that has not happened, while the revenue raise alongside it is an exchange rate rather than a volume. The headline net figure went up, from +33.4% to +34.8%, which is why this revision reads better in a summary than in the filing. Tokyo is shut on Monday 12 October for Sports Day, so the first session to price it is **Tuesday 13 October**.

1H op **−5.5%** FY2/27e rev **¥600bn** FY2/27e op **¥57.5bn** 2H op implied **+47.8%** 2Q orders **¥185.4bn** Motion Ctrl orders **+56%** Robotics op **−43.3%** Off high **−59.8%**

**21 October: −1.71% to ¥4,204 on 1.06 times average volume, the only tier-one name in Tokyo to fall on a day the index rose 0.85%.** It is the lowest close since **31 March 2026**, and the drawdown widened to **46.9%**, the second deepest in tier one behind Schaeffler. The twelve-month cell improved a point even so, −4.4% to −3.4%, on a year-ago base rolling ¥4,474 to ¥4,352, so the column moved opposite to the price for the second run running. Nothing is dated to the session.

**Watch**Reported, priced, and now the back half. The revised guide needs **+47.8%** operating profit growth in the six months to February against a first half that delivered **−5.5%**, so the question is no longer the guidance but whether the ERP disruption it blames is behind the company or still in front of it. Third-quarter results, in January, are the first check. Treat the revenue raise as currency until a volume line says otherwise, and treat the record order book as a data-centre number until the segment split says otherwise: Motion Control orders up **56%** on semiconductor and data-centre demand is a genuinely strong figure about a market that is not humanoids. The humanoid line to follow here is no longer an order disclosure but a product: the **actuator under development** shown in the second-half pack, and whether it acquires a grade, a date or a customer at the January results. **26 October took the price below a level it had held for eleven months.** **−1.04% to ¥3,980** on average volume, its **lowest close since 27 November 2025**, in a Tokyo session where most constituents rose and the index fell on chip weakness. That widens the drawdown half a point to **49.7%**, within half a point of Schaeffler’s page record, and takes the year to date to **−16.3%**. The columns moved further than the price again: the four-week cell to **−11.2%** on a base rolled to 25 September, two of those three points base rather than price, and the twelve-month to **−12.5%**. The revised FY2/27 arithmetic is unchanged and unhelped by this: a back half that has to deliver +47.8% operating profit growth, with the stock now pricing the first half rather than the forecast. **27 October added 2.76% to ¥4,090** on **0.68 times** average volume, its highest close since 21 October, which lifts the twelve-month cell 3.2 points to **−9.3%** and brings the drawdown in 1.4 points to **48.3%**. A rebound on two thirds of average turnover, four sessions after a close that was the lowest since November 2025, is the selling pausing rather than the guidance being believed. **28 October: +2.35% to ¥4,186 on 0.64 times average volume, the lightest turnover anywhere on the page, and its highest close since 21 October.** Nothing was filed. The row is **3.79%** above its 22 October close and still **0.43%** below the ¥4,204 it printed the day before the rout, so this is the one tier-one gearbox-or-motor line that has not yet repaid the fall. The drawdown narrows **1.2** points to **47.1%** and the twelve-month cell gains **5.4** to **−3.9%**, of which 3.3 is a year-ago base cut ¥4,509 to ¥4,354. **One circulating item names this company and is three months old.** A daily market brief dated **27 October** reports, in the present tense and undated, that Fanuc and Yaskawa “are teaming up” with other Japanese firms on robot AI using NVIDIA’s technology. The tie-up is real, it is the **Cosmos Coalition**, and it was announced on **15–16 July 2026** with no binding commitment and no money disclosed. A name rising 2.35% on the lightest volume on the page is not a market reacting to it. The claim audit carries the row. **4 November took this line to within a point of a twelve-month low, and the back half of the guidance is what it is pricing.** **−1.99% to ¥3,839** on 1.14 times average volume, its **lowest close since 21 November 2025**, which widens the drawdown to **51.5%**, the second-widest on the page behind Schaeffler, and takes the year to date to **−19.3%**. The twelve-month cell fell **−7.7% to −12.2%**, and most of that is the base rather than the session: the year-ago line stepped up 3.0% onto 4 November 2025’s ¥4,374. Nothing was filed. The **9 October** revision still stands as the thing to measure against, and it needs **+47.8%** operating-profit growth in the back half against a first half that delivered −5.5%.

Perception On-robot vision, no cloud

### Ambarella

AMBA

**19 November.** Ambarella closed at $58.93 (+7.85%) on 2.01 times its prior 20-session average volume.

**18 November.** Ambarella closed at $54.64 (+2.05%) on 0.58 times its prior 20-session average volume.

**17 November.** Ambarella closed at $53.54 (−2.71%) on 1.22 times its prior 20-session average volume.

**16 November.** Ambarella closed at $55.03 (−8.47%) on 0.53 times its prior 20-session average volume.

**13 November.** Ambarella closed at $60.12 (−1.54%) as the PHLX Semiconductor Index fell 3.43%. The [issuer-confirmed results call](<https://investor.ambarella.com/node/17436>) is 24 November at 13:30 PST. Bid talks remain press-sourced.

The only US-listed name and the only one with a clean, current, verifiable set of numbers. **CV7** launched at CES on 5 January 2026 — an 8K edge AI vision SoC for multi-sensor perception across robotics, drones, industrial automation and automotive. Fiscal 2026 revenue was **$390.7m, up 37.2%**, with fiscal 2027 growth guided at 10–15%. Q2 FY2027, reported on 3 September, came in at **$108.1m (+13.2%)** with non-GAAP EPS of $0.18 — both a shade ahead of consensus — and Q3 guided to $115–124m.

The robotics pipeline is disclosed rather than implied: 15+ robotics design wins with lifetime revenue above $100m, and 30+ customers in the pipeline. The honest caveat is that robotics is still the smaller part of a business anchored in security cameras and automotive — and the deceleration from +37% to a guided 10–15% is the number bulls have to explain.

Nor did the follow-up. Fermi Wang appeared at Citi's Global TMT Conference on **9 September** and was blunter there than on the call: robotics has “a couple design wins,” but the volume is not going to change the financial performance materially, and humanoids are, on his account, harder from a technology standpoint than a level-5 autonomous car. He also named the constraint that is not in the guidance — memory. Customers, he said, could not get allocation commitments from their memory suppliers even for November, which he called the biggest uncertainty and expected to run into 2027. For a fabless designer that is a customer-side supply problem, not a demand one, but it lands on the same revenue line.

The Q2 call moved the narrative without moving the disclosure. Fermi Wang said edge AI revenue “reached record levels” on balanced sequential growth in Auto and IoT and very strong growth from the 5nm CV75 and CV72 SoCs, and the company raised its five-year serviceable-market forecast for edge AI and _physical_ AI, citing higher-value products and new distribution through Macnica and CapGemini. Margins went the other way: non-GAAP gross margin of **59.3%** against 60.5% a year earlier, with the Q3 guide at 59–60%. Still a GAAP loss — $6.7m, or $0.15 a share — against $8.2m of non-GAAP net income, on $272.3m of cash.

FY26 rev **$390.7m** Growth **+37.2%** FY27e **+10–15%** Q2 FY27 rev **$108.1m** Design wins **15+**

**20 October: +0.73% to $63.22 on 0.62 times average volume, and this time the sector went the other way.** The PHLX Semiconductor index _fell_ 0.29% the same session, so seven tenths of a per cent against it is the first move on this card in a fortnight that beta does not account for — and seven tenths of a per cent on well under average turnover is not a deal being priced either. A third reported bidder entered the sales process last week, after the **31 July** Financial Times report of NXP talks and the **9 October** Qualcomm speculation. Both October items trace to unverified market alerts repeated by aggregators, no named party has filed or said anything, and one of them characterises the process as near completion, which is a claim about a contest none of the parties has acknowledged. They are carried here as circulating talk. This card’s standing read was that Ambarella’s falls are sector beta with the robotics line doing none of the work; applied to a rise, the same test gives the same answer.

**23 October: +5.35% to $66.50 on 1.15 times average volume, the best session on the page and twelve times its own sector.** That is its **highest close since 9 October** and **+7.78%** on the week, against a PHLX Semiconductor index that rose **0.44%**. It inverts Thursday exactly, when the name fell 4.36% into a 6.72% sector rout, taking about two thirds of it. Two consecutive sessions have now gone against this card’s standing read that Ambarella’s days are mostly sector beta, in opposite directions, and neither carried a disclosure. The drawdown narrows **3.5** points to **31.2%**, the tightest on the page. The only live story is still the one with no document behind it: the **21 October** report naming **Nvidia** as the third party evaluating the company alongside NXP and Qualcomm, said to be working with Goldman Sachs, traces to a _Betaville_ alert rather than a filing or a named source, and nobody involved has commented. Carried here as dated market chatter, not as a transaction.

**30 October: −2.26% to $63.12 on 2.30 times average volume, and the M&A story is now reported as exclusive.** The report has the board choosing **NXP** after weighing several offers, which would retire the Qualcomm and Nvidia strands above. It is still not a document: Ambarella’s only 30 October filing is a third-party **Schedule 13G/A**, NXP filed six Form 4s and nothing else, and neither company has commented. The participation is the useful detail rather than the price. **3,274,100 shares** against a twenty-session average of 1,422,615 is the heaviest turnover on this card since **23 September**, on a move of barely two per cent, which is a tape reading a story it has read three times before. Lowest close since **19 October**, and **34.7%** below the high. One aggregator put the fall at 5.5%: it was 2.26%, and the claim audit carries it.

**Watch**Robotics revenue disclosed as its own line — and Q2 did not do it. “Physical AI” now appears in the serviceable-market framing, but there is still no robotics number to hold anyone to, so this name remains a narrative read rather than a volume read. Then the tape did something the disclosure did not explain. Ambarella rose **8.4% to $68.80 on 9 September**, the largest single move on this page in weeks and its highest close since 31 August. The sequence is worth recording, because the obvious attribution is wrong: the stock opened flat at $63.07 against a $63.48 previous close, ran roughly 10% in the first thirty minutes to $70.28 by 10:00, then faded to $66.43 by 13:00 — before the Citi appearance began at 13:55 — and closed at $68.80 on heavy final-hour volume. The move was made in the opening half-hour, hours ahead of the only scheduled event of the day, and the stock actually _fell_ through the appearance itself. It was not a sector move either: the semiconductor ETFs added 0.7% and 0.1%, and NXP, still the reported suitor, closed lower. No company announcement accompanied it — and a sixth of it came back the next session, **−1.4% to $67.87** on 10 September, against a semiconductor index down 2.7%. On 11 September it did nothing at all, a single cent lower at **$67.86**, while the semiconductor index rose 1.8% — the one name on this page that joined neither the rare-earth selling nor the chip rebound. Then on **14 September** the whole unexplained 9 September gain came back at once, **−5.7% to $63.98**, its worst session since 16 July and within fifty cents of the 8 September close it started from; 15 September then steadied it, +0.6% to $64.33 — before **16 September took it to $62.77, down 2.4%**, its lowest close since 17 July. This time the attribution is easy and it is not about robots: public calls from several AI-lab executives for a slower pace of frontier development took the semiconductor complex down with them — SOXX −5.6%, Nvidia −3.4%, Broadcom −4.8%, AMD −4.4% — and Ambarella tracked the sector almost exactly. That is the useful fact about this position. On the way up, its 9 September move was idiosyncratic and unexplained; on the way down it was pure sector beta. That last session broke the pattern and was worth marking — the semiconductor index _rose_ 0.6% on 16 September while Ambarella fell 2.4% — and then three sessions ran the other way just as hard: **+6.6% to $66.91** on 17 September, the day after the Federal Reserve's first hike since 2023, −1.1% to $66.15 on 18 September, and **+4.1% to $68.85** on 21 September with SOXX up 4.9% on AI optimism ahead of the 24 September state dinner — its highest close since 31 August. Three sessions, **+9.7%**, and again not a word from the company. It gave most of that back at once, **−2.6% to $67.06** on 22 September, then steadied, +0.8% to $67.62 on 23 September — the only name on this page to rise on a session that took the whole minerals tier down — then added **+3.3% to $69.86** on **24 September**, and then **+4.0% to $72.68** on **25 September**, its **highest close since 21 August**. Neither had a company announcement attached, and the second is worth one line of method: the obvious candidate — the Developer Zone expansion on Google Cloud, with the Gemini Enterprise tooling, the ZEDEDA fleet integration and the Ultralytics YOLO collaboration — is dated **15 September 2026**, seven sessions earlier, and aggregator write-ups published on the 25th that place it vaguely “in September” are how a stale release gets re-read as a same-day cause. The sector took about a quarter of it, SOXX +1.2%. Which settles the read on this position: down it was sector beta, up it is sector beta, and the robotics line inside it is doing none of the work in either direction. The next session took all of that back and then some. On **28 September** Ambarella fell **5.5% to $68.68** against SOXX −2.1% and SMH −1.1% — again roughly three times the sector, again with nothing published by the company, and this time in the opposite direction, which is the cleanest possible confirmation of the read above: whatever is moving this position is not the robotics line and is not company news. **29 September** then did nothing at all with it, **+0.1% to $68.73** against a semiconductor complex up **1.2%**, so the name-specific leg was neither extended nor taken back. Three more sessions did very little either way: **+1.1% to $69.51** on 30 September, **+1.1% to $70.30** on 1 October and **−2.4% to $68.65** on **2 October**, which gives back the two days before it and leaves the name roughly where the 28 September fall put it. That left Ambarella, as of 2 October, **9.2%** higher over a month on a base sitting on 4 September's $62.89, **29.0%** off its high, with a twelve-month column that fell seven and a half points on the 22 September session purely because the year-ago base rolled onto a +7.1% day, recovered to −12.9%, and then read −20.3% after another base step. It remains the only tier-one name whose drawdown is set by the AI trade rather than by industrial demand. Then **9 October** broke the pattern this card has spent six weeks establishing, and it took a rumour to do it: **+7.22% to $67.43** on **2.02m shares**, the heaviest volume in a month, against SOXX **−0.7%** and SMH −0.6%. Up was sector beta and down was sector beta; this was neither. The cause in circulation is a **Betaville “uncooked”** alert, that outlet’s lowest-confidence tier, putting **Qualcomm** with advisers on a possible bid and Ambarella with bankers on talks involving Qualcomm and **NXP**, described as “at an advanced stage”. Nothing there is confirmed and none of the three companies has commented. What is on the record is the _Financial Times_ report of **31 July 2026** that NXP was in talks at above **$3bn**, the session that carried the stock **+16.1% to $86.00**. So this position now carries a second rumoured suitor and still no disclosure, and the M&A overhang has become the dominant term in its price rather than the robotics line. The CPI session then took it back down: **−4.42% to $61.46** on **14 October**, its **lowest close since 23 April**, on **0.88 times** average volume against SOXX −4.50% and SMH −3.98%, which is the sector almost exactly and is the read this card has settled on. **15 October then added 3.08% to $63.35** on average volume, rising while the semiconductor indices fell again, which is the first session in a while that the sector read does not cover. **16 October put the sector read back in charge:** **−2.60% to $61.70** on **1.57 times** average volume, against SOX **−3.83%**, SOXX −4.14% and SMH −3.41%, so roughly two thirds of its sector in a week the Associated Press records as the worst since June for the S&P and the Nasdaq. On the current close the name reads **−1.7%** over four weeks against a 16 September base of $62.77, **−12.9%** year to date, **−26.6%** over twelve months against a 16 October 2025 base of $84.09, and **36.2%** off its high, still the narrowest drawdown in tier one after Nabtesco. Ambarella is still the name closest to breaking robotics out, which would make it the cleanest read on real humanoid volume anywhere on this list — the next chance is Q3, in December, assuming it is still independent enough to report one. **27 October: +0.45% to $67.12 on 1.09 times average volume, a third consecutive gain and its highest close since 9 October.** The S&P 500 rose 0.60%, so this is the one line on the page that simply tracked its market on a session when everything below it moved six to seventeen per cent. That is the useful property of this row rather than a disappointment in it: at **0.20** against REMX it is the least rare-earth-sensitive name here, and it behaved like it. The drawdown comes in **0.3** points to **30.6%**, the tightest anywhere on the page. Still no robotics revenue line, and the 1m cell read **−7.6%** that run against a base that had not moved in two runs, so for once that column was clean.

**2 November is the first session in a fortnight the bid report does not explain, because the chip tape explains it better.** **−2.30% to $61.67** on **0.79 times** average volume, 1,203,600 shares against 1,518,300, between **$61.18** and $65.12, its **lowest close since 14 October**. The PHLX Semiconductor index fell **3.37%** the same session, half again as far, with the Nasdaq Composite down 1.07%. The drawdown widens to **36.2%** below the $96.69 high, the year to date to **−12.9%** and the four-week cell to **−10.2%** on a base rolled onto 2 October’s $68.65. Three sessions after the exclusivity report put 3.27m shares through the tape at 2.30 times average, turnover is back below average and the exclusivity is still a report rather than a document. A semiconductor name trading with the semiconductor tape, and under it, is what this row looked like before July.

03 Claim audit — the posts, and the coverage

Wrong

Lynas and the US had finalised a four-year, US$96m rare-earth oxide purchase agreement on 16 March 2026, as previously described in this watchlist’s Seadrift audit entry

Lynas’ [March-quarter ASX report](<https://announcements.asx.com.au/asxpdf/20260421/pdf/06ypfwgh3p6xmg.pdf>) describes a binding letter of intent to finalise a supply agreement. It describes the approximately US$96m previously allocated to the Texas plant as intended for oxide purchases over four years. The earlier audit entry has been corrected to reflect the status disclosed on that date.

Wrong

TDK up 4.5% YTD and down 9.5% over one year, and Aichi Steel down 10.0% over one year, as shown in the 16 November company cards

Those card figures were stale. At the 16 November Tokyo close, TDK was down **0.4% YTD** and **13.2% over one year**, and Aichi Steel was down **8.6% over one year**. The Tier Three table already carried those values. The cards now use the same 17 November data as the table. [TDK history](<https://finance.yahoo.com/quote/6762.T/history/>) and [Aichi Steel history](<https://finance.yahoo.com/quote/5482.T/history/>).

Wrong

MP at 0.57 to REMX and 0.22 to robotics, TDK at 0.56 to robotics, and the basket at 0.37 to REMX, as repeated in the 15 November tier commentary and captions

Those figures were stale for the 103 returns through 13 November. Exchange-local daily and native weekly closes both give MP **0.549548** to REMX and **0.238331** to robotics, TDK **0.572185** to robotics, and the basket **0.357566** to REMX. The charts and table coefficients already reflected this window. [Yahoo Finance history](<https://finance.yahoo.com/quote/MP/history/>).

Wrong

Schaeffler at 0.62, TDK at 0.56 and Aichi Steel at 0.14 to the robotics basket, as shown on this page’s 14 November company cards

Those card values conflicted with the charts and tables for the same window. Both exchange-local calculations through 13 November give **0.610120**, **0.572185** and **0.148566**. The cards now show **0.61**, **0.57** and **0.15**. Schaeffler’s REMX correlation is **0.46** on this window. The 103 returns and the headline 0.3576 coefficient stand. [Yahoo Finance history](<https://finance.yahoo.com/quote/SHA0.DE/history/>).

Wrong

“Perpetua secured a $2.9bn EXIM loan” — this page’s Perpetua card

The [13 November Form 10-Q](<https://www.sec.gov/Archives/edgar/data/1526243/000110465926127248/ppta-20260930x10q.htm>) distinguishes the EXIM board’s 21 May approval from the outstanding closing conditions and documentation. Perpetua now expects progress to closing in Q1 2027. The final amount and timing remain uncertain.

Wrong

“\[Nabtesco\] bought Slovak cycloidal maker Spinea in 2023” — this page’s Nabtesco card

[Timken’s closing announcement](<https://news.timken.com/Timken-Completes-Acquisition-of-Spinea>) identifies Timken as the buyer and dates the completion to 31 May 2022.

Wrong

“13 November \[is\] the first utilisation read since the RV capacity doubling” — this page’s Nabtesco card

The [August Q2 briefing](<https://www.nabtesco.com/cms/wp-content/uploads/Results_Briefing_Material_for_FY2026_Q2e.pdf>) already showed the same 1.15m-unit annual capacity and reported utilisation at Tsu (85%) and Changzhou (100%), with Hamamatsu ramping. The [Q3 briefing](<https://www.nabtesco.com/cms/wp-content/uploads/Results_Briefing_Material_for_FY2026_Q3e.pdf>) updates those figures.

Wrong

“US defence sourcing rules bar Chinese-origin NdFeB magnets, tungsten, molybdenum and tantalum across the supply chain” — this page’s 1 January 2027 catalyst

[DFARS 225.7018](<https://www.acquisition.gov/dfars/subpart-225.70-authorization-acts-appropriations-acts-and-other-statutory-restrictions-foreign-acquisition>) does not list molybdenum as a covered material. The January expansion reaches earlier stages for the listed materials, including the full NdFeB chain. The rule applies to covered defence acquisitions and has exceptions, including some off-the-shelf items.

Wrong

“FY3/26 landed at ¥59.6bn revenue (+7.0%) with operating profit down 94.4% year on year” — this page’s Harmonic Drive card

The revenue is correct. The [13 May FY3/26 results](<https://finance-frontend-pc-dist.west.edge.storage-yahoo.jp/disclosure/20260513/20260513529667.pdf>) report operating profit of **¥2.567bn**, up from **¥6m** in FY3/25. The **94.4% decline** belongs to [FY3/25](<https://www.hds.co.jp/Portals/0/files/english/ir/data/Consolidated%20Financial%20Results/tanshin_20250514.pdf>).

Wrong

“The €400m stands on a preliminary release” — this page’s catalyst summary as of 10 November 2026

The [20 October Taicang release](<https://www.schaeffler.com/en/media/press-releases/press-releases-detail.jsp?id=88222851>) already repeated the approximately **€400m** order figure, as the company card recorded. The [10 November release](<https://www.schaeffler.co.in/en/news-media/press_releases/press_releases_detail.jsp?id=88226241>) repeats it again. The 3 November nine-month statement omitted the amount. The two press releases describe order intake and give no humanoid revenue figure.

Wrong

“NJDCY has not printed since February” — this page’s method note as of 4 November 2026

The OTC history shows trades in late October and November. The **9 November** session ended at **$2.99**, following **$2.50 on 6 November**. Yahoo Finance and the [MarketScreener OTC history](<https://www.marketscreener.com/quote/stock/NIDEC-CORPORATION-120787337/quotes/>) agree on both closes. The latest print updates the stale label. The Tokyo primary remains the source for the watchlist’s prices and correlation.

Wrong

“The first magnet invoice, which on the company’s own account does not yet exist” — this page’s USA Rare Earth card as of 4 November 2026

The [9 November earnings exhibit](<https://www.sec.gov/Archives/edgar/data/1970622/000197062226000066/exhibit991-earningsrelease.htm>) dates the first Stillwater commercial magnet sales to **September 2026**. The earlier Q2 filing supported the statement for Q2. Carrying it into November without a date overstated what was known. The new release identifies the initial product as sintered NdFeB block magnets and leaves the sales value undisclosed.

Wrong

“Schaeffler shares fell **a little over 2%** in Frankfurt” — published 3 November 2026

Wrong as a session, and it is the oldest species in this audit: an intraday figure carried with no session label in a piece about the day’s announcement. **Xetra closed 3 November 5.19% lower at €5.66**, from €5.97, and the **Frankfurt** line the report names closed **5.47%** lower, so the quoted move is under half the fall on either German venue. The figure is traceable to the open rather than invented: the first five-minute bar of the session carries **€5.82**, 2.5% below Monday’s close, and the stock fell steadily from there to a €5.62 low. The piece is otherwise accurate on the restructuring programme it reports, which is the usual pattern here, a correct slow-moving account beside a wrong fast-moving number. A wire filing before the 17:30 auction cannot report the day, and this one does not say that it is not.

Wrong

USA Rare Earth’s third quarter lands on 5 November 2026 — “verified”

Wrong, and the label is the fault rather than the guess. bullstory.io carries the date under the headings “Upcoming verified earnings release” and “Verified data”. The company’s own release, datelined Stillwater **2 November 2026**, sets results for **after the close on Monday 9 November 2026** with the call at **4:30pm ET**. Two other trackers carried **5** and **13 November** on the same morning, and one of them has since corrected to the 9th. An estimated date is ordinary; an estimated date printed as verified is not.

Checks out

Nabtesco operating profit rose 60% year over year

Correct. FY2025 operating profit +60.3% to ¥20.7bn on revenue of ¥307.9bn.

Checks out

Ambarella's CV7 launched at CES 2026 for on-robot perception

Correct. Announced 5 January 2026, an 8K edge AI vision SoC explicitly targeting robotics and industrial automation.

Checks out

MP Materials revenue jumped 89% to $108.5m

Correct and current. Q2 2026, reported 7 August — up from $57.4m, with NdPr output +41% and adjusted EBITDA turning positive.

Half right

Schaeffler is signing partnerships with multiple humanoid makers

True — Hexagon Robotics and Humanoid, both confirmed. But the ticker given, SFHLF, is **SAF-Holland**, a truck-parts supplier. Schaeffler is SHA0.DE or ADR SFFLY. Buying the quoted ticker buys the wrong company.

Half right

Nidec has ¥2.6 trillion in annual revenue and the scale to dominate

The revenue figure is right (the post's "$2.6 trillion yen" is a units error — it is ~$17bn). Omitted: an accounting investigation opened in September 2025 whose final report landed on 17 April 2026 and whose restatements are still not filed, results delayed again on 5 August 2026, a separate quality-misconduct investigation whose report was published on **4 September 2026** and which found falsified test results and shipments of unauthorised items across multiple bases (though no material financial impact), a shareholder demand on 5 August that directors be pursued for liability, and a second, joint demand from multiple shareholders on 20 August.

Stale

"Breaking: the Pentagon bought a $400,000,000 stake in America's only rare earth mine"

Real, but announced **10 July 2025** — thirteen months before the post. The DoD preferred-stock purchase plus warrant took it to ~15% of MP on an as-converted basis. It is not news, and it is already in the price.

Wrong

Yaskawa's operating profit rose roughly 70% in its most recent fiscal year

Inverted. For FY ended February 2026, operating profit **fell more than 5%**, with pre-tax and net profit down more. The ~70% figure resembles the forward net-profit guidance for FY2/27, not a delivered result.

Wrong

USA Rare Earth has commercial production expected by 2028

Out of date by two years on the plant, though this page over-corrected in the other direction and has now fixed it. USAR commissioned its Phase 1a magnet line at Stillwater in **Q2 2026** and says it has commenced commercial production, guiding to 600 mtpa by end-Q4 2026 — so the 2028-style timeline belongs to the Round Top mine, not to the magnet plant, and the post conflates two separate assets. But commissioning is not selling: the company states it has “not commenced sales of neo magnets.” See the dedicated row below.

Overstated

Perpetua controls the only major domestic reserve of antimony

Stibnite is the largest known US antimony resource and is expected to cover ~35% of US demand in its first six years, so the thrust is fair. Omitted: it is **predominantly a gold mine** — ~4.8Moz reserves, ~450koz a year — and first production is **2029**. Antimony goes into flame retardants and munitions, not magnets.

Out of date

Lynas is now building a US processing facility to fill the gap

It is not currently building one. On **16 March 2026** Lynas announced a binding letter of intent to finalise a US rare-earth oxide supply agreement. Its [March-quarter report](<https://announcements.asx.com.au/asxpdf/20260421/pdf/06ypfwgh3p6xmg.pdf>) says approximately US$96m previously allocated to the Seadrift, Texas facility was intended to fund four years of oxide purchases from existing facilities. The genuine 2026 development is that Lynas became the first commercial producer of separated dysprosium and terbium outside China.

Wrong

"Trump just signed a 15% tariff on Chinese materials"

The 6 August 2026 Section 232 proclamation puts a 15% tariff on **polysilicon** and its derivatives — a solar and semiconductor input, with minimum import prices attached, effective December. It is not a rare-earth tariff and not a broad materials tariff. It has essentially no bearing on any of the five minerals names listed under it.

Not investable

Proterial listed among "publicly traded material & magnet producers"

Proterial has been **private since October 2022**, when a Bain Capital-led consortium completed a ~$7.5bn take-private of Hitachi Metals and the shares were delisted. Three of the six magnet names in that summary — Proterial, Niron and Noveon — cannot be bought at all, and a fourth, Arnold, only through its parent. Screening lists routinely mix public and private companies without flagging it.

Overtaken

"Ambarella beats Q2 EPS, revenue in line — shares edge higher" (3 Sep 2026)

The beat is right: $108.1m against $107.75m expected, non-GAAP EPS $0.18 against $0.17. The share-price half was written off the _first_ post-market prints and did not survive the evening. Ambarella opened the extended session at **$64.50**, up 1.8% on a $63.38 close, then gave the whole move back to finish at **$61.24** — down 3.4%. A second wire, filed later, ran the opposite headline. Neither is a forecast; they are the same session read at different minutes, and only one of them is still true. Reaction headlines pinned to a single early print are not evidence of how a print was received.

Overtaken

"Japan Stock Market Today, September 8: Nikkei 225 gains 0.4% while Topix falls 0.6%"

A mid-session snapshot filed as the day's market report. The index was at **66,615** when the piece was written; it closed at **65,269.33**, down **1.70%** — so the headline has the day the wrong way round by more than two points. The article's own list of fallers was already arguing with it: Murata −4.8%, Taiyo Yuden −5.4%, Toyota −3.2%, Sony −2.7% against a supposedly rising index. It matters here because four of the six tier-one names price in Tokyo, and the index print is the tape their moves get read against. This is the second such figure this page has had to correct — the first was 19 August's widely repeated "Nikkei −2.64%" against an actual −3.16% — which is why index moves are checked against the index's own close and not against the coverage of it.

Context missing

"Trump's administration is cutting China out of the US materials supply chain"

Directionally real: [DFARS 225.7018](<https://www.acquisition.gov/dfars/subpart-225.70-authorization-acts-appropriations-acts-and-other-statutory-restrictions-foreign-acquisition>) expands the existing covered-country sourcing restrictions on **1 January 2027** to include earlier production stages. That is a defence-procurement rule, not a commercial-robotics one — a humanoid built for a warehouse is not covered.

Half right

“Rare earth stocks tumble on U.S.–China thaw hopes: USA Rare Earth sinks 4%, MP Materials drops 5%” (10 Sep 2026)

One fault stands; the other this page has had to withdraw. The figures are a mid-session snapshot filed as the day's move — the piece is stamped **12:41pm ET**, and by the close USA Rare Earth was down **5.98% to $16.04** and MP **5.49% to $51.32**. That half is still wrong. On the cause, this page overreached. It argued the driver was purely a rates event — August producer prices a tenth above forecast, a $6bn Treasury buyback under what dealers had positioned for, the ten-year at **4.94%**, silver −5.3%, the gold miners −3.5% — and that a summit date fixed since 23 July could not be news on the day. The macro leg was real and is not in question. The _next_ session settled the rest: on **11 September** the whole complex reversed, the S&P up 0.86%, gold up 0.6% and the gold miners up 1.1%, and the rare-earth names fell again regardless — REMX **−2.6%**, USA Rare Earth −3.0%, MP −1.6%, Lynas −2.2%. A liquidation that has already reversed cannot explain a second day of selling. So there is a sector-specific leg, it outlived the macro one, and ruling the coverage's driver out altogether was wrong. What survives is the narrower lesson, and it cuts both ways: a mid-session print is not a close, and a same-day cause asserted without a same-day disclosure is inference rather than evidence — including when the inference is this page's own.

Overtaken

“Nikkei 225 Falls Today, September 14: Japanese Benchmark Drops 1.7%” (14 Sep 2026)

The third mid-session index print this page has had to correct, and the second from the same outlet. The Nikkei 225 **closed at 63,492.99, down 0.81%**. The 1.7% figure — and the “below 63,000” and “six-week low” framing attached to it — belongs to the opening half-hour: the index bottomed at **62,810.29 at 09:25** and then spent the whole afternoon between 63,300 and 63,600. The piece is stamped 10:04 IST, about two hours before the Tokyo close, and its own third paragraph contradicts the headline, putting the index “around 63,364 points, down approximately 1.01%”. A Reuters-sourced wrap the same day carried **62,979.17** and −1.61%, also a morning level. The close is corroborated twice over: the two Tokyo-listed Nikkei 225 ETFs, 1321.T and 1330.T, fell **0.59%** and **0.69%**, which no 1.6% index day produces. It matters more than the arithmetic: four of the six tier-one names price in Tokyo, and “flat against an index down 0.8%” is a different observation from “outperformed an index down 1.7% by a point”. Always in the same direction — the mid-session number is the more dramatic one, and it is the one that gets filed.

Wrong

“Nikkei 225 Today, September 15: Japan Stock Market Rises 0.4% Above 63,700 As Tech Stocks Rebound” (15 Sep 2026)

The fourth mid-session index print this page has had to correct, and the third from the same outlet — two of them on consecutive sessions. The Nikkei 225 **closed at 63,484.10, down 8.89 points, −0.01%**. Flat, not up 0.4%, and the direction is wrong as well as the size. The piece is stamped **09:55:02 IST**, which is 13:25 in Tokyo, two hours and five minutes before the 15:30 close, and its body is explicit about what it is quoting: the index “stood at 63,735, up 242 points or 0.38%”. The index did trade there; it gave the whole gain back by the bell. The two Tokyo-listed Nikkei 225 ETFs settle it exactly as before — 1321.T closed **−0.03%** and 1330.T **−0.12%**, which no +0.4% index day produces. A second aggregator carried the same session as **63,850 and +0.56%**, also a level from before the close. Four of these now, from three different filers, and the failure mode has never once varied: an intraday quote is published under a headline that describes the day. For a page where four of the six tier-one names price in Tokyo, the index a reader is handed decides whether “flat” counts as strength or weakness — and on 15 September the honest answer was that Tokyo did nothing at all.

Wrong

“Nabtesco announced the acquisition of Harmonic Drive Systems, strengthening its precision gear portfolio for humanoid robot actuators”

No such transaction exists, and the truth runs the other way. The two companies did integrate their strain-wave gear businesses into a US joint venture in **2006**, and Nabtesco took Harmonic Drive to equity-method affiliate status in **2008** — but Nabtesco **dissolved** that relationship in **2021**, removing Harmonic Drive from its equity-method affiliates and booking a valuation gain on the shares. Both remain separately listed in Tokyo today; 6324.T and 6268.T are the two largest positions on this page's tier one and they are not related parties. Neither company has filed anything resembling a merger, and the tape agrees: Harmonic Drive traded **+0.3%** on the day the claim was circulating, which is not what a takeover of a ¥500bn company looks like. The claim traces to a single market-research “forecast report” page, which carries a second invented event alongside it — a Yaskawa–Siemens tie-up to co-develop servo gear reducers, for which neither company has issued a release. Two fabricated corporate actions on one page is a source-quality signal, not a pair of unlucky errors. It matters here because the whole structure of tier one assumes these are independent suppliers: if the gear duopoly had consolidated, the diversification argument on this page would be wrong.

Misattributed

“Nidec faces potential write-downs amounting to ¥250bn ($1.6bn) as a result” of the quality investigation (Sept 2026)

The number is real; the attribution is not, and the two Nidec investigations are being welded together. Nidec's **4 September** quality report states the opposite of this on its own terms: its evaluation of individual matters “revealed no issue with any material impact on the Company's consolidated and other financial statements.” The ¥250bn belongs to the _separate accounting_ strand, where Nidec discloses it not as a write-down taken but as an **amount subject to examination** for possible additional impairment, mainly on automotive-related goodwill and fixed assets. The figure genuinely booked from that strand is a **¥160.7bn** expected reduction in consolidated net assets. So a contingent exposure under review in one investigation is being reported as a realised write-down caused by the other. The confirmed count from the quality report is **844 cases** over about a decade, mostly unapproved changes to materials and processes, with falsified tests and country-of-origin mislabelling among them. Both strands matter here — but one is a governance problem with a filing deadline, and the other is a ¥250bn question mark over the automotive book, and a reader who merges them gets the size of neither right.

Overstated

“USA Rare Earth began filling customer orders for sintered NdFeB magnets at Stillwater in Q2 2026” — a claim this page itself carried

Commissioned, yes. Selling, no — and the company says so twice. The Q2 10-Q filed **10 August 2026** states that the Stillwater facility “has recently been commissioned and has commenced commercial production; _however, we have not begun generating revenue from_ sintered neodymium-iron-boron (‘NdFeB’) permanent magnets.” The risk-factor supplement filed on **15 September 2026** is blunter still: Stillwater and Blacksburg “are under development and are not yet completed, we have not commenced sales of neo magnets,” and “we have not realized any revenues to date from the sale of neo magnets, critical minerals or rare earth minerals.” The quarter's **$5.8m** of revenue, which reads on this page and elsewhere as evidence of magnet sales, is nothing of the kind: the 10-Q attributes it “all” to **Less Common Metals**, the Cheshire subsidiary acquired in November 2025, and its own disaggregation note says every dollar of company revenue “is derived from sales of casting and strip casting” — booked at a gross _loss_ of $1.6m. So the correct statement is narrow: a magnet line exists, is commissioned and is producing; no magnet has been sold, and the **600 mtpa** figure is guidance for end-Q4 2026 rather than an achieved run-rate. This page had the stronger version in three places and has corrected all of them. It matters because USAR is one of only two names here that make finished magnets, and the entire tier-two case rests on whether ex-China magnet tonnage is arriving or merely announced.

Context missing

“The robotics basket's correlation to REMX is 0.41” — this page, through 16 September 2026

The figure was real but under-specified, and under-specified is not reproducible. On the stated window — 104 ISO-week-aligned weekly returns to the week ended 11 September 2026, primary listings throughout — an equal-weighted basket _rebalanced weekly_, meaning the mean of the six names' weekly returns, gives **0.3889**. The same six names held _without_ rebalancing give **0.4068**. Both bucketing methods agree to four decimal places on each, so this is not a data problem; it is two different definitions of “equal-weighted basket,” and the page never said which it meant. The ambiguity surfaced because the published figure could not be rebuilt from the method note: swept across window lengths from 100 to 106 returns and end weeks 36 to 38, the rebalanced construction stays inside **0.363–0.398** and never reaches 0.41. A third check found a related slip — a two-year daily request yields at most **103** weekly returns, not the 104 this page claimed. This row then recorded that the window “now comes from a three-year request”, and on **6 October 2026** that part was itself shown to be false: the published figures reproduce only at **103** returns, so the stated fix was never in the numbers. The window was 103 returns and the page was corrected to say so throughout. This row then claimed it was **104** as of the 16 October window, on the reasoning that a Saturday run can count a week whose trading has finished, and that too was wrong: a two-year request rolls forward at _both_ ends, so a week gained at the close of the sample is a week lost at the start of it. The aligned count held at **104 closes and 103 returns**, capped by Schaeffler’s Xetra line, the shortest of the fifteen series. Corrected again on the **19 October** run, across the chart subtitles, the two table captions and the method note, which were the places still carrying 104. The page publishes the weekly-rebalanced figure, **0.39**, and names the construction. No conclusion changes at either value; what changes is that a reader can now reproduce it.

Wrong

“Japan’s JP225 rose to 65,665 points on 24 September 2026, gaining 0.99%”

That is a late-session print, not a close, and the distinction cost roughly a quarter of the day’s reported gain. The Nikkei 225 closed **65,513.99** on 24 September, **+0.76%** and 495.04 points above 18 September’s 65,018.95 — verified two ways, from the daily bar whose quote metadata is stamped 15:45 JST, and from the five-minute series whose final print is 15:30 at the same 65,513.99 with previousClose at 65,018.95. The index was indeed near 65,650–65,680 at 15:10–15:25 and reached **66,249.11** intraday, **+1.89%**; the closing auction then took about 130 points out in the last five minutes. This is the second Nikkei figure to fail this check, after a press report of −2.64% on 19 August against an actual −3.16%, and the third instance on this page of a widely-quoted number turning out to be a continuous-trading print rather than a settlement. Check an index against the index, and check it at its close.

Wrong

“Japan shares lower at close of trade; Nikkei 225 down 4.77%” (15 Oct 2026)

The **eighth** Nikkei index figure this page has had to correct for being a print taken before the bell, and the smallest of the eight, which is why it is worth being precise about what is wrong with it. The Nikkei 225 **closed at 64,425.95, down 3,318.34 points or 4.90%**, and that figure is not this page’s reconstruction: it is what the **index provider’s own daily table** publishes for 15 October. The article says in terms “at the close in Tokyo, the Nikkei 225 declined 4.77%”, and −4.77% is a level the index held at **15:20**, where the five-minute series prints **64,507.56**, ten minutes before the bell. The closing auction then took roughly **82** points out. The quoted level that travels with the piece, **64,510.00** on **−3,234.29**, comes from a live-updating sidebar widget rather than the body, and it is internally consistent with −4.77% while being inconsistent with the settlement. Two independent routes give the close as the provider does: the daily bar whose quote metadata is stamped **15:45:03 JST**, and the five-minute series whose final print at **15:30** is the same 64,425.95 against a previousClose of 67,744.29. A CFD-based tracker carried the same session as **64,795** and **−4.35%**, further from the cash close still, and it is a contract rather than the index. **The honest qualification is that this one does not change the read.** −4.77% and −4.90% tell the same story about the same day, which several of the earlier cases did not: a 1.7% headline on a 0.8% session and a +0.4% headline on a flat one both inverted the meaning outright. What is identical is the mechanism, and on this index alone it has now failed the same way eight times, from 19 August through 8, 14 and 15 September, 24 and 30 September and 6 October to here: a continuous-trading print published under a headline that says “at the close”. For a page where six of its listed names price in Tokyo, the index a reader is handed sets the baseline every one of them is measured against, so it gets checked against the index, at its close, every run.

Context missing

“China rare earth magnet exports drop 21% before trade talks” (circulating 21–22 September 2026)

The 21% is real and the framing is not: it is _US-bound_ shipments only. China exported **512 tonnes** of rare-earth permanent magnets to the United States in August, down from 647 tonnes in July, on customs data released 20 September. China’s _total_ magnet exports that month were **5,010 tonnes**, down **6%** month on month and 18% year on year. Dropping “to the US” turns a 6% decline into a 21% one and makes a bilateral licensing signal read as a collapse in global supply — a distinction that matters directly to tier two, because the US-bound figure is the one the scarcity thesis rests on and the global figure is the one magnet prices respond to. The underlying article states the scope correctly in its first sentence; the syndicated headline does not. The same piece dates the Trump–Xi meeting to 25 September, which is a day late: Xi’s state visit ran 23–25 September and the leaders met on the **24th**.

Misattributed

“Critical Metals Surges 25% as Greenland Security Pact Puts Tanbreez Rare Earth Permit in Focus; USA Rare Earth Climbs 6%, MP Materials Rises 4%” (21 September 2026)

The session is real, the attribution is loose and the two figures are intraday prints. Trump announced on **18 September** an agreement with Denmark and Greenland giving the United States “permanent control over security” on the island, and the rare-earth complex was bid when New York reopened on Monday. But **USA Rare Earth owns nothing in Greenland**: its assets are Round Top in West Texas, the Stillwater magnet line in Oklahoma, the Blacksburg site in South Carolina, Less Common Metals in Cheshire, a minority stake in Carester in France and Serra Verde in Goiás, Brazil. A security treaty over Greenland changes none of them, so a nine-per-cent move in USAR is sector momentum on a policy headline, not exposure being repriced — and treating it as the latter is the same error this page's whole tier-two section exists to refuse. The figures are wrong too, in the ordinary way: measured on closes, **USAR rose 9.2% to $16.78** and **MP 5.8% to $50.00**, against the “6%” and “4%” in the headline. Those are prints taken while the session was still open, published as the day's move. This page prices the last _completed_ close for exactly this reason.

Context missing

“Over the past four weeks REMX is 11.0% lower while the basket has fallen 5.2%” — this page, through 17 September 2026

The second number cannot be rebuilt, which makes it the same fault as the 0.41 correlation one row up, in a different column. On the figures that run published — Tokyo at 17 September, Xetra and the US at 16 September — an equal-weighted basket of the six tier-one names gives **−8.4%** on a 30-day lookback, **−5.2%** on the 28-day lookback the table's 1m column actually uses, **−4.5%** as a geometric mean of the same six ratios and **−10.6%** as a weekly-rebalanced four-week compound. None of them is 5.2%, and the REMX figure it was set against was computed on a 28-day lookback while the tables alongside it used 30 days. So two numbers presented as a comparison were measured over different windows, and one of them on no stated construction at all. The page now uses a single stated construction for it — the mean of the six names' own 1m cells, equal-weighted and unrebalanced, each on its own venue's four-week (28-day) lookback — which on the current window puts the basket at **−2.7%** against REMX's **−13.6%** on the same basis. The conclusion is unchanged and in fact strengthened; what changes is that both halves are now measured the same way and can be checked against the table.

Wrong

“Nidec has disclosed nothing at all since 4 September” — this page, accurate from 6 September, false from 10, repeated to 25 September 2026

True when written, false when repeated, and the thing it was missing was the most consequential disclosure on this name since the quality report. On **9 September 2026** a shareholder instituted a **derivative action** in the **Kyoto District Court** against two former directors, founder **Shigenobu Nagamori** and **Hiroshi Kobe**, demanding they pay the company **¥28,730,558,300** plus delinquent charges under **Article 462, paragraph 1** of the Companies Act over the treasury-share buybacks of September 2022 and February–March 2023 and the December 2022 midterm dividend. Nidec disclosed it on **10 September**, at which point this page's standing line stopped being accurate. It was nonetheless carried for another fifteen days, and an argument was built on it — that the stock was being repriced purely on interpretation because there was no new information — while the largest single gain anywhere on the page, **+5.6% on 10 September**, landed on the day the suit became public. The conclusion happens to survive, because a ¥28.7bn claim against two individuals is not what moved a ¥2,700 stock up 5.6% and nothing in the tape suggests it was read at all. But it survives by luck rather than by method: from 10 September the premise was false and was checkable on the company's own disclosure page. The lesson is specific to negative claims. “Nothing has been disclosed” is not a background condition that stays true until someone notices otherwise; it is an assertion about the record with a silent expiry date, and restating it on day fifteen is a new claim requiring the same check as on day one. This page audits press coverage for exactly this and had been reprinting its own unverified line while doing so.

Wrong

“The only analyst action on the tape is a week old — Jefferies cutting its target to €8.70 while keeping a Buy” — this page, on the Schaeffler session of 24 September, published 25 September 2026

Wrong on the date, and the date was the whole load-bearing part. Jefferies cut Schaeffler to **€8.70 from €9.45** on **23 September 2026** — _one_ session before the −6.2% move, not a week. The error has a clear mechanism: there were two notes from the same analyst, **Vanessa Jeffriess**, and this page merged them. On **5 August 2026** Jefferies went €10.45 to €9.45; on **23 September 2026** it went €9.45 to €8.70. The target quoted came from the September note, the age from the August one, and the two spliced together produced a catalyst that could be set aside as stale. It was not stale. The September note kept a Buy, said Schaeffler’s own reduction of its 2028 targets had damaged credibility, credited the company with self-help potential, and called the non-automotive robotics business the most attractive investment story in the sector. Against the Xetra line the sequence is unambiguous: **−2.24% to €6.97** on 23 September, the session of the note, which contemporaneous German coverage headlined as the stock falling on the Jefferies cut; **−6.17% to €6.54** on 24 September; **+1.38% to €6.63** on 25 September. So two claims on this page need separating. The conclusion that no disclosure explains the 6.2% session survives — a one-day-old target cut does not ordinarily deliver its damage on the second day, and nothing else was published. But the 23 September fall, which this page attributed to the dollar-led metals liquidation that took the minerals tier down, is far better explained by the note: a German bearing and auto-parts supplier is not a rare-earth proxy, and that attribution was co-timing dressed as cause. The failure mode is worth naming separately from the Nidec row above. That one was a negative claim repeated past its expiry; this one is a dismissal — the relevant item was found, misdated, and discarded for being old. Both come from the same habit of inheriting a characterisation instead of re-reading the source, and on a page whose method note insists every figure be dated, an analyst note is a dated document like any other.

Wrong

“Novonix 3.4% to $2.801” — this page, on the session of 23 September, published 24 September 2026

The price is right, the comparison is not. Novonix closed at **$2.85** on 22 September and **$2.801** on 23 September, so the session was **−1.7%**. The adjacent figure in the same sentence, Lynas easing 2.6% to $10.06, checks out exactly, which is what makes this one worth recording rather than shrugging at — the arithmetic was being done correctly one clause earlier. Two corroborations put the base beyond doubt: this page's own tier-two table published **$2.85** as Novonix's last close on the 23 September run, and the vendor, whose 22 September daily bar then went missing for four days across seven US tickers, has since backfilled that session at exactly **$2.85**. The rest of the claim survives — $2.801 was indeed the lowest close since 9 September's $2.74. What does not survive is any account of where −3.4% came from. It implies a prior close of about **$2.90**, and there is no $2.90 anywhere in the series: 18 September was $2.885, 21 September $2.84, 22 September $2.85. Every other row in this audit that concerns this page's own work names a mechanism — a merged pair of analyst notes, a negative claim past its expiry, a base rolled without noticing. This one has none, and that is the finding. A figure with a traceable bad method can be guarded against by fixing the method; a figure that matches no input at all can only be guarded against by recomputing every percentage from the two closes it sits between, which is now the rule for the prose as well as for the tables.

Overtaken

“A board resolution has been passed to dismiss President Mitsuya Kishida”, and Michio Kaida named as incoming CEO — aggregated coverage, 29 September 2026

Both halves have now happened, and the row stays because one of them was still false when it was published. Nidec’s only statement on the matter at the time, released **28 September 2026**, said the opposite in terms: it was “true that change in executive members and a large-amount impairment are being discussed,” but “none of them has been decided by the Company at this moment.” _Nikkei_’s 29 September report, the source for Kishida going, named no successor at all. Kaida was a real and plausible internal candidate — First Senior Vice President, and vice-chair of the committee running the company’s reform programme — which is exactly what made the claim durable rather than obviously wrong. It was then **confirmed on 29 September**, when an extraordinary board meeting accepted Kishida’s resignation and appointed Kaida president and chief executive. So the aggregators were about a day early rather than inventing it, and that is the useful lesson rather than a vindication: a claim that turns out true was still not established when it was asserted, and nothing in the 29 September record distinguished it from the ones that do not land. One detail circulating with it is wrong on its own terms: Kaida is described in places as Nidec’s _chief technology officer_, and the company’s own release identifies him by the First Senior Vice President role and the reform committee, not that title.

Wrong

“REMX −0.4% to $66.66, a fourth consecutive fall” — this page, on the session of 25 September, published 28 September 2026

The price, the percentage and the “lowest close since 31 July” all check out; the streak does not. REMX fell on **23 September** (−3.11% to $68.76), **24 September** (−2.66% to $66.93) and **25 September** (−0.40% to $66.66) — three sessions, not four. The session before them, **22 September**, closed _up_ **1.0% at $70.97**, and that base is not in doubt from either direction: it is the figure this page reconstructed from previousClose when the 22 September US bar went missing across seven tickers, and the figure the vendor backfilled to the cent when the hole healed. So the error is not a data artefact and has no reconstructable mechanism beyond counting a streak forward from the previous run's copy rather than reading it off the series. It is corrected by the tape rather than by this page: **28 September**'s −3.3% to $64.47 is the fourth consecutive fall, and it is the lowest close since **6 November 2025**. A run-length is a claim about a series, and like the negative-disclosure claim two rows up it has to be re-derived each time it is repeated rather than incremented.

Wrong

“Nidec surged 6.2% to ¥2,389” and “the Nikkei climbed 2.10%” — market coverage of the 30 September 2026 Tokyo session

Two figures, one mechanism. Nidec closed at **¥2,355** on 30 September, **+4.67%** on the ¥2,250 of the day before; the session high was **¥2,398**. The Nikkei closed at **66,753.72**, **+1.94%**, with a high of **66,946.14**. Both quoted numbers sit between the close and the high of the instrument they describe, which is the signature of an intraday print carried into a report as the day’s move. Neither is a rounding difference: ¥2,389 is a level the stock traded at and did not settle at, and 2.10% resolves to about 66,856, likewise inside the range. Checked against the index and the stock themselves rather than against each other, because a wire figure and an aggregator figure repeating it are one source. This is not the first Nikkei percentage on this page to have been taken before the bell, and the check is the same every time: a move is a claim about a settlement, so price it off the close or do not price it.

Wrong

“Nidec stock slumped 17.2% to ¥1,949” and “MP Materials surges 6% to $48.29” — market coverage of the 1 October Tokyo session and the 30 September US session

The same error as the row above it, one run later, on two venues at once, which is why it gets its own row rather than a footnote. Nidec closed at **¥2,100** on 1 October, **−10.8%** on the ¥2,355 of the day before, having traded as low as **¥1,888**. ¥1,949 is neither of those: it sits between the close and the low, and the report that carries it names the ¥1,888 low separately in the same paragraph, so the 17.2% is being offered as the day’s result rather than as an intraday reading. MP closed at **$47.29**, **+4.1%**, with a session high of **$48.665**; $48.29 was never a settlement, and the quoted +6% and +5.8% resolve to roughly $48.16 and $48.06, both inside the range. Two things are worth separating. The _directional_ reporting is sound in both cases, and on Nidec the widely quoted “as much as 20%” is exactly right against the ¥1,888 low, which this page confirms to the cent from an independent wire report of the same figure. What fails is the settlement. A percentage and a price attached to a named session are a claim about where trading _ended_, and on both of these names the stock ended somewhere else: Nidec recovered **11.2%** off its low into the bell and MP gave back a point and a half off its high. That recovery is not a detail. A reader told Nidec fell 17.2% on an audit disclaimer and a reader told it fell 10.8% are being handed different facts about how the market received the disclaimer, and only one of them is the market’s. The check has not changed and is now stated for the third run: price it off the close, or do not price it.

Wrong

“Lynas shares fell 7.1% to A$12.85” — market coverage of the 1 October 2026 Sydney session

The fourth consecutive run on which this page has had to record the same error, now from a fourth outlet and on a third venue. Lynas closed at **A$12.64** on 1 October, **−8.60%** on the A$13.83 of the day before, in a session that ran a high of **A$13.35** and a low of **A$12.60**. **A$12.85** sits inside that range, so it is a level the stock traded through and did not settle at, and the quoted −7.1% resolves to about A$12.85 as well, which makes the two figures one reading rather than two checks. The ADR agrees with the close and not with the report: LYSDY settled **−8.85% at $8.75** the same day, and a ratio line and its primary do not both miss by a point and a half in the same direction by accident. What separates this row from the three before it is the sign of the gap. On Nidec and MP the reported move was _larger_ than the settlement, because the quote was taken before a recovery into the bell; here it is a point and a half _smaller_, because the quote was taken before the last leg down. A reader who takes the published figure therefore understates the market’s reaction to an all-share acquisition that hands 5.9% of the company to the target’s holders, which is precisely the number that deal has to be judged on. The check does not change with the direction of the error: a move is a claim about a settlement, so price it off the close or do not price it.

Wrong

“Nikkei 225: 70,576 (+0.90%)”, and elsewhere 70,650, “+827.47 / +1.18%” and “rose to 70665 points… gaining 1.03%” — coverage of the 6 October 2026 Tokyo session

A fifth consecutive run of the mid-session quote, and the widest spread of versions yet: **four distinct numbers** for one close, none of which is it. The Nikkei settled at **70,683.98**, **+737.12 points** and **+1.05%** on the 69,946.86 of 5 October, in a session that ran from **69,813.16** to **70,798.62**. A daily market note carries 70,576 and +0.90%. A market-data aggregator carries three more on a single screen, and they disagree with each other before they disagree with the index: a headline value of **70,650**, a change of **+827.47 points / +1.18%**, and prose putting the index at **70,665** for **+1.03%**. Every one of 70,576, 70,650 and 70,665 sits inside the day’s range, which is the usual signature of a figure read off a live screen. The point change is the more instructive error: +827.47 on the prior close implies about **70,774**, so it matches neither of that page’s own levels, and a point change that cannot be reconciled with the level printed beside it is not a rounding difference but two unrelated reads stitched together. This one matters more than a single-stock miss because an index figure is what the day’s session gets described by, and the error understates the move that cleared 70,000 for the first time since 1 July 2026. Price it off the close, or do not price it.

Wrong

“A _Nikkei_ report of final-stage talks to sell Nidec Components to Carlyle… a press report, not a disclosure” — this page, carried from 30 September to 5 October 2026

This page’s own, and the most substantial of the eight it has had to record. The caveat was correct for exactly one session and then stopped being so, which is the harder kind to catch: **Nidec’s own release of 1 October 2026** states that the board resolved that day to transfer all **66,302,120 shares** of **Nidec Components Corporation** to **TCG2603 Co., Ltd.**, indirectly owned by an investment fund related to The Carlyle Group, for **¥102,980m**, with a contract day of 1 October and a **transfer date of 1 December 2026**. So from 1 October a verified company disclosure existed, carrying a precise price the press reports rounded to “more than ¥100bn”, while this page went on labelling the transaction unconfirmed through the three runs that followed. The irony is exact and worth keeping: the rule this section applies to everyone else is to verify against the company rather than the wire, and the failure here was checking the wire twice and the company not at all. Two things the correction adds that the press reports did not: the unit earned **¥5,465m** of operating profit on **¥32,092m** of sales in the year to March 2026 against **¥67,319m** of net assets, both down for a third straight year, which prices the deal at about **1.53 times book**; and its end markets include **collaborative robots**, making this the disposal of the most robotics-adjacent sensing business inside a tier-one name on this page. Neither detail was available from the coverage, because neither was in it.

Wrong

“Anchoring a true 104-return window needs a three-year request, and that is how the figures here are built” — this page’s method note, carried from 16 September to 5 October 2026

Also this page’s own, and caught by rebuilding the claim rather than re-reading it. The note correctly identified the trap: a two-year daily request spans 105 ISO weeks including the in-progress one, so excluding that leaves **103** weekly returns, not 104. It then said the page had moved to a three-year request to get a true 104. It had not, or the figures were never recomputed on it. Rebuilt from a three-year request at a genuine **104** returns, the window as it stood when this row was written (ending 2 October 2026) gave basket-to-REMX **0.3670**, Lynas **0.6322**, Schaeffler-to-basket **0.5991** and Harmonic Drive-to-REMX **0.1581**. The figures published against it were **0.3681**, **0.6326**, **0.5997** and **0.1596**, which reproduce at **103** returns and at no other length. So the window has been 103 returns throughout, while four charts, two captions and the note all said 104. The figures themselves were never wrong and no conclusion moves by a thousandth; what was wrong was the description of how they were made, which is the thing this page asks its readers to check it on. The counts have been corrected to 103 everywhere rather than the window changed, because changing it would move every figure on the page to fix a stated number that was the only thing broken. The general lesson is the one this row is the second instance of: an error too small to show up in the output survives every re-read and dies only to a rebuild.

Wrong

“Seagate Tops Toshiba in Bidding for TDK’s HDD Head Unit” — aggregated coverage, 6 October 2026

A sourcing failure rather than an arithmetic one, which makes it the first of its kind in this section. Every other wrong row here is a number read off the wrong instrument or the wrong moment. The underlying _Bloomberg_ report of **6 October** describes an auction _in progress_: Toshiba opened talks in the spring, Seagate bid higher over the summer, several billion dollars, all of it attributed to people it does not name. The aggregated version states the outcome in its headline as settled. Two things were available to contradict it on the day it published. **Toshiba** told the same outlet that the report was **“inconsistent”** with its understanding and that no trilateral discussions between Toshiba, TDK and Seagate over the future supply of hard-disk heads had taken place, with **Japan Industrial Partners**, its owner, saying the same; and neither **TDK** nor **Seagate** has commented at all, then or since. So the record holds one unnamed-source report, one on-the-record denial of the process, and no word from either party whose signature a transaction would need. “Tops” is a verdict on a contest none of the three has acknowledged. The denial deserves the same reading, for the record: it is a spokesperson's comment to the outlet that ran the story, not a release or a filing, so it establishes what Toshiba is willing to say rather than what is true. This page priced the session the report caused and wrote in the same paragraph that an auction reported through unnamed sources is not an agreed transaction; the Tokyo session after the denial took **2.70%** back off TDK. The check this row adds to the ones above it has nothing to do with closes: before repeating a transaction claim, find the parties who would have to confirm it, and record what they actually said.

Half right

“China’s rare-earth export truce expires November 10 with no renewal in sight” — aggregated coverage, 6 October 2026

**This row was published as “Wrong” and that verdict was too strong; it is downgraded on the 20 October run, and the reason is a distinction this page should have drawn the first time.** There are two instruments here and they are not the same. MOFCOM **Announcement No. 70 of 2025**, dated 7 November 2025, suspended Announcements 55 to 58, 61 and 62 until **10 November 2026**, and Beijing has published nothing extending, replacing or amending it: trackers reading the ministry’s own 2026 announcement list on **1 October** and again on **16 October** found it running to **Announcement No. 48** with no such notice, and the US has filed no matching extension of its own measure. The 10 January 2027 date comes from **Scott Bessent** on **23 September 2026** and from a ministry official on **28 September**, both describing the political arrangement around the suspension rather than the suspension. So the headline date is the only expiry in force, and “no renewal in sight” is defensible as to the legal text. What the piece still gets wrong is what it leaves out: the stated extension exists, it is the market-moving fact of the period, the same outlet had reported it, and a countdown written as though nothing had been said is not the state of play. The original framing error stands too, because the piece’s presentation — a cliff edge four weeks out — is exactly the scarcity premium this page recorded _coming out_ of the single names on 23 September, when USA Rare Earth fell 6.8% and MP 4.2% on the extension. Reporting the deadline as a live cliff without the stated extension re-sells a premium the market had already taken out, which is a different fault from getting the date wrong, and the one that survives here. Two things the extension genuinely does not cover, and neither is what the piece claims: the **April 2025** controls putting seven medium and heavy rare earths under case-by-case licensing remain fully in force, and the separate suspension covering gallium, germanium, antimony and superhard materials runs to **27 November 2026** on its own clock. The underlying worry is real. It is attached to the wrong date. **9 November update.** The [joint announcement](<https://www.mofcom.gov.cn/zwgk/zcfb/art/2026/art_91466027f465467985942a639a9485f4.html>) continues the suspension through 10 January 2027.

Wrong

“Each name on its own venue’s 30-day lookback” — this page’s description of its own 1m column, through 7 October 2026

The column is computed on a **28-day** (four-week) lookback, not a 30-day one, and the note has said otherwise since 17 September. This was tested rather than re-read: rebuilt against the ten 1m cells the **7 October** run actually published, the 28-day convention reproduces **all ten** to a tenth of a point and the 30-day convention reproduces **one** — Schaeffler, and only because both lookbacks happen to land on the same bar. Where they differ they differ by more than the figures this page argues over: Harmonic Drive **+13.9%** against +11.1%, Yaskawa **+9.8%** against +6.1%, Nidec **−16.4%** against −23.0%, six and a half points on the last. The irony is specific. The audit row five above this one exists because a basket figure was quoted against a REMX figure measured over a different window, and the fix it announced named the _wrong one_ of the two conventions it was choosing between — so the construction was standardised correctly and then described incorrectly. No published number moves: the cells were right throughout and the four-week base dates this page keeps narrating were always 28 days back. The labels are corrected throughout, and “four-week” and “28-day” now appear together wherever the lookback is named, so the two cannot drift apart again.

Wrong

“These circumstances led Nidec’s stock to be designated on **28 October 2026** as a security on special alert” — Nidec’s own internal-control improvement page, read 9 October 2026

A primary source contradicting itself on the one date that decides a listing, which makes it the first row here taken from a company’s own disclosure page rather than from coverage of one. The prose dates the designation a year ahead: 28 October 2026 is still in the future. The timeline lower down the _same_ page gives **“Oct. 28, 2025 Nidec Corporation’s stock is designated as a security on special alert”**, and every entry around it is consistent with 2025, from the 27 June 2025 reporting extension to the 27 April 2026 revised improvement plan. The timeline is right and the prose is wrong, and it is not the only one: the same introduction opens “On September 03, 2026… established an independent third-party committee” against its own **Sep. 03, 2025** entry. Two dates, both a year out, both contradicted a few hundred words below. This matters more than a typo normally would because of what hangs off it. The Tokyo Stock Exchange requires the written confirmation of internal management system _one year after_ designation, so the prose date would move Nidec’s delisting examination to October 2027 and erase the only listing-risk deadline on this page. A reader checking the deadline against the company’s own explanation of it would get the wrong year. This page dates it from the timeline, **28 October 2025**, which is also what the exchange’s designation notice and contemporaneous reporting carry. The lesson is the one this section least expected to have to write down: a primary source is the right place to check a date, and it is still a document that can be wrong, so the internal consistency of the page you are reading is part of the check.

Wrong

“China Rare Earth Group is in talks to acquire Shenghe Resources, which would put the 3% stake in MP Materials under a Chinese state group” — aggregated coverage, 8 October 2026

The report is real, three weeks old, and was answered on the record by the company it is about; the recirculated version carries the report and not the answer. _Reuters_ reported on **18 September 2026**, on sources it does not name, that state-owned **China Rare Earth Group** was negotiating to buy **Shenghe Resources** and wanted a controlling stake, which would carry Shenghe’s holdings in foreign companies with it, among them roughly **3%** of MP Materials. Shenghe then answered it in a filing to the **Shanghai Stock Exchange**: its controlling shareholder had confirmed in writing that there are _no plans_ to transfer control to external parties, and the company said it was aware of “false” reports in overseas media. China Rare Earth Group, Shenghe and China’s Ministry of Commerce did not comment, and MP’s standing position is that neither Shenghe nor the Chinese government has any control over its operations. So the record is one unnamed-source report, one exchange-filed denial from the subject of it, and silence from everyone else. That is the same shape as the TDK auction row above and it gets the same treatment, with one difference that runs in the opposite direction to that case: Toshiba’s denial was a spokesperson’s comment to the outlet that ran the story, whereas this one is a stock-exchange filing, which is the strongest form the denial can take. A version that omits it is not reporting the story. Offered as a cause of Thursday’s selling it fails on a second count anyway, the one this page keeps having to apply to this tier: it is dated 18 September, and the session was 8 October.

Wrong

“Ambarella (AMBA) shares rise **5.3%** on Qualcomm acquisition speculation” — aggregated coverage of the 9 October 2026 session

The session settled at **+7.22%**, $62.89 to **$67.43**. A 5.3% move resolves to about **$66.22**, a level the stock traded through on its way to a $67.83 high and did not settle at. The figure was true when it was written, as an intraday print, and stopped being true at 16:00 ET; what makes this a row is what happened next. It was syndicated as the session’s move, and the _same publisher_ carried both numbers in two separate pieces about the same day, one headlined 5.3% and one 7.2%, with no indication that the first had been superseded. A reader taking the headline understates by nearly two points the only move in tier one that a takeover rumour has produced, on the one name whose drawdown is set by the AI trade rather than by industrial demand. The rule this breaks is already written down here four times over, for Nidec, MP, Lynas and Novonix: a move is a claim about a settlement, so price it off the close or do not price it. The novelty is only in the mechanism. This is not a stale quote or a misread venue but a _live_ article overtaken by its own subject, which is the failure mode of publishing a percentage before the bell rather than of getting it wrong.

Wrong

“One dated company item existed for the session and it was not a catalyst” — this page, on the Tokyo session of 8 October, published 9 October 2026

There were two, and the one this page missed is much the more consequential. **Oasis Management**, a holder of about **8%** of Nidec, published a letter on **7 October 2026** urging the board to begin a preliminary review of privatization options, with outside directors retaining their own financial advisers and soliciting non-binding proposals from strategic and private-equity buyers in Japan and overseas. _Bloomberg_ reported it the same day and it went out on Business Wire, so it was on the record and findable well before Tokyo opened on 8 October. The entry it belongs in is the one that called Nidec the best relative performer in tier one on the day the gearbox layer de-rated, and read that as a measure of what had already been priced into it. That reading still stands, but it was made without the only piece of news the session actually had: the largest activist on the register asking the board to explore a sale, with the stock trading **3.1%** below its previous close intraday. The mechanism is worth naming because it is not this section’s usual one. This was not a bad source, a misread date or an intraday print. It was a _negative_ claim, “one dated company item existed”, of a kind this page has made casually and repeatedly, and which can only be supported by a search complete enough to prove an absence. An absence is the hardest thing to verify and the cheapest thing to assert. The fix is to stop asserting it: this page now reports what it found and not what exists, and “nothing was filed” is from here on a statement about the filing feeds it checked rather than about the world.

Wrong

“Yaskawa said June–August quarter operating profit of **¥136 billion** rose approximately **5.8%** year-over-year… the full-year operating profit forecast was revised down from **¥600 billion to ¥575 billion**… the stock rose **2.0% to ¥4,879**” — aggregated coverage of Yaskawa’s first-half results, 13 October 2026

Three errors in one article about a tier-one name, and each is a different species, which is why it earns a row rather than a correction. **The scale is wrong by a factor of ten.** Yaskawa’s full-year operating-profit forecast was revised from **¥60,000m to ¥57,500m**, not from ¥600bn to ¥575bn; ¥600bn is the _revenue_ line, which was revised _up_ from ¥580bn. So the sentence takes the revenue figure, attaches it to operating profit, and reports a raise as a cut. **The sign is wrong.** First-half operating profit was **¥22,058m, down 5.5%** year on year, as both the tanshin and the briefing deck state in terms; “rose approximately 5.8%” inverts the one line that measures the business, and the ¥136bn attached to it is impossible on its face, since a single quarter cannot earn more than twice the full year’s guided profit. **And the price is an intraday print.** The article is stamped 13 October at 00:06, inside the Tokyo afternoon session; ¥4,879 is indeed **+2.0%** on the ¥4,783 previous close, and the session settled at **¥4,840**, **+1.19%**. That last one is the fourth time this section has had to record a pre-bell percentage syndicated as a session, after Ambarella, Nidec and the Nikkei, and the arithmetic was right each time; what was wrong was the clock. The reason to spell all three out is that the rating in the same article is real and is carried on the Yaskawa card: _Jefferies_ analyst **Sho Fukuhara** did upgrade the stock to **Hold from Underperform** with a **¥5,300** target. An article can be sound on the fact a reader cannot check and wrong on all three they can, which is the argument for reading the filing and not the write-up. Every Yaskawa figure on this page comes from the 9 October tanshin and briefing deck.

Wrong

“Nothing was filed, which is the point… new lows on doubled volume are the market shortening the odds on a deadline _rather than reacting to news_” — this page, on Nidec’s session of 9 October, published 9 October 2026

The first clause is true and the inference from it is false, and the two are separated by about four days of this page repeating it. Nothing _was_ filed by Nidec on 8 or 9 October. But there was news, it was named, dated and market-moving, and this page reported its absence: **UBS** cut Nidec to **Neutral from Buy** on **8 October 2026**, analysts **Shingo Hirata** and **Kohei Takahashi** taking the price target to **¥2,400** from ¥2,800 and putting their operating-profit forecasts **16%** and **22%** below consensus for this year and next. _Bloomberg_ reported the stock falling as much as **9.3%** on it. Separately **R&I** cut the issuer rating one notch to **A+** and placed it on monitor with negative implications over the impairments. A broker downgrade is not a filing, so the distinction this page drew is a real one and worth keeping; what does not follow from it is “rather than reacting to news”. A sell-side capitulation on a name whose accounts carry a disclaimer of opinion is precisely news, and a 6.07% session with a 9.3% intraday low is what it looks like being priced. The aggravating detail is the timing. On **9 October** this page published, in the row directly above, a commitment that “nothing was filed” would from then on be “a statement about the filing feeds it checked rather than about the world” — and then, in the same run, used the absence of a filing to rule out the existence of a cause. The commitment was correct and was broken in the same breath, which suggests the fix is not a resolution but a habit: when a session moves on no disclosure, check the brokers and the rating agencies before concluding there was nothing to react to.

Wrong

“The drawdown column is the one that does not move with the calendar, which is part of why this finding is framed on it” — this page’s Finding 02, carried for weeks and repeated on 13 October 2026

This page’s own, and the one it had most reason to catch, because it was the stated justification for framing a finding on that column. The 52-week high used here is meta.fiftyTwoWeekHigh, and that field’s window is anchored on the **request date**, not on the venue’s last completed close. On 14 October 2026 the window rolled past the **14 October 2025** session highs onto the **15 October 2025** ones, and three cells moved on the calendar alone: **USA Rare Earth**’s high fell from **$42.00** to **$35.92**, taking its drawdown cell _in_ 3.3 points to −68.6% on a session its price fell 4.57%; **Novonix**’s from $31.90 to **$27.90**; **MP**’s from $100.25 to **$97.00**, for a cell four tenths of a point better on a 2.15% fall. The weaker claim survives and is now the one made: the drawdown column rolls less often than the return columns, and when it rolls it is a high coming off the window rather than a base coming on. The stronger claim, that it does not move with the calendar at all, was false every time it was published.

Wrong

“At the close in Tokyo, the Nikkei 225 added **1.66%**” — wire coverage of the Tokyo session, published 16 October 2026

The **ninth** Nikkei index figure this page has had to correct for being a print taken before the bell, from the same outlet as the eighth and on the very next session. It is also the first for which the mechanism can be named to the minute, and the first that indicts this page as well as the filer. The Nikkei 225 **closed at 65,324.29, up 898.34 points or 1.39%**. The 1.66% is the **15:00** print of **65,493.63**, which is **+1.657%** on the previous close: the quoted figure to two decimal places, not a rounding difference. The Tokyo Stock Exchange extended its afternoon session by thirty minutes on **5 November 2024**, so continuous trading now ends at **15:25** and an itayose closing auction settles the market at **15:30**. A “close” read at 15:00 is therefore taken before continuous trading has even finished, let alone settled. That is the part this page got wrong too: the row eight above this one dates a Tokyo session against “the 15:00 close”, which has not been the close since November 2024, and it has been corrected to 15:30. Eight of these rows have been written about other people reading the index early while this page carried the wrong closing time in its own copy. The five-minute series shows the whole give-back: 15:00 **65,493.63**, 15:10 65,401.91, 15:15 65,287.84, 15:30 **65,324.29**, the last of these matching regularMarketPrice stamped 15:45 JST and the next session’s previousClose. A second aggregator carried **65,605 and +1.83%**, a level the index held around 14:35–14:50. Both are afternoon quotes published under a headline describing the day, and both overstate it. The two Tokyo-listed Nikkei 225 ETFs corroborate the direction as before, 1321.T **+1.19%** and 1330.T **+1.13%**. Six of the names here price in Tokyo, four of them in tier one, so the index a reader is handed decides whether a 3.68% session in Harmonic Drive counts as leading the market or merely following it. On 16 October it led, and by how much depends entirely on which figure you take: two and a half times a 1.39% index, or two and a quarter times a 1.66% one. Price it off the close, or do not price it.

Wrong

“USA Rare Earth Climbs **4%** as Month-Long Selloff Pauses; MP Materials Ticks Up” — with USA Rare Earth at **$10.77**, MP **up 2% at $40.82** and REMX **down 0.9% at $55.26**, published 15 October 2026

All four figures are mid-session prints filed as the day, and the piece is stamped **1:53pm ET**, more than two hours before the bell. At the close **USA Rare Earth was up 7.46% at $11.09**, nearly double the headline; **MP up 4.16% at $41.61**, twice its stated move; and **REMX down 0.07% at $55.74**, flat rather than down nine tenths of a point. The direction is right in all three and the magnitude wrong in all three, understating a session that was still building: the tier was mid-reversal, having traded as low as **$9.82** on USA Rare Earth and **$38.60** on MP that morning. This is the second mid-session piece from this outlet the section has recorded, after the 10 September snapshot. The aggravating detail is internal: the article’s own live sidebar carried the correct closes, **$41.61**, **$11.09** and **$55.74**, directly beneath a body contradicting all three. The framing is the second fault. “Month-long selloff pauses” attributes the session to the sector, and the session was not the sector’s: the **ten-year yield fell about 10bp to 5.09%**, the dollar index **0.54%** to 102.00, gold rose **1.79%**, the gold miners **4.34%**, the junior miners **4.64%** and silver **3.39%**. A rates-driven bid across the whole metals complex is not a rare-earth event, whatever the rare-earth names did inside it, and the names that did run past the complex were the small ones rather than the large.

Context missing

“Nidec fell 1.04% to ¥1,800 … not a new seller arriving, the bid going away” — this page’s account of the 16 October Tokyo session

The tape facts are exact and the inference was drawn straight over a filing. Nidec published “Nidec Receives an Investigation Report from the Executive Responsibility Investigation Committee” on **16 October**, carried on Business Wire at **12:23 AM EDT**, which is **13:23 JST**: inside the afternoon session, two hours and seven minutes before the 15:30 auction. The Committee, established **13 March 2026**, exists to determine whether current and former directors, auditors and vice presidents “bear legal liability, including whether they breached their duties” over the improper accounting, and the release states that Nidec will “deliberate legal actions such as lodging a claim for damages”. On a name whose central question is whether the listing survives October, that is the most consequential document available, and it landed on a session this page characterised as carrying no disclosure. Two things keep this short of a plain _Wrong_. The numbers check out to the share: **11,851,000** against a twenty-session average of **16,355,065**, or **0.72 times**. And the market genuinely did not react to it, which the five-minute series settles: ¥1,818 at 12:30, ¥1,806 by 13:15 on that afternoon’s heaviest bar of 160,100 shares, ¥1,805 at 13:20 and still ¥1,805 at 13:25, the bar containing the release, on 46,500 shares, then ¥1,800 by 13:45. No gap, no volume spike, no acceleration; the drift was running before the release and continued at the same rate after it. So the _conclusion_ about the character of the session stands and the _method_ that reached it does not, for the third time in nine days. This audit already carries the Oasis privatization letter, a dated item sitting on the public record and missed because nothing was checked, and the 8 October UBS downgrade, a session declared newsless because no filing existed. This instance is the sharpest of the three, because here there was a filing, in English, on the company’s own feed, on the day, and it was still the absence of a check rather than the absence of a document. The standing fix is narrow and mechanical: before any session on this page is given a character, the disclosure feed of the company it belongs to is read for that date.

Wrong

“Nidec’s board resolved on October 16 to pursue litigation against nine former executives” — an aggregator, 19 October 2026

Wrong on both halves, and the primary document is a page long. Nidec’s release of **16 October**, “The Investigation Report Issued by the Executive Responsibility Investigation Committee”, says the company “will _consider_ necessary legal actions such as filing a claim for damages against those directors, etc., and respond swiftly”. Considering is not resolving, and no resolution to litigate sits on Nidec’s disclosure feed. Nor are the nine all former executives: the Committee lists **Kazuya Murakami** and **Hiroyuki Ochiai** without that qualifier, and the ninth is a **former vice president**, a former CFO of the company from May 2022 to March 2026, who was not a director at all and whose liability is framed under the Civil Code and Nidec’s Vice President Regulations rather than the Companies Act. The count of nine is the one part that holds.

Wrong

“Nidec fell 2.8% to 1,760.0 yen on Monday” — a wire, 19 October 2026

An intraday print published as a close. Nidec closed **¥1,789** at the 15:30 JST auction on 19 October, down **0.61%** from Friday’s ¥1,800. The session ranged ¥1,749 to ¥1,803, so ¥1,760 was touched on the way through, and −2.8% from ¥1,800 is ¥1,749.6, which is the day’s low rather than its close. The error matters more than the 29 yen: a reader told Nidec fell 2.8% on the session its own committee named nine people as liable in damages would conclude the market traded the news, when the close moved six tenths of a per cent on **0.32 times** average volume. Same class as the “Nikkei −2.64%” figure already in this audit, and the same fix: price the venue’s completed close, not a quote taken while it was open.

Wrong

“Neither the redacted Report nor the promised actions has posted” — this page, 17 October 2026

Wrong when it was published. Nidec issued _two_ releases on **16 October**: the receipt, which this page carried, and “The Investigation Report Issued by the Executive Responsibility Investigation Committee”, which it did not. The second sits one item above the first on the same feed and carries the Committee’s conclusion in full, naming **Nagamori, Kobe, Sato, Yoshimoto, Seki and Kishida** under Article 423(1) of the Companies Act, **Murakami and Ochiai** under the same article, and an unnamed former CFO under the Civil Code. The redacted Report proper was indeed still pending, and is now scheduled for **20 October** “or so”, so that half of the sentence stood; the conclusions and the names, which the run described as absent, had been public for a day. That is the fourth instance of one mechanism this audit now records, after the Oasis privatization letter, the Schaeffler order book and the receipt released on this same 16 October, and the cause is identical each time: the feed was checked for one expected document instead of being listed.

Wrong

“104 weekly returns to 16 Oct 2026” — this page’s correlation chart subtitle

Off by one, contradicted elsewhere on the page, and a repeat. The aligned sample is **104 weekly closes**, which yields **103 weekly returns**, and the prose of this same section says so twice, describing “a 103-week window”. The count was raised to 104 for the 16 October window on the reasoning that a Saturday run picks up a week whose trading has finished; the flaw is that a two-year request rolls forward at both ends, so the week gained at the end is paid for by one dropped at the start, and the total does not move. The ceiling is Schaeffler’s Xetra line, the shortest of the fifteen series, so it cannot be lifted by asking for more history either. Corrected in the subtitle on the 19 October run. No correlation figure changes, because each was computed from the returns themselves rather than from the count.

Wrong

“Nikkei 225 down 1.16%” on 20 October 2026, and an historical table showing 64,224.00 for the same session — wire and aggregator coverage

Neither is the close. The index settled at **64,075.42** against Monday’s **64,916.70**, which is **−1.30%**, confirmed on two routes: the 20 October daily bar and regularMarketPrice stamped **15:45 JST**, fifteen minutes after the 15:30 auction. The published 1.16% implies 64,163 and the tabled 64,224.00 implies −1.07%; both sit inside the day’s 63,901 to 64,810 range, which is what an intraday snapshot looks like when it is printed as a close. Two other snapshots circulating on the same session, “around 64,300” at one point and “63,980, −1.4%” at another, bracket the real figure from both sides and neither is it either. This is the third instance of one error class in this audit, after “Nikkei −2.64%” and Nidec at “¥1,760”: an index or a stock quoted mid-session and reported as settled. The check is the same every time, and it costs one request: convert the quote timestamp to exchange-local time and reject anything stamped before the closing auction.

Wrong

Lynas at 0.64 to REMX, Nidec at 0.61 and Harmonic Drive at 0.81 to the basket — this page’s own correlation charts, 17 and 19 October

All three round a point low. On the window those runs used, **104 aligned weekly closes and 103 returns from week 2024-43 to week 2026-42** — the window has since rolled forward one week, to 2024-44 through 2026-43 — the figures are **0.6471**, **0.6229** and **0.8152**, which round to **0.65**, **0.62** and **0.82**. Both bucketing methods agree, and the headline basket-to-REMX figure they sit around, **0.3885**, was right. What makes this a clean finding rather than a judgement is that the window did not move between those runs and this one: on an identical sample a correlation cannot change, so a cell that recomputes differently was carried wrong rather than overtaken by new data. The likely mechanism is a cell edited by hand from a prior run’s rounded figure instead of being rewritten from the recomputation, which is the same failure the “104 returns” row above records in a different column. Corrected in both bar charts and in the tier-two table this run. No conclusion in section 04 depends on any of the three: on the window used for that October note the robotics names spanned 0.19 to 0.47 against REMX and the minerals names 0.37 to 0.65.

Wrong

“Nikkei 225 up 0.97%” at the close on 21 October 2026, with an index table reading 64,696.00 — wire coverage

Neither figure is the close, and this is the third time this error class has earned a row. The index settled at **64,621.24** against the previous day’s **64,075.42**, which is **+545.82 points, or +0.85%**, confirmed on three routes: the 21 October daily bar, regularMarketPrice stamped **15:45 JST**, and Nikkei Inc.’s own index page. The wire’s level and its point change agree with each other and with the correct previous close, so this is a different print rather than a transcription slip — one that sits inside the session’s 64,558.60–65,404.63 range and is not the 15:30 auction. It joins the “−2.64%” row of 19 August and the “down 1.16%” row of 20 October. Check the index against the index.

Wrong

“Tokyo’s Nikkei 225 lost **1.6%** to **63,595.58**”, with SoftBank Group “lost 5.6%”, Tokyo Electron “fell 5.9%” and Advantest “lost 5%” — Associated Press market report, published 22 October 2026

The **tenth** Nikkei index figure this page has had to correct for being a print taken before the bell, the first from a wire of this reach, and the first in which _four_ quoted lines are all wrong in the same direction by almost exactly the same amount. The Nikkei 225 **closed at 62,907.84, down 1,713.40 points or 2.65%**. SoftBank Group closed **−6.86%**, Tokyo Electron **−6.99%** and Advantest **−6.01%**. Each published figure understates the fall, by **1.05, 1.26, 1.09 and 1.01** points respectively, and four errors of one size are one snapshot rather than four mistakes. The piece is stamped **12:43 AM ET**, which is **13:43 JST**, one hour and forty-seven minutes before the 15:30 itayose auction. What makes this instance unusually clean is that the session only went one way after it was filed: the index and all three named stocks **closed at their lows of the day**, so there is no moment after 13:43 at which any of the four figures was ever right. The quoted 63,595.58 implies −1.587% and sits inside the day’s **62,907.84 to 64,866.44** range, which is what an intraday level looks like when it is printed as a close — and that range is 1,958 points wide, because the index opened up 0.38% before giving the whole session back. Two independent routes carry the close: the 22 October daily bar and regularMarketPrice stamped **15:45 JST**. A third, written the same session, agrees without being asked: _Il Sole 24 Ore_ put Tokyo at **2.65%** lower. This row joins the “−2.64%” of 19 August, “down 1.16%” of 20 October and “up 0.97%” of 21 October — three consecutive sessions now, from three different filers. Four of this page’s six tier-one names price in Tokyo, so the index a reader is handed decides whether Harmonic Drive’s 9.55% fall is three and a half times the market or merely six times a mild one. Check the index against the index.

Wrong

“Japan shares lower at close of trade; Nikkei 225 **down 0.25%**”, index line **62,752.00**, **−155.84** — market wrap dated 23 October 2026, and the same outlet’s **−2.40%** for 22 October

The **eleventh and twelfth** Nikkei index figures this page has had to correct for being prints taken before the bell, and the first time one vendor has done it on two consecutive sessions. The index **closed 62,693.16 on 23 October, down 214.68 points or 0.34%**, confirmed on two routes: the 23 October daily bar and regularMarketPrice stamped **15:45 JST**. The published 62,752.00 and its −155.84 agree with each other and with the correct previous close of 62,907.84, so this is a different print rather than a transcription slip, and it sits inside the day’s range of **61,462.61 to 62,786.89** without being the 15:30 auction. The day before, the same outlet headlined Thursday **−2.40%** where the index closed **−2.65%**, 62,907.84 against 64,621.24. Both errors understate the fall. What makes the pair instructive is that the _same_ 23 October report has **TDK down 5.61%, −156.00, at ¥2,624.50**, which is correct to the yen and matches this page’s own figure exactly: the single-stock table is right and the index line is not, in one document. Check the index against the index, even when the stocks around it check out.

Wrong

“On October 23, 2026, General Motors Co (NYSE: GM) **announced the termination** of its long-standing supply agreement” with “a U.S. manufacturer of rare-earth magnets” — aggregator piece dated 23 October 2026

Three errors in one sentence, on a story this page carries in full two sections below, and the version that gets it right was published the same day by someone else. **The date is wrong.** The termination notice was delivered on **1 July 2026** and took effect then, nearly four months before this piece ran. **The announcement did not happen.** The account became public when _Reuters_ reported it on **22 October** from SEC filings and people familiar with the matter; GM has published nothing, and both sides decline to comment because the arbitration is confidential. An aggregator turning a reported-from-filings story into a same-day company announcement is the single most common failure in this section, and it is the one that makes a stale fact look like a catalyst. **And the counterparty is missing.** It is **e-VAC Magnetics**, the US unit of Vacuumschmelze, whose plant is the $500m South Carolina facility; without the name there is nothing in the sentence a reader can check. The facts were plainly available: a trade outlet published the same story the same day with the SEC provenance, the 1 July effective date and e-VAC all intact. Nothing here changes the substance, which matters on this page because the cancelled offtake is real and is recorded in tier three. What it changes is the date a reader would attach to it, and a four-month-old termination read as Friday’s news is exactly the kind of thing that gets priced twice.

Wrong

“The standing tally is **thirty holes, thirty reconstructions, twenty-nine confirmed to the cent**, with 20 October the only one open” — this page’s method note, 22 Oct 2026

One short, and contradicted a few lines above itself. The same run’s venue note reported Schaeffler’s **21 October** daily bar null as well, and gave the close it had to be recovered at, **€5.82**. That gap was the **thirty-first** and was never added to the count, so the tally understated the number of times the German feed has dropped a session and described a closed case as the only open one. This is the same failure mode the audit already carries against this page twice over: a figure maintained in one paragraph and used in another without the two being read together. Corrected, 20 October was the thirtieth and 21 October the thirty-first; both have since backfilled at exactly the figures published here, and **22 October** is the thirty-second. Nothing priced on the page ever depended on the count, which is precisely why it drifted.

Half right

“**Nidec Files Suit for Damages against Former Executives**” — headline of the company’s own release, 27 October 2026

The event is real and the headline is ahead of it. The body of the same release says Nidec “has **decided today to file** a suit for damages” against nine named former officers, and then that it “plans to examine, among others, the **amounts** of compensation to be sought” and will “file the Lawsuit **as soon as the Company is ready to do so**”. So no suit has been filed, no court is named and no amount is set. The nine are right: six former directors (Nagamori, Kobe, Sato, Yoshimoto, Seki, Kishida) and two former corporate auditors (Murakami, Ochiai) under Article 423(1) of the Companies Act, plus a former Vice President who was CFO from May 2022 to March 2026 under the Civil Code or Nidec’s own Vice President Regulations. This is the same distinction this page keeps on Perpetua’s EXIM facility, where _approved_ is not _drawn_: here _decided_ is not _filed_. The trap that follows is a figure. The nearest circulating number, **¥28.7bn** or **$182m**, is a **shareholder derivative** action filed in the Kyoto District Court on **17 July 2026** against Nagamori and Kobe only, and the Investigation Committee’s **¥27,633,480,861** is its estimate of what the scandal cost including investigation expenses. Neither is Nidec’s claim, because Nidec has not made one. Second row on this page sourced to a company’s own disclosure.

Wrong

“USA Rare Earth **Just Broke Ground** on a $1.2 Billion Magnet Complex” — headline dated **27 October 2026**, published 18:54 EST

Seven weeks late, on the evening of the largest session this page has recorded in the name. The groundbreaking at Blacksburg, South Carolina happened on **9 September 2026**, per USA Rare Earth’s own release: about **$1.2bn** of investment, roughly **490** jobs, an 800,000 sq ft plant on a 124-acre site in Bailey Industrial Park, Cherokee County, targeting **6,400 t** a year of sintered NdFeB and **5,000 t** of strip-cast metal and alloy, commissioning **2028**. Every one of those figures is right. Only “just” is wrong, and the article’s own body says “last month”, so the error is in the headline rather than the text. It matters because of when it landed: USA Rare Earth closed **+16.72%** that session, and a headline published hours afterwards announcing a groundbreaking as new supplies a cause that cannot be one. No name in tier two filed or released anything dated 26 or 27 October.

Wrong

“**Fanuc** and **Yaskawa** are teaming up with other Japanese firms to develop AI for robots using **Nvidia’s** tech” — daily market brief dated **27 October 2026**

Real, present tense, and three months old with no date attached. Fanuc, Yaskawa, Fujitsu and Kawasaki Heavy joined NVIDIA’s **Cosmos Coalition** on **15–16 July 2026**, during Jensen Huang’s week in Tokyo, with Fujitsu leading a collaborative control platform and NVIDIA showing a 4-billion-parameter on-device model. No binding commitment and no money were disclosed then, and none has been since. Presented undated in a 27 October brief it reads as a fresh catalyst for a name on this page: Yaskawa rose **2.35%** the following session, on **0.64 times** average volume, which is the lightest turnover anywhere on the page and not what a market reacting to news looks like. The failure mode is the same as the row above it, from a different kind of publisher: a dateless sentence about a dated event.

Half right

“Lynas Rare Earths posts **record** $310.1M quarterly revenue” — and, from a second outlet, “Lynas Rare Earths posts **record sales in Q1 2026**”, both covering the quarterly report of **29 October 2026**.

The figure is right and the word is not, and the company says so itself in the same sentence the figure comes from. Lynas’ own release reads: quarterly sales revenue of A$310.1m “was 55% higher than in the prior corresponding period (Q1 FY26) and is the **highest quarterly sales revenue achieved since Q4 FY22**”. Highest in fourteen quarters is not a record, and the release is careful about the difference, because it does use the word once: “the average selling price across all rare earth products during the period was **a record A$102.1/kg**”. So the record in this quarter is the _price_, and the coverage has moved it onto the _revenue_. The second headline adds a period error on top: the quarter reported is **Q1 FY27**, and **Q1 FY26** is the comparison it beat by 55%, at **A$200.2m** — so the label attaches the claim to the quarter that was a third lower. Both are checkable against one paragraph of the primary document, which is why they are here, and neither needs an outside source to settle it: the company wrote the correction into the sentence being quoted.

Overstated

“Nikkei 225 **plunges 3%**”, and in the body “the benchmark was **down 2.8%**” — coverage of the 30 October 2026 Tokyo session

The **ninth** Nikkei figure this page has had to correct for describing a session by a level the index held before the bell, and the first where the filer never actually claims a close. The Nikkei 225 **closed at 63,290.11, down 1,369.36 points or 2.12%**, verified two ways: the daily bar whose quote metadata is stamped **15:45:03 JST**, and the five-minute series whose final print at 15:30 is the same figure against a previousClose of 64,659.47. Both Tokyo-listed Nikkei 225 ETFs settle it as they have every other time, 1321.T at **−2.17%** and 1330.T at **−2.20%**, which no 3% index day produces. The quoted levels are locatable: −2.8% is around **62,850**, which the index held mid-morning, and the intraday low was **62,762.46 at 10:20**, or −2.93%, so “plunges 3%” describes the worst tick of the day and “down 2.8%” a mid-morning quote, after which the closing auction added about **208** points to the 15:25 print. **This one is treated more gently than the previous eight, and the reason is on the page.** The article never says “at the close”, and its timestamp of 30 October 00:12 sits before the session it describes, so the figure reads as a live number that was never updated rather than a settlement misreported. Overstated, then, not wrong: direction and rough shape are right and the read does not change. What does change the read is the breadth the headline hides. The two TOPIX trackers fell **1.16%** and **1.12%**, barely half the Nikkei’s move, so this was a large-cap technology event rather than a market-wide one. That distinction decides whether a name is leading a rout or merely sitting in one, and six of this page’s names price in Tokyo.

Context missing

“The average selling price rose 88 per cent to a record **$102.10 a kilogram**” — wire and aggregator coverage of Lynas’ September-quarter report, 29–30 October 2026

The figure is Lynas’ own and it is right; the bare dollar sign is not, and on this page it collides with a number it has nothing to do with. The company reports **A$102.1/kg**, an _all-products_ average selling price across its whole basket, up 88% year on year, lifted by floor-price agreements that now cover more than half of production. At an A$/US$ rate of about **0.70** that is roughly **US$72/kg**. The **US$110/kg** figure this page tracks is a floor on **NdPr specifically**, under the US arrangement and the Japanese JARE deal. A reader carrying “$102.10” across to that floor therefore makes two errors at once, a currency and a product basket, and lands on a conclusion that sounds precise and researched: a record realised price still sitting comfortably below the policy floor. The direction of the error is the awkward one, understating Lynas’ realised pricing by about thirty per cent while appearing to compare like with like. Quote the currency with the number, and never measure a basket average against a single-oxide floor.

Wrong

“Schaeffler … 1m **−10.6%**” — this page, tier-one table, published 29 October 2026

Wrong by 2.4 points, and the cause is a feed fault this page documents at length elsewhere reaching into a column nobody was checking. The 1m cell takes the last close on or before the same calendar day one month earlier. For the 28 October close of **€5.93** that base is **28 September**, when Schaeffler closed **€6.46**, which gives **−8.2%**. The published −10.6% implies a base of **€6.63**: the **25 September** close, three sessions early. The mechanism is the German daily hole. The 28 September bar was evidently null when the series was read, and the lookup walked back to the last surviving close instead of stopping, so the window silently stretched to one month and three days. The bar has since backfilled at €6.46, which is the only reason the error became visible at all. Two things earn it a row rather than a shrug. Every other name on both tiers reconciles to the stated definition exactly, so this was a single-name failure that gave no sign of itself. And this page had already published €6.46 for 28 September in its own prose, so the correct base was sitting a few paragraphs from the wrong figure. A documented data fault is only actually handled once the handling covers every column it touches.

Wrong

“REMX is **27.3%** lower on the four-week lookback the tables use, against the basket **18.1%** lower … a **9.2-point** gap” — this page, section 04, published 29 October 2026

Both inputs wrong, the subtraction right, and the first number was sitting two rows away in this page’s own table. On the figures that run published, REMX’s four-week cell was **−15.7%** ($54.35 on 28 October against $64.47 on 28 September) and the basket’s was **−10.1%**, the mean of the six tier-one 1m cells recomputed on the stated definition, or −10.5% taken exactly as displayed. The gap was **5.6 points**, not 9.2. The **27.3%** was USA Rare Earth’s 1m cell, printed two rows below in the tier-two table, and **18.1%** cannot be rebuilt from any construction of the six tier-one cells this page offers, rounded or unrounded, arithmetic or geometric. This is the **second** failure of this one cell: the audit already carries a basket four-week figure of 5.2% from September that could not be rebuilt either, and the row near it records a correlation quoted the wrong way round. The pattern is specific enough to name. The prices, percentages and correlations are recomputed from source every run and have held up under exactly this kind of checking; the _derived_ cross-tier figures in the surrounding prose are typed rather than computed, and that is where the errors cluster. Corrected, the series reads **5.6** points for 29 October and **5.2** for 30 October, and the earlier entries in it have not been re-derived.

Wrong

“Ambarella (AMBA) in Exclusive Talks with NXP Semiconductors, **Shares Drop 5.5%**” — aggregator coverage of the 30 October 2026 session

Wrong by a factor of two and a half, from the same outlet this page audited a week ago over the GM magnet story. Ambarella closed **30 October at $63.12, down 2.26%** from a 29 October close of **$64.58**, verified two ways: the daily bar, and meta.regularMarketPrice stamped **16:00:01 ET**, which agree to the cent. A 5.5% fall from $64.58 would have closed the name at **$61.03**, a level it has not closed at or below since **21 April 2026**, so this is not a rounding question but a different six months. Two other outlets carrying the same report got the tape roughly right, one at “down about 2%” and one at a range of “2% to 3.6%”, so it is a single filer’s error rather than a bad print in circulation, and it sits in the headline, which is the half of an article that travels. What earns it a row is the thing it is attached to. The exclusivity claim has no filing behind it on either side, which leaves the share move as the only element of the piece that looks like corroboration, and an inflated one invites the wrong inference: 5.5% reads as a market pricing a disappointing outcome, while 2.26% on 2.30 times average volume reads as a market that has heard this rumour three times before. The second is what the tape shows.

Wrong

“USA Rare Earth Tumbles **10%**, MP Materials Falls **6%**” — with USA Rare Earth “at **$9.52**” and MP “at **$39.21**”, published 30 October 2026

The third instance of the same fault from the same outlet, and it is a mid-session snapshot filed as the day’s move: the piece is stamped **2:08pm ET**, nearly two hours before the bell, and gives its prices in the present tense. At the close **USA Rare Earth was down 8.56% at $9.62** and **MP down 4.97% at $39.57**, so both headline percentages overstate the session and neither quoted price is a close, an open or a session low. USA Rare Earth did trade through $9.52 on the way to a **$9.19** low and recovered 4.7% off it into the auction, which is precisely what a snapshot cannot see. The year-to-date figures in the same piece, “down 20%” and “down 22%”, are close to the closing values of −19.2% and −21.7%, which is the tell: a correct slow-moving figure sitting beside a wrong fast-moving one. The session was real and it was the largest on this page in a month. Published two hours early, it is still not reportable as a day.

Wrong

“The Dow Jones Industrial Average **closed at 52,342.56** on Friday, October 30, 2026, down just **0.25%**” and “the Nasdaq Composite **finished at 26,184.72, down 1.38%**” — published 30 October 2026

Both figures are the **29 October** session, level and percentage together, presented as Friday’s closes. The Dow closed 29 October at exactly **52,342.56, down 0.25%**, and the Composite at exactly **26,184.72, down 1.38%**; Friday 30 October closed the **Dow at 51,770.31, down 1.09%** and the **Composite at 25,437.00, down 2.86%**. So the piece understates the session by 0.8 and 1.5 points on the one day of the month whose entire interest is its size, which also makes it the article least able to explain the tier-two rout it sits alongside. This is a new species for the audit: not a stale article but a _mixed-date_ one. Its Intel figure, “fell 28.8% during Friday’s session”, is genuinely Friday’s (−28.58%), while its Amazon figure, “fell 4.4% during Friday’s session”, is Thursday’s (−4.25%, against Friday’s −1.41%), and its Alphabet figure is correctly dated to 29 October. One true figure among four makes the rest read as checked, which is why this is harder to catch than a wholly stale piece and worse when it is missed.

04 The REMX Question (Rare Earth ETF)

Every humanoid needs roughly 1.3 kg of NdPr magnet material, and China makes about 90% of the world's rare-earth magnets. That is a real physical dependency. It does not follow that a rare-earth ETF is a way to own robotics — so I tested it rather than assuming it.

### Correlation with REMX

Weekly return correlation, 103 weekly returns from 104 aligned weekly closes, ISO weeks 2024-47 through 2026-46, ending 13 Nov 2026. Daily closes bucketed in each exchange’s local time and cross-checked against exchange-local native weekly bars put the basket-to-REMX figure at **0.3576**. Friday was the last trading session of week 46 in every tracked venue.

Robotics component names Rare earth & critical minerals names

**Lynas, MP Materials and Perpetua have the highest REMX correlations on the watchlist.** Their coefficients are 0.63, 0.55 and 0.55. The six robotics suppliers range from 0.16 to 0.46 against REMX. The weekly-rebalanced robotics basket correlates to REMX at 0.36.

Those six suppliers correlate to their basket at 0.54–0.83. On the table’s 30-calendar-day lookback, the mean of the six suppliers’ returns is **−18.2%** and REMX is **−18.0%**, a gap of **0.2 percentage points**. This comparison averages the individual listing returns using the venue dates in the method note. The weekly correlation calculation rebalances each week.

**Week 41, in the October window. The reason the figure barely moved is not the one the word “cancels” implied, and the distinction is the whole lesson of this run.** Week 41 did not cancel inside the basket. It was a broadly _down_ week in both tiers: the basket fell **3.27%** and REMX **3.54%**, with five of the six tier-one names lower (Schaeffler 5.62%, Harmonic Drive 5.33%, Nidec 5.05%, Ambarella 1.78%, Nabtesco 1.36%, Yaskawa 0.48%) and four of the five minerals names lower (USA Rare Earth 7.58%, Novonix 5.02%, Lynas 2.78%, MP 1.77%, against Perpetua _up_ 1.28%). A week in which both sides fall by almost exactly the same amount is not a quiet observation, it is an observation sitting almost on top of the existing relationship, so it confirms the correlation rather than changing it. That is a different fact from the one this note predicted, and the predicted number was right for the wrong reason. Had the two tiers fallen by materially different amounts the same busy week would have moved the figure hard in one direction or the other.

**October window comparison for Schaeffler.** Its correlation to the basket rises from 0.5997 to 0.6070 and to REMX from 0.4564 to 0.4645, the largest single change on the page when the window last advanced, and the only one big enough to round differently, from 0.60 to 0.61. It earned that by being the worst line in the basket in the week the basket fell, which is what a high-beta member does to its own correlation. Three other cells round differently on much smaller moves: Nabtesco to REMX from 0.22 to 0.21, Yaskawa to the basket from 0.82 to 0.81, and MP to the basket from 0.19 to 0.20; in the magnet tier TDK goes from 0.52 to 0.51 and Aichi Steel from 0.17 to 0.18. Everything else is unchanged at two decimal places: Lynas 0.6333, MP 0.5524, Perpetua 0.5248, USA Rare Earth 0.3678, Novonix 0.3603. The robotics names now run 0.18 to 0.46 against REMX, and 0.55 to 0.82 against the basket. On the 16 October window five cells round differently again, all by a single point and none of them changing a reading: Lynas to 0.64, MP to 0.56, Perpetua to 0.53, Novonix to 0.37 and USA Rare Earth to 0.38, with Nabtesco back out to 0.24 and Nidec to 0.35 against REMX. **No conclusion in this section moves**, which is the honest outcome of advancing a 103-week window by one week, even a week with this much in it. Separately, the four-week columns did move this run, and for once without any help from the calendar: the gap widened from 10.9 to **13.0 points** on price alone, with all seven bases still sitting on 11 September. See the decomposition above.

**Part of the reason is what REMX actually holds.** Roughly 24% of the fund is in rare-earth companies and about 38% in lithium names — more lithium by weight than rare earths, as of April 2026. Top holdings run to MP Materials, Lithium Americas, Lynas, China Northern Rare Earth and Albemarle.

So a bet on REMX is substantially a bet on the lithium cycle and on Chinese miners, with roughly a third of the fund in Chinese companies. It is weighted to miners rather than to the separators and magnet makers a robot actually consumes.

**What REMX is really tracking.** Its drivers are Chinese export policy, MIIT production quotas and the NdPr price — not humanoid order books. Chinese Pr–Nd oxide was assessed at **$97.40/kg on 4 August 2026**, down about 12% from July and still _below_ the $110/kg floor set under the US DoD–MP price-protection agreement; China's rare-earth price index closed the same day at 267.0, well below the early-2026 peak near 310. What has moved the ETF since is policy, not price: an upsized US government commitment behind USA Rare Earth's Brazilian offtake, and Beijing's own 2026–2030 materials strategy set out at a State Council briefing on 26 August. What has moved it over the past fortnight is neither, and the four-session run that closed September has now ended. REMX fell **5.1% to $72.42** on 10 September, its worst session since 23 June, inside a broad metals liquidation: US producer prices a tenth above forecast, a light Treasury buyback, the ten-year at 4.94%, silver −5.3% and the gold miners −3.5%. Then on 11 September the macro reversed — the S&P rose 0.86%, gold 0.6%, silver 1.1%, the gold miners 1.1% — and REMX fell anyway, **2.6% to $70.56**. That second session is the tell. A liquidation accounts for the first day; only something specific to rare earths accounts for the second. Three smaller falls followed, **2.3% to $68.93** on 14 September alongside the whole precious complex, 0.9% on 15 September and **1.0% to $67.61** on 16 September — the lowest close since 31 July, and the sixth down session running. Then the run ended, and policy is what ended it: **+2.2% to $69.10** on 17 September, flat at $69.06 on 18 September, and **+1.8% to $70.27** on 21 September, when Trump's **18 September** security agreement with Denmark and Greenland reopened the trade in anything holding an Arctic deposit. Greenland-linked micro-caps moved in multiples that session on press accounts not verified here, while the ETF moved 1.8% — roughly the difference between owning the story and owning a diversified basket. A third rise followed, **+1.0% to $70.97** on 22 September, and it was the weakest of the three: the whole minerals tier was bid that session but none of it led, and the day's largest mover on this page was a gold name moving with gold miners rather than with rare earths. Then the run ended on **23 September**, −3.1% to $68.76 — and the size of it is the point. Hawkish Fed remarks put the dollar index back above 100 and liquidated the whole metals complex that session, taking the gold miners down **4.4%** and silver **4.2%**; REMX fell _less_ than either. The rare-earth-specific headline of the day — Bessent extending the Busan Agreement from 10 November to **10 January**, removing a cliff edge that had been part of the scarcity premium — shows up not in the ETF but in the single names that ran past it, USA Rare Earth **−6.8%** and MP **−4.2%**. Then **24 September** extended it without the same excuse: **−2.7% to $66.93**, its worst session since 10 September and its **lowest close since 31 July**, on a day the dollar index merely held above 101 rather than surging and the gold miners fell 1.3% — a third of REMX's move. So this one is not a dollar session and not a precious-metals session; it is rare-earth selling on its own account, and the two largest names in the fund did not lead it, MP closing _higher_. Then **25 September** took it a _third_ session lower and barely at all, **−0.4% to $66.66**, extending the lowest close since 31 July on no rare-earth news of any kind — this page called it the fourth at the time, and the claim audit carries the correction. The fourth came on **28 September** and came hard, **−3.3% to $64.47**, its **lowest close since 6 November 2025**, on a day whose cause was an oil price and a rates repricing rather than anything upstream of a magnet. Then the run ended on **29 September**, **+0.5% to $64.79**, and the way it ended is the more interesting half: every US rare-earth and minerals name on this page fell that session, MP **2.2%**, USA Rare Earth **2.4%** and Lynas **1.0%**, while the fund holding them rose. A diversified basket can go up on a day its rare-earth constituents go down, which is the whole of this section in one session. Then **30 September** ran the same experiment backwards and got the mirror answer: REMX fell **0.8% to $64.26**, extending the lowest close since 6 November 2025, on a session in which **MP rose 4.1%** and Lynas 1.3%. Two consecutive sessions in which the fund and its largest holdings moved in opposite directions, once each way, is a better demonstration of a 0.37 correlation than any single day could be. Two sessions since have added almost nothing either way: **−2.0% to $62.95** on 1 October, its lowest close since **23 September 2025**, on the session Lynas fell nine per cent on its own announcement, **+0.4% to $63.21** on **2 October**, a fund barely moving while two of its holdings rose two and a half per cent and a third fell one and a half, **+0.5% to $63.54** on **5 October**, a third consecutive session in which the fund and its largest holdings disagreed about direction — MP, Lynas and Novonix all fell while the fund that holds them rose — and **+0.4% to $63.79** on **6 October**, which makes the same point by magnitude rather than by direction: **MP rose 3.6%** and **Lynas 2.3%**, and the fund holding both of them added four tenths of a point. Then **7 October** ended the disagreement and ended it downwards: **−3.0% to $61.88**, the **lowest close since 23 September 2025**, and this time every minerals name on this page went with it — MP **−4.6%**, USA Rare Earth **−3.7%**, Perpetua −3.1%, Novonix −2.5%, Lynas −1.1%. After three sessions in which the fund and its largest holdings disagreed about direction, one in which they all agreed is the less interesting outcome statistically and the more informative one here, because of what was _not_ moving: the S&P 500 fell **0.2%** that session and the gold trust 1.7%. A fund that falls thirteen times its equity index on a day the broad market is flat is not being liquidated along with everything else; it is being sold on its own account. No rare-earth disclosure of any kind is dated to it, and no company on this page announced anything. That is the fourth distinct cause this section has had to record in a month, after a metals liquidation, a security treaty and a dollar move. Then **8 October** ran it a second day, **−2.1% to $60.58**, the **lowest close since 12 September 2025**, and this session closes off the one explanation that was still available. On 10 and 23 September this page could attribute rare-earth selling to a liquidation of the whole metals complex. Not here: **gold rose 0.73%** and the **gold miners 1.47%** while REMX fell, with the S&P 500 down 0.47%. A fund that falls two per cent on a day the precious complex is bid and the broad market is flat is not being liquidated alongside anything; it is being sold on its own account for a second consecutive session. The single names went with it, USA Rare Earth **−4.5%** to its lowest close since 31 December 2025, Lynas **−2.9%** to its lowest since 2 January 2026 and Novonix **−2.8%** to a new two-year low, while MP sat almost still at −0.4%. Two causes are circulating for it and neither is dated to the session: August customs data showing **512 tonnes** of Chinese magnets shipped to the United States, released **20–21 September**, and the Xi–Trump state visit of **23–25 September**. Both are standing conditions of this tier rather than events in it, and the next scheduled meeting is APEC on **18–19 November**. Then **9 October** ended the run, and the way it ended is the useful part: **+0.6% to $60.97**, a rise into a session in which the whole precious complex was bid, **gold +1.57%** and the **gold miners +2.95%**, with the S&P 500 up 0.59%. A fund that adds six tenths of a point on a day the gold miners add three is not leading a recovery in rare earths; it is declining to fall while everything adjacent rallies. Its own constituents split underneath it, **Lynas +1.5%**, **Novonix +3.2%** and **Perpetua +2.6%** against **USA Rare Earth −0.6%**, which extended its lowest close since 31 December 2025 on the day its tier recovered, and **MP exactly unchanged** at $46.14. Then **12 October** resumed it and took the fund lower than any session in this whole account: **−4.6% to $58.18** on **1.18m shares** against 571,200, roughly double the volume, closing within a quarter of its low, and its **lowest close since 10 September 2025**. This time the constituents did go with it, and hard: **MP −6.0%**, **USA Rare Earth −5.9%**, **Lynas −3.1%**, **Novonix −3.1%**, **Perpetua −1.9%**. And the adjacent complex again refuses to explain it: **gold fell 0.63%** and the **gold miners 0.84%**, a fifth and a seventh of the fund’s move, so this is neither a dollar session nor a precious-metals liquidation. It is a fourth distinct day of rare-earth selling on its own account, with no company disclosure and no policy item dated to it. Two sessions since have taken it lower still, **−2.91% to $55.78** on **14 October**, its **lowest close since 8 August 2025** and again a smaller fall than every producer it is meant to proxy. Then **15 October** made the same point with the sign reversed: the fund closed **−0.07% at $55.74**, effectively unchanged, on **2.75 times** average volume and after an intraday low of **$53.75**, while **MP rose 4.16%** and **USA Rare Earth 7.46%**. Then **16 October** resumed the fall and took it through the August 2025 floor: **−2.71% to $54.23** on **1.34m shares** against a twenty-session average of 753,045, which is **1.78 times** average, and the **lowest close since 7 August 2025**. This time it did not fall less than its constituents, it fell more than most of them — MP −3.46%, USA Rare Earth −4.96%, Novonix −1.38% and Lynas −0.52%, with Perpetua _up_ 3.75% — which after seven runs of the opposite complaint is worth recording as the exception rather than a change of character. Then **19 October** put the character straight back and removed the last macro excuse available to it: **−1.60% to $53.36** on 702,000 shares, **0.89 times** a twenty-session average of 788,965, and the **lowest close since 7 August 2025**. This was not a risk-off session in any direction. The **S&P 500 rose 0.41%**, the **Nasdaq Composite 1.11%** and the **PHLX Semiconductor index 1.61%**; a fund that falls while its own equity market and the chip complex both rise is not being liquidated with anything. Nor is it gold: the **gold trust fell 0.61%** and the **gold miners 1.41%**, both less than the producers underneath the fund. And underneath it the tier split again, which is the eighth session in this account where the fund and its largest holdings disagreed: **USA Rare Earth −4.27%** to its **lowest close since 9 July 2025** and **MP −3.04%** to its lowest since 29 July 2026, against **Lynas +1.83%** on 1.29 times average volume and **Novonix +0.47%**. The fund fell a third as far as the worst name it holds and rose with none of the best. That is the fifth distinct cause this section has had to record, and it is the thinnest of them: no disclosure, no policy item and no macro, on turnover below average. Then **20 October** broke the run, and it is the first up-session of any size in this whole account: **+1.31% to $54.06** on 716,000 shares, **0.89 times** a twenty-session average of 802,845, and its best session since **21 September**. The split underneath it went the other way this time, which makes it the ninth disagreement here and the first of them in the fund’s favour: REMX rose more than four of the five producers it holds — MP +0.26%, USA Rare Earth −0.30%, Perpetua −0.27%, Novonix −2.31% — and less than one, Lynas at +3.35%. Nor did the tape hand it the move: the Nasdaq Composite was flat at −0.02%, the chip index fell 0.29%, the gold trust fell 0.54% and the gold miners 0.43%. A fund that rises 1.3% while its two largest producers sit still, the chip complex falls and gold falls is being bought on its own account, which is the same finding as the selling days with the sign reversed. Year to date REMX is now **32.0%** lower with a drawdown of **54.9%**, against Harmonic Drive up 37.7% YTD. Still not the same exposure — and note what moved this fund across the whole run: a metals liquidation, then rare-earth-specific selling, then a security treaty, then a dollar. No humanoid order book appears anywhere in that list.

If your intent is to own the magnet bottleneck specifically, REMX is a blunt instrument for it: the cleaner expressions are a rare-earth pure play or a magnet maker, not a diversified strategic-metals basket. If your intent is to own robotics, the component names above already carry the magnet input inside their own cost base — and rising NdPr prices are a _cost_ to them, which is part of why the co-movement between the two groups, while positive at 0.36, stops well short of making one a stand-in for the other — and 23 September showed the same thing from the other end, the minerals tier down 3–7% on a dollar move while Tokyo, closed that day, reopened bid. On the magnet input itself, the position has changed, and in the direction that matters. The last like-for-like SMM figure verified here is still 4 August's **$97.40/kg**, and that series remains behind a paywall with only aggregators quoting it. But a second price-reporting agency has now published on the record: **Benchmark Mineral Intelligence**'s Q3 2026 rare-earths review, dated **25 September 2026**, assesses Chinese PrNd oxide at **RMB729,500/t**, about **$108.7/kg**, up 1% through September, and ex-China PrNd at **$106,500/t** CIF, up 12%. That is a different assessor on a different basis, so it does not retire the SMM number or get spliced onto it; it is carried as its own series. What it does retire is this page's standing claim that NdPr sits comfortably _below_ the **$110/kg** DoD floor. At $108.7/kg it is within two dollars of it, which makes the floor close to the market rather than well under it, and that is a materially different fact about MP's price protection. The same review carries the number this section had been missing entirely. Heavy rare earths outside China have decoupled from Chinese domestic prices to a degree no NdPr figure captures: dysprosium oxide at **$3,500/kg** CIF Europe, **+56%** in September and **16.3 times** the Chinese domestic level, terbium at **$6,745/kg**, +23%, and yttrium oxide CIF North America at roughly **196 times** the Chinese level. That is the export-control premium priced, and it lands on exactly one name here: Lynas, the first and so far only commercial producer of separated dysprosium and terbium outside China, whose ADR fell **17.1%** over the same month the ex-China price of its scarcest product rose by half. A third assessor has now printed in October, and it moves nothing: the **China Rare Earth Industry Association** posted its own price index at **263.6** on **9 October 2026**, against the **267.0** this page carries from 4 August, with NdPr oxide quoted at **¥729.8–749.8/kg**. That brackets Benchmark’s 25 September RMB729,500/t rather than contradicting it, so the Chinese domestic price has been roughly flat for a month while the index has eased about one per cent, and the **$110/kg** DoD floor stays close to the market rather than comfortably under it. One figure circulating alongside it is not an assessment and is not carried as one: small-volume terbium transactions reported above **$12,000/kg**, against Benchmark’s $6,745/kg, are trade talk relayed by a paywalled outlet with no assessor behind them. None of this turns the divergence between the two tiers into a magnet-price story, but it does mean the magnet input is no longer the quiet part of this page.

05 Tier two — the minerals list, assessed

The five critical-minerals names get the same treatment as the component makers — but they stay on their own tier, because the data says they are an upstream policy trade rather than robotics exposure. Three of the post's five descriptions turned out to be wrong or out of date, and two of the companies have nothing to do with magnets at all.

| Company | What it actually is | Last | 1m | YTD | 1y | vs 52w high | Corr REMX | Corr robotics |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| MP MaterialsMP · NYSE | Mountain Pass mine + magnet plant; DoD-backed | $36.66 | −6.1% | −27.4% | −42.3% | −52.3% | 0.55 | 0.24 |
| USA Rare EarthUSAR · Nasdaq | Stillwater sold its first magnets in September; Round Top exploration | $6.87 | −31.7% | −42.3% | −52.0% | −79.1% | 0.38 | 0.16 |
| Lynas Rare EarthsLYSDY · ADR | Largest producer outside China; first Dy/Tb | $8.46 | +5.4% | +2.3% | −16.0% | −46.3% | 0.63 | 0.17 |
| Perpetua ResourcesPPTA · Nasdaq | Antimony & gold, Idaho — not a rare earth | $15.47 | −17.4% | −36.1% | −35.2% | −58.6% | 0.55 | 0.18 |
| NovonixNVX · Nasdaq | Synthetic graphite for batteries, Tennessee | $1.90 | −10.0% | −81.2% | −83.5% | −88.2% | 0.37 | 0.27 |Correlations are weekly returns over the same 103 weekly returns, re-bucketed by ISO calendar week. "Corr robotics" is against an equal-weighted basket of the six tier-one names. Novonix is shown restated for the one-for-ten ADS ratio change effective 27 August; its daily bars around that change keep appearing and vanishing, so the 28 August close is pinned to the verified figure rather than re-derived each run — see the method note.

**Sort them by whether they touch a magnet.** Only three of the five sit anywhere near the chain a robot actually consumes — and only two of those, MP Materials and USA Rare Earth, make finished NdFeB magnets. Perpetua is antimony and gold; Novonix is battery graphite. Both are legitimate critical-minerals businesses and neither goes into a robot joint.

The cards below rank them on that basis, closest to the magnet first. Note the correlation asymmetry throughout: even MP, the most robotics-adjacent name here, sits at 0.55 to REMX and 0.24 to the robotics basket. These trade as mining and policy assets, which is the whole point of keeping them on their own tier.

Mine to magnet Closest to the robot

### MP Materials

MP · NYSE

**19 November.** MP Materials closed at $36.66 (+1.97%) on 0.85 times its prior 20-session average volume.

**18 November.** MP Materials closed at $35.95 (+3.13%) on 1.19 times its prior 20-session average volume.

**17 November.** MP Materials closed at $34.86 (−3.41%) on 1.07 times its prior 20-session average volume.

**16 November.** MP Materials closed at $36.09 (−6.89%) on 0.92 times its prior 20-session average volume.

**13 November.** MP closed at $38.76 (+3.50%) while REMX moved −1.69%.

**5 November.** MP reported $147.7m in Q3 consolidated revenue and $18.6m in Magnetics revenue, all from precursors. The company completed GM qualification and expects saleable finished-magnet deliveries in Q4. [Form 10-Q](<https://www.sec.gov/Archives/edgar/data/1801368/000180136826000057/mp-20260930.htm>).

The only vertically integrated Western play, and the one genuine bridge between "rare earths" and "robot actuators." Q2 2026 revenue of **$108.5m (+89%)** came with NdPr production of 840 tonnes (+41%) and sales of 1,006 tonnes — more than double the prior year — pushing adjusted EBITDA positive and narrowing the diluted loss to $0.11 from $0.19.

The government backstop is unusually deep: the Department of Defense took **$400m of preferred stock** in July 2025 to become the largest shareholder, with a ten-year **$110/kg NdPr price floor**, and has since signed a ten-year offtake for 100% of the output of the planned **10X facility** in Northlake, Texas — a $1.25bn-plus campus commissioning from 2028 that is meant to lift MP's _total_ US magnet capacity to roughly 10,000 tonnes a year, not to produce 10,000 tonnes on its own. Beijing noticed: China's Ministry of Commerce added MP to its export-control blacklist in June 2026, and the stock fell nearly 30% through July.

The 13 Aug earnings call surfaced a new line: a multi-year, nine-figure agreement signed in July to supply separated gadolinium oxide to an undisclosed US aerospace and defense manufacturer, recovered as a co-product of the existing Mountain Pass ore stream. Terms are undisclosed so it cannot be sized precisely, but it is the clearest evidence yet that MP is building toward a multi-element refinery rather than a single-product (NdPr) business.

**Two sessions have taken the stock back to where it started the week, and neither carried a disclosure.** **6 October** added **+3.61% to $48.53**, the largest Western gain on this page that day and nine times REMX's own move; **7 October** handed all of it back and more, **−4.57% to $46.31**, its **lowest close since 1 October**, inside a session in which every minerals name here fell and the S&P 500 moved 0.22%. The standing story is unchanged and is still the company's own: commercial magnet shipments to **General Motors** from the Fort Worth Independence plant are guided for the fourth quarter, which is now running, and the magnetics segment booked **$16.5m** of the $108.5m Q2 revenue. On the input side the **$110/kg** NdPr floor under the DoD agreement is no longer comfortably above the market: Benchmark's 25 September assessment puts Chinese PrNd at about **$108.7/kg**, within two dollars of it, which makes the price protection worth materially less than it was when this card was first written. A third session followed on **8 October** and it barely moved at all, **−0.37% to $46.14**, on a day REMX fell 2.1% and the rest of the tier fell three to five times as far: the fund and its largest holding disagreeing about magnitude rather than direction, which is this card’s argument restated. Four sessions since took it a good deal lower, the last of them in the CPI selloff: **−5.84% to $39.95** on **14 October**, on **1.88 times** average volume and its **lowest close since 29 July**. **15 October reversed part of it, +4.16% to $41.61** on 1.16 times average, off an intraday low of **$38.60**, its best session since **21 September** — and it arrived with the gold miners up 4.34% and the ten-year yield 10bp lower rather than with anything rare-earth. **16 October gave most of that back**, **−3.46% to $40.17** on 0.97 times average volume, inside a chip-led risk-off session rather than anything rare-earth. Twelve-month return **−52.1%** against a 16 October 2025 base of **$83.84**, and **52.2%** below the high, the third-deepest drawdown in tier two behind Novonix at 91.0% and USA Rare Earth at 67.9%. Part of the previous run’s improvement was a date rather than a price: this name’s own 52-week high rolled from $91.99 to **$84.15** on 16 October, which was **4.1** of the six points the drawdown cell came in, and it rolled once more on 17 October to **$83.99**, the only one of eleven to move that run.

Q3 rev **$147.7m** Q3 precursor sales **$18.6m** NdPr floor **$110/kg** Gadolinium deal **9-figure, multi-yr**

**20 October: +0.26% to $39.05 on 0.68 times average volume, effectively still, in a session the Dow rose 1.02% and the Russell 2000 1.65%.** Nothing is dated to it. Both lookback columns moved far more than the price did, and in opposite directions, which is the clearest single illustration this page has of why they are read with the base in hand. The 52-week high rolled **$80.89 to $76.80** as the 20 October 2025 session dropped out of the window, taking the drawdown cell from 51.9% to **49.2%** and under fifty per cent for the first time in this account, on a quarter-point session. The twelve-month base rolled the other way on the very same day, $80.79 to **$82.90**, so the 1y cell _worsened_ from −51.8% to **−52.9%** on a session the stock rose. Neither number is a judgement about the business.

**Watch**First revenue from saleable finished magnets for GM, expected in Q4 2026 following qualification. Q3 segment sales were from magnetic precursors.

**2 November: a third consecutive fall, three sessions before the print that settles this card.** **−3.08% to $38.35** on **0.81 times** average volume, 4,227,300 shares against 5,224,795, opening $40.19 and closing near its $38.13 low, its **lowest close since 26 October**. The drawdown widens to **50.1%**, the four-week cell to **−18.4%** on a base rolled onto 2 October’s $46.97, the year to date to **−24.1%** and the twelve-month to **−39.2%**. Nothing was filed, and it is not a gold session or a market session: the gold trust rose 1.65% and the gold miners 2.67% while the S&P 500 fell half a point. **Third-quarter results are confirmed for after the close on Thursday 5 November 2026**, call 5:00pm ET, and the magnetics revenue line against the $16.5m of the second quarter, plus any first commercial Fort Worth shipment to General Motors, are the only figures in it that move this row.

Magnets Already shipping

### USA Rare Earth

USAR · Nasdaq

**19 November.** USA Rare Earth closed at $6.87 (+4.73%) on 1.56 times its prior 20-session average volume.

**18 November.** USA Rare Earth closed at $6.56 (−1.13%) on 1.96 times its prior 20-session average volume.

**17 November.** USA Rare Earth closed at $6.64 (−6.94%) on 1.02 times its prior 20-session average volume.

**16 November.** USA Rare Earth closed at $7.13 (−12.62%) on 1.13 times its prior 20-session average volume.

**13 November.** USAR closed at $8.16 (+1.12%) on 1.30 times its prior 20-session average volume. Its [12 November 10-Q](<https://www.sec.gov/Archives/edgar/data/1970622/000197062226000072/usar-20260930.htm>) confirms the November start of Serra Verde’s Phase 1 carbonate sales.

**9 November.** the Q3 report disclosed September sales of sintered NdFeB block magnets from Stillwater and October commercial-scale heavy mixed rare-earth carbonate shipments from Serra Verde. Q3 consolidated revenue was $3.0m. The release gives no separate magnet-sales figure. [Issuer results](<https://investors.usare.com/news-releases/news-release-details/usa-rare-earth-reports-third-quarter-2026-financial-results>).

The source post is still wrong about the timeline, but this page was wrong about the other end of it, and the correction matters more. USAR did **commission its Phase 1a commercial magnet line** at Stillwater, Oklahoma in **Q2 2026**, pulling the plant forward by about two years from the 2028 date the post implies — and the company says it “has commenced commercial production.” What it has _not_ done is sell any of it. The Q2 10-Q is explicit that Stillwater “has recently been commissioned and has commenced commercial production; however, we have not begun generating revenue from” sintered NdFeB magnets, and the **15 September** risk-factor supplement goes further, describing Stillwater and Blacksburg as “under development and… not yet completed” with “no… revenues to date from the sale of neo magnets, critical minerals or rare earth minerals.” This page previously said USAR “began filling customer orders” in Q2 2026. It did not. Phase 1a is _guided_ to a 600 mtpa run-rate by the end of Q4 2026 and 1,200 mtpa combined with Phase 1b in Q1 2027, against roughly 5,000 tonnes at full build — every one of those a target, none of them yet a shipment.

Q2 2026 revenue was $5.8m against a $10.3m net loss (down from $142.5m a year earlier, when the company had no commercial revenue at all). None of that $5.8m is magnets: the 10-Q attributes it “all” to Less Common Metals, the Cheshire metals subsidiary acquired in November 2025, and states that all revenue “is derived from sales of casting and strip casting” — at a **gross loss** of $1.6m on a negative 27.2% gross margin. The balance sheet is the actual headline: **$1.53bn of cash** at quarter-end, up from $359.9m at year-end 2025. USAR is deploying it into corporate structure as much as capacity — it closed the acquisition of Texas Mineral Resources Corp on **7 Aug**, giving it outright ownership of Round Top, and finalized a roughly 13.6% stake in French rare-earth processor Carester. It agreed to buy Brazil's Serra Verde, the only scaled ex-China producer of all four magnet rare earths, for about $2.8bn; shareholders approved the share issuance on **28 August** — 108,248,297 votes for against 1,403,269 opposed, with 16,879,393 abstentions — and the combination **completed on 3 September 2026**, announced the following morning and filed the same day on Form 8-K under Item 2.01. This page carried it as pending for a week longer than it was; that was wrong, and the correction is the more important half of this card.

What actually closed is bigger than the price tag suggested. USAR paid **$300m in cash plus 126,849,307 new shares** to Serra Verde's shareholders, lenders, warrant holders, employees and consultants, and took on Serra Verde's financing with it: a facility of up to **$565m from the US International Development Finance Corporation**, sized to fund the Brazilian optimisation and expansion programme through to positive cash flow. The Goiás operation has been producing since January 2024 and is finishing a commissioning and optimisation programme whose first stage targets a run-rate of about **4,000 tpa of total rare earth oxide by the end of 2026**. Sir Mick Davis joins the board; Thras Moraitis becomes President immediately and succeeds Barbara Humpton as chief executive on **1 October**. USAR now owns an operating heavy rare-earth mine, a magnet line and a separation stake — upstream, midstream and downstream in one company, which is what MP has and no one else outside China does.

The financing behind that deal landed on **24 August 2026** and is the most consequential development on this page since the last refresh. The special-purpose vehicle that will buy 100% of Serra Verde's Phase 1 output completed an upsized **$1.55bn capitalisation**: a **$750m Department of War investment**, raised from the $500m originally contemplated, a bank senior secured borrowing-base revolver of up to $500m, and a US government forward purchase contract for not less than $300m of rare-earth product over five years. The offtake runs **fifteen years** on take-or-pay terms with guaranteed price floors on the magnetic rare earths — including, on the company's account, the first such floors anywhere for **dysprosium and terbium**. Read against the $110/kg NdPr floor MP already has, Washington is now underwriting the price of the heavy elements too, which is the part of the chain China actually controls.

And on **9 September 2026** it broke ground on a second magnet plant, which this page should have carried a run earlier. The site is **Blacksburg, South Carolina** — 124 acres in Bailey Industrial Park, Cherokee County, about **800,000 square feet**, roughly **$1.2bn** of investment and some **490** manufacturing jobs, commissioning in **2028**. The capacity is the part that matters: **6,400 tpa of sintered NdFeB magnets** plus 5,000 tpa of strip-cast metal and alloy. On its own that is larger than Stillwater's full build of roughly 5,000 tonnes, and it places a second and bigger magnet line behind the company's own metal. Measured against MP's 10X campus, which is meant to lift MP's _total_ US magnet capacity to about 10,000 tpa from 2028, Blacksburg is the largest single increment to announced ex-China magnet tonnage anywhere on this page. The company says it screened nearly 275 sites before choosing it. None of this moved the stock, which fell 8.8% across the two sessions that followed.

The Round Top deposit in West Texas is the part that is genuinely years out, and the distinction matters: the magnets Stillwater is producing are made from purchased feedstock, not from USAR's own mine — and the 15 September supplement flags that too, warning that feedstock delays at Less Common Metals have already constrained the plant, and that China's **22 June 2026** export-control designation “has had and is expected to continue to have an adverse impact on our ability to source key raw materials.” Round Top did produce a first dysprosium oxide sample at 99.1% purity in January. Like MP, USAR was added to China's export-control list in June 2026. The stock sits **66.2% below its 52-week high**, the steepest drawdown on this page bar Novonix, and closing the deal did nothing to lift it. It fell 5% on the day the SPV funding was announced, 6.5% on the day of the vote, 7.4% on 31 August when the result was disclosed and 3.1% on 1 September; it then drifted through six sessions either side of completion, fell 3.4% to $17.06 on 9 September, a further **6.0% to $16.04** on 10 September — the largest fall on the page that day, though that one arrived with the whole metals complex rather than alone — and **3.0% to $15.56** on 11 September, which did not: the complex rebounded that session and USAR made a fresh low regardless, its weakest close since 31 July. The one-way tape broke only briefly — +1.0% to $15.71 on 14 September, one of only two tier-two names to hold a session that took REMX to a six-week low, then −2.0% to $15.40 on 15 September and −1.9% to $15.10 on 16 September, a fresh low for the run. Then the move finally came, and not from the company: +3.5% to $15.63 on 17 September, −1.7% to $15.37 on 18 September, and **+9.2% to $16.78** on 21 September — the largest single gain anywhere on this page in this period — on Trump's security agreement with Denmark and Greenland, followed by a much quieter +1.3% to $16.99 on 22 September — and then the whole policy bid came out at once, **−6.8% to $15.83** on **23 September**, the largest fall anywhere on this page that session and its lowest close since 18 September. That one is not market beta: the gold miners fell 4.4% and REMX 3.1% on the same dollar move, and USAR went half as far again as either, on the day Bessent extended the trade truce past its November expiry. The symmetry is the lesson — two sessions of policy bid on a treaty about an island it has no assets on, one session of policy unwind on a truce it is not party to either. USA Rare Earth owns nothing in Greenland and is not a signatory in Busan. A fourth session then took it lower again with no policy in it at all, **−2.8% to $15.38** on **24 September**, and the run has not stopped since: −1.3% to $15.18 on 25 September, **−5.0% to $14.42** on **28 September** on the oil-and-rates repricing, and **−2.4% to $14.08** on **29 September**, a fifth consecutive fall and its **lowest close since 29 July**, on a session when the benchmark that holds the tier closed higher. Three more sessions have carried it lower still: **−2.0% to $13.80** on 1 October and **−1.5% to $13.59** on **2 October**, its **lowest close since 29 July 2026**. The selling has continued without a disclosure attached to any of it, and **14 October** was the heaviest of it: **−8.51% to $10.32** on **20.92m shares**, **1.56 times** average, its **lowest close since 9 July 2025**. **15 October then reversed it harder than anything else in the tier: +7.46% to $11.09** on **23.58m shares**, **1.72 times** average, after first trading as low as **$9.82**. **16 October resumed the fall**, **−4.96% to $10.54**, and the turnover is the part worth recording: **10.96m shares** against a twenty-session average of 14.32m, or **0.77 times** normal, so Friday extended Thursday’s selling without fresh participation behind it. That leaves it **67.9%** below its high, the steepest drawdown on the page bar Novonix, and **30.2%** lower over a month. Its twelve-month return now reads **−63.1%**, and that figure needs its date more than any other on this page: the year-ago comparison has rolled onto **16 October 2025's $28.54**, a close struck inside the October 2025 spike this name has never been back to, so the column is measuring the stock against its own mania rather than against a normal session. The year to date is **−11.4%**. The market is pricing the dilution ahead of the assets — 126.8m new shares against 244.7m outstanding is arithmetic, the assets require execution — and pricing the policy tape ahead of both. The claim audit carries the Greenland attribution.

Line **commissioned Q2 2026** Phase 1a capacity **600 tpa** First magnet sales **September 2026** Cash, equivalents and restricted cash at 30 Sep **$1.16bn** SPV funding **$1.55bn** Serra Verde **closed 3 Sep** Blacksburg **6,400 tpa** Off high **−79.1%**

**20 October: −0.30% to $10.06 on 0.70 times average volume, extending the lowest close since 9 July 2025 by three cents.** The drawdown is **69.3%**, still the second deepest here behind Novonix. The twelve-month cell fell four and a half points to **−68.2%** and none of that is the session: the year-ago base rolled $27.73 to **$31.59**, a 14% one-day step onto the day the market traded China’s export controls. Nothing is dated to it. The open items are unchanged and both are now close: third-quarter results are not scheduled, and the Stillwater **600 mtpa** run-rate and the first magnet invoice are each guided to a quarter with ten weeks left in it.

**Watch**Finished magnet prototype sales to multiple Stillwater customers by year-end. The Q3 release targets Q2 2027 for both Line 1b commissioning and the 4,000 tpa Serra Verde TREO run rate.

Separation Heavy rare earths

### Lynas Rare Earths

LYSDY · ADR

**20 November in Sydney.** Lynas settled at A$12.44 (−2.74%) in Sydney. The ADR closed on 19 November at $8.46 (+1.32%) on 1.41 times its prior 20-session average volume.

**19 November in Sydney.** Lynas settled at A$12.79 (+3.40%) in Sydney. The ADR closed on 18 November at $8.35 (+2.96%) on 0.85 times its prior 20-session average volume.

**18 November in Sydney.** Lynas settled at [A$12.37](<https://stockanalysis.com/quote/asx/LYC/history/>) (+2.83%). The ADR closed on 17 November at $8.11 (−7.63%) on 2.36 times its prior 20-session average volume.

**17 November in Sydney.** Lynas settled at A$12.03 (−8.24%). The ADR closed on 16 November at $8.78 (−1.90%). The company’s [AGM notice](<https://announcements.asx.com.au/asxpdf/20261029/pdf/075xzndxqk7000.pdf>) sets 25 November at 10:00 AEDT.

**16 November in Sydney.** Lynas closed at A$13.11 (−0.61%). Its US ADR last closed on 13 November at $8.95 (−0.89%).

Operationally the strongest of the five, and now with a full year on the board. FY26 results, reported **26 August 2026**, were records across the line: revenue of **A$977.9m, up 76%**, net profit after tax of **A$222.4m** against A$8.0m the year before, REO production of 13,089 tonnes (+25%), NdPr production of 7,260 tonnes (+11%) and a record average realised price of A$80.7/kg. More strategically, it became the **first commercial producer outside China of separated dysprosium and terbium** — first terbium oxide in June 2026, on a heavy rare-earth circuit of about 1,500 tonnes a year.

The market did not take it as a clean beat. The ADR fell **5.9%** on results day before recovering 5.0% the next session, then gave back another 8% into 1 September with the rest of the mining complex, and rose in four sessions to $11.23 on 8 September. That run ended abruptly: −2.9% to $10.91 on 9 September, then **−4.3% to $10.44** on 10 September in the broad metals selloff, **−2.2% to $10.21** on 11 September after it reversed, −3.2% on 14 September, −0.7% on 15 September and **−2.9% to $9.53** on 16 September — its lowest close since **6 January**. It has since recovered the handle: +2.5% to $9.77 on 17 September, **+2.9% to $10.05** on 18 September, its first close back above $10 since 11 September, then the one faller on the page's Greenland session, −0.3% to **$10.02**, then the tier's second-strongest session the next day, **+3.1% to $10.33** on 22 September, and then **−2.6% to $10.06** on 23 September in the dollar-led liquidation. Nothing from the company behind any of them, in either direction; the Sydney ordinary line followed a session later, **−1.9% to A$14.39** on 24 September, in an Asian rare-earth session the wires attributed to the truce extension. Over a month the ADR is **18.2%** lower against REMX's **16.5%** — and that ordering, which had flipped for a week, has flipped back: the single name is falling marginally faster than the fund again, which is what a 0.63 correlation and a higher beta ordinarily produce. A fifth session followed, **−1.2% to $9.94** on **24 September**, its lowest close since **17 September** and its first below $10 since then. Its twelve-month column crossed below zero on 23 September, from +1.0% to −8.1%, the year-ago base moving up from $10.23 to $10.95 doing most of that, and it now reads **−38.1%** against a 16 October 2025 base of $12.33. Three more falls have taken it lower still — −3.8% to $9.58 on 28 September and **−1.0% to $9.48** on **29 September**, its **lowest close since 6 January 2026**, on a session when the benchmark that holds it closed higher — while the Sydney ordinary line went the other way a day later, **+1.4% to A$13.83** on 30 September. Nothing from the company behind any of it. The reason is on the cost side: capital expenditure on the heavy rare-earth circuit has escalated **63%, to A$294m from A$180m**, which the company attributes to sourcing equipment outside China. That is the price of the scarcity premium, stated in cash.

That heavy-element capability is the real scarcity, and it has just been repriced. **Benchmark Mineral Intelligence**'s Q3 2026 review, dated **25 September 2026**, assesses terbium oxide at **$6,745/kg** CIF Europe, up **23%** in September, and dysprosium oxide at **$3,500/kg** CIF Europe, up **56%** in a single month and **16.3 times** the Chinese domestic level, with yttrium oxide CIF North America at roughly **196 times** its Chinese price. Those ex-China terbium figures are well above the $4,500–$4,700/kg range this page carried through the summer, so the scarcity premium on the heavy elements is widening rather than holding. Dysprosium and terbium are what keep magnets stable at motor operating temperatures, which is why this is the part of the chain a robot actually cannot substitute away from. The uncomfortable part for the thesis is the tape: over the month in which the ex-China price of its scarcest product rose by half, the ADR fell **14.5%**. Either the market does not price a CIF Europe assessment as revenue for an Australian producer selling largely under contract, or it is not reading it at all. The post's claim that Lynas is "now building a US processing facility" is out of date: on **16 March 2026** the US government redirected the US$96m earmarked for the Seadrift, Texas heavy rare-earth plant into a four-year purchase agreement for Lynas product instead, with a US$110/kg NdPr floor. The Texas plant is in limbo, not under construction.

Q1 FY27 rev **A$310.1m** NdPr **2,082t** ASP **A$102.1/kg** Cash **A$1,212m** Seadrift **unmentioned**

**20 October: +3.35% to $8.03 on 1.14 times average volume, the largest rise anywhere on this page and a second consecutive green session while the rest of the tier sat still.** Its twelve-month cell fell five and a half points regardless, −36.4% to **−42.0%**, on a year-ago base rolling $12.21 to **$13.84**: a 13% one-day step, because 20 October 2025 is the session this ADR re-rated on the Chinese export controls. The best day on the page and the largest base roll on it are the same name. The September-quarter report is due late October and is the next dated test of the heavy rare-earth output this card turns on, including whether the ex-China dysprosium and terbium premium recorded in section 04 shows up in a realised price.

**Watch****The Meteoric scheme**, which as of 1 October is the largest single thing on this card. Lynas signed a binding scheme implementation deed to acquire 100% of **Meteoric Resources** at **0.0207 new Lynas shares** per Meteoric share: no cash, an implied **A$968m** on a 60-day VWAP, a **68.4% premium**, and about **5.9%** of Lynas to Meteoric holders. The asset is **Caldeira** in Minas Gerais, on Meteoric's description the largest known ionic-clay JORC rare-earth oxide resource outside China, with a definitive feasibility study completed in July 2026. Three dates to hold: a scheme meeting in **January 2027**, implementation targeted for **March 2027**, and Brazilian critical-minerals approval somewhere in between, any of which can move. The tension is worth stating plainly rather than resolving. Ionic clay is the cheap end of the industry's cost curve and a second geography outside Malaysia and Australia, so the strategic case is real; but production is years out, the consideration is paper, and the dilution lands now. Shareholders voted the second half, 8.6% of it in a session. What the deal does _not_ do is touch the heavy circuit, which is the part of this thesis that matters here, so everything previously on this line still stands: whether **Seadrift** is revived or quietly shelved, and the dysprosium/terbium ramp against that 63% capex escalation, now against an ex-China heavy price that moved more in September than the ramp will in a year. The heavy circuit is the part China cannot easily replace and the part no one else outside China has, though USA Rare Earth's US-backed Dy/Tb price floors are the first sign that someone else intends to be paid for it too. The scheme has now also acquired a downstream half, reported rather than released: on **12 October 2026** Minas Gerais mines and energy minister **Alexandre Silveira** told _MINING.COM_ that Lynas has proposed a rare-earth oxide separation plant in the **Poços de Caldas** region, with development bank **BNDES** evaluating financing on unnamed sources and a technical visit to the Australian operations due before year end. Hold it at its actual weight: a minister’s written comment and an evaluation, with no Lynas release, no approval, no construction decision and no capital figure, so it is a proposal on this card rather than capacity, and it does not touch the heavy circuit either. And a permanent CEO is still outstanding: **Pol Le Roux** has run the company on an interim basis since 1 July, which is a notable thing to remain true of a company signing a billion-dollar scheme. **27 October: +7.69% to $8.40 on 1.69 times average volume, the heaviest relative turnover in tier two.** Nothing was filed. The row is **8.25%** above its 22 October close, the largest three-session recovery in the tier, and the drawdown comes in **3.8** points to **46.7%**. The twelve-month cell gains **12.8** points to **−25.4%**, of which 8.1 is a year-ago base cut $12.63 to **$11.26**, so most of that column is the calendar rather than the tape. **The next real test is two days away:** the September-quarter activities report is scheduled for **29 October 2026**, with an analyst and shareholder briefing at 10am Sydney time. That is where heavy rare-earth output and any movement on **Seadrift** would appear. The company’s own standing language on Seadrift is that there is “significant uncertainty as to whether the construction… will proceed and, if so, in what form”, and no decision has been published; the March 2026 binding letter of intent with the Department of War for about **$96m** of oxide over four years is the only firm US commitment behind it. **The report arrived on schedule, it is a good quarter, and Seadrift does not appear in it.** Lodged with the ASX at **21:12 UTC on 28 October**, an hour and twelve minutes after the New York close, so no close on this page has priced it. Gross sales revenue **A$310.1m**, up 55% on the prior corresponding period and the highest since Q4 FY22; total rare earth oxide production **4,254t** against 3,993t a year earlier and 3,481t in the June quarter; **NdPr production 2,082t** against 2,003t and 1,857t; **23 tonnes of Dy and Tb**; sales receipts A$266.6m; closing cash and short-term deposits **A$1,212m**; capex, exploration and development A$28.7m. The number this card asked for is the realised price, and it answers directly: the average selling price across all rare earth products was **a record A$102.1/kg**, up **88%** year on year, which the company puts down to market pricing, to bundling light with heavy rare earths for magnet customers, and to floor-price agreements that now cover **over 50%** of production. So the ex-China heavy premium recorded in section 04 is showing up in a realised number, not just in an assessment. The June quarter’s ore-concentrate quality problem is resolved on the company’s own account. **On Seadrift the document is silent, and the silence is structured.** Searched in full it contains no instance of _Seadrift_, _Texas_ or the United States. Its “Lynas Growth Projects” section names one asset, the expanded heavy rare earths facility at Lynas Malaysia, sequenced to gadolinium then yttrium. The Towards 2030 partnerships it names are **JS Link** in Kuantan, **Malaysia** (3,000 tpa sintered NdFeB, about A$50m of equity, exclusive supply to January 2038), **LS Eco Energy** in **Vietnam** (metal making including heavy rare earth metal, about A$29m of cross-subscribed convertibles each way), and, after the quarter end, **Meteoric Resources** in **Brazil**. Three growth geographies, none American. Read that as what it is: not a cancellation, and not the “significant uncertainty” language being withdrawn either, but a quarterly that lists where the next capacity is going and leaves the Texas plant off the list.

**2 November: one of only two names in tier two to rise, on a session four of the five fell between three and eight per cent.** The ADR closed **$8.39, up 0.60%** on **1.22 times** average volume, 552,900 shares against 452,730, which leaves the row **1.5%** higher year to date, the only positive year-to-date cell in tier two, and **46.7%** below its high, the shallowest drawdown on the tier. The primary LYC.AX line has since closed **3 November at A$11.97**, down 1.24% from A$12.12, so the two lines disagree on direction across the one-session ADR gap: up 0.60% on Monday in New York, down 1.24% on Tuesday in Sydney. Nothing was filed on either side of it, and Seadrift is still absent from the capacity list.

Antimony Not a magnet input

### Perpetua Resources

PPTA · Nasdaq

**19 November.** Perpetua closed at $15.47 (−8.05%) on 1.68 times its prior 20-session average volume.

**18 November.** Perpetua closed at $16.83 (−1.95%) on 1.27 times its prior 20-session average volume.

**17 November.** Perpetua closed at $17.16 (+0.88%) on 1.53 times its prior 20-session average volume.

**17 November tungsten update.** The [non-binding memorandum](<https://perpetuaresources.com/perpetua-resources-announces-non-binding-mou-with-a-major-american-aerospace-defense-company-to-evaluate-domestic-tungsten-opportunities/>) could fund work on potential domestic tungsten supply. It contemplates up to $55.5m of non-equity funding and a possible right to acquire up to half of future tungsten output. Definitive terms and a separate permitting process would be needed for development.

**16 November.** Perpetua closed at $17.01 (−4.33%) on 1.82 times its prior 20-session average volume.

**16 November project update.** Perpetua’s [Stibnite construction notice](<https://perpetuaresources.com/project/local-notices/>) says critical-path work on Burntlog Route is continuing. The 17 November webinar is scheduled for 11:00 EST.

**13 November.** Perpetua closed at $17.78 (−0.22%). The [13 November 10-Q](<https://www.sec.gov/Archives/edgar/data/1526243/000110465926127248/ppta-20260930x10q.htm>) puts expected completion of EXIM documentation and conditions in **Q1 2027**, after the earlier Q4 2026 target. The final investment and construction decision is also expected in Q1. The $2.9bn facility has EXIM board approval and remains subject to closing conditions.

**4 November.** Perpetua published Stibnite drill results including 21.3 metres grading 3.2 g/t gold and 0.9% tungsten at SB597, measured downhole. The disclosure is an exploration result. The next financing milestone remains EXIM financial close. [Issuer release](<https://perpetuaresources.com/perpetua-announces-new-gold-discovery-at-stibnite-gold-project-continued-high-grades-from-clark-tunnel-fault-zone/>).

A well-financed critical-minerals project that has almost nothing to do with robotics. The EXIM board approved a **$2.9bn loan** in May 2026. Perpetua expects definitive documentation and closing conditions in Q1 2027. Earlier financing included a $255m private placement in October 2025 backed by Agnico Eagle and JPMorgan. Construction at Stibnite in Idaho is underway.

Two caveats the post skips. First, this is **predominantly a gold mine**: reserves of roughly 4.8 million ounces and about 450,000 ounces of gold a year over the first four years, with antimony supplying an estimated 35% of US demand alongside it. Second, production does not begin until **2029**. Antimony goes into flame retardants, ammunition and alloys — not robot joints. It earns its place on a critical-minerals list, not on a robotics one.

The early October comparison adds a third caveat. At that point Perpetua showed **−16.7%** over the past month, and that relationship inverted for a week before reasserting itself: bullion was **4.4% lower** over the same 30 days and the gold miners **10.4% lower**, so Perpetua fell further than both — a mild amplification of the equities rather than of the metal, which is the ordinary configuration for a high-beta gold miner. The explanation is in the sessions below: several of them moved on something other than gold. On 31 August gold _rose_ and Perpetua fell 4.7% anyway, alongside the gold miners; on 1 September gold fell 1.9%, the miners **3.9%**, and Perpetua **5.8%**. Then it ran in reverse: on 2 September the miners rose **3.1%** and Perpetua **4.1%**. On 3 September the relationship held in direction but not in size — gold **1.9%**, the miners **4.0%**, and Perpetua only **2.4%**, for once lagging the equities it usually amplifies. Then on 4 September it decoupled altogether: gold fell **1.4%** and the miners **2.2%**, and Perpetua did nothing at all, +0.2% to $25.09. On 8 September the usual relationship resumed quietly: gold eased 0.8% and Perpetua 1.3% to $24.76. On 9 September it went missing again: bullion added 0.9% and the gold miners 1.1%, and Perpetua closed **unchanged at $24.76**, to the cent. Then on 10 September it came back with force — bullion **−1.7%**, the gold miners **−3.5%** and Perpetua **−4.4% to $23.67**. On 11 September it went quiet once more: bullion rose 0.6% and the miners 1.1%, and Perpetua eased **0.2% to $23.63**, its lowest close since 1 September and the only tier-two name that barely moved on a day the rest of the tier fell. Then on **14 September** it amplified the complex again and harder than either leg of it: bullion **−1.3%**, the gold miners **−3.1%**, and Perpetua **−4.6% to $22.55**, its lowest close since 4 August — before it recovered 0.6% to $22.69 on 15 September, the only tier-two name to rise on a session when gold fell again. The three sessions since split the same way. On **17 September** it tracked the equities closely — the miners **+3.4%**, Perpetua **+3.8% to $23.37** — and on 18 September it tracked them again, the miners −0.5% and Perpetua −0.6% to $23.24. Then on **21 September** it decoupled upward for the first time in this run on something that was not gold at all: bullion **−0.7%**, the gold miners **−1.1%**, and Perpetua **+1.9% to $23.68**, moving with the rare-earth complex on the Greenland pact rather than with the metal it is actually levered to. On **22 September** it went back to the miners and amplified them mildly: bullion **+0.4%**, the gold miners **+3.6%**, and Perpetua **+4.6% to $24.76** — the largest single gain anywhere on this page that session, and still an ordinary gold-equity beta rather than an antimony event, with no company disclosure attached to it. That is the exception that proves the rule this card exists to state — antimony is a critical mineral, so it catches a critical-minerals bid, and none of that is the gold mine that will supply 450,000 ounces a year from 2029. Four times in thirteen sessions it has ignored a moving metal and a moving miners index, and twice it has multiplied one. That session carries a caution worth recording, because this page names GOLD as the instrument to check before attributing a Perpetua move. GOLD is Barrick, and Barrick rose **11.2%** that day against a falling metal and a falling miners index. A single-name divergence that wide is company-specific, not a gold signal — which is why the comparison above is drawn against the miners index rather than against Barrick alone. Then **23 September** put it back in its ordinary place and took the year with it: bullion **−1.8%**, the gold miners **−4.4%**, and Perpetua **−5.8% to $23.32**, its lowest close since 18 September — a mild amplification of the miners in a dollar-led liquidation, with no antimony in it and no company disclosure, exactly as on the way up the session before. Its year-to-date column crossed back below zero in the process, from +2.3% to **−3.7%**, one session after crossing above. Then **24 September** broke the pattern in the one direction it had not yet gone: Perpetua rose **0.4% to $23.42** while the gold miners fell **1.3%** and bullion **0.3%** — the first session in this run where it outperformed gold rather than amplifying it, and again on no company disclosure. Then **25 September** put it straight back in line and lagging slightly: the gold miners rose **0.6%**, bullion **0.4%**, and Perpetua **0.3% to $23.49**. Then **28 September** delivered the largest single fall this page has recorded on any name this run, and it was still the same relationship: **−7.1% to $21.82**, its lowest close since **4 August**, against gold miners **−5.4%** and bullion **−3.9%** — an amplification of the equities, on a session whose cause was an oil price and a rates repricing rather than anything in the ground. Brent went back toward $106 after the White House rejected Iran's proposal to reopen the Strait of Hormuz, the ten-year Treasury yield passed 5.2%, and gold had its sharpest fall in this run. Not one part of that chain is antimony, and no company disclosure accompanied it. Across eighteen sessions Perpetua tracked the mining equities, not the metal, and one session of mild outperformance does not change that. The exposure is to gold _equities_, not to gold. Then **29 September** made the point from the other side: Perpetua fell **1.3% to $21.53**, its lowest close since **4 August**, on a session when bullion rose 1.3% and the gold miners rose 1.3% — the first time in this run the name has gone the opposite way to the equities it normally amplifies, and again with nothing published. Eighteen sessions of tracking the miners, one of tracking nothing. Two more sessions have continued it and the second is the one to read: **−3.0% to $20.93** on 1 October on a session the gold trust _rose_ 0.5%, and **−2.6% to $20.38** on **2 October** against a gold trust down only **0.7%** — roughly four times the metal, with nothing published either day. Two quieter sessions have put it back in line. It eased **0.2% to $20.33** on **5 October** against a gold trust down 0.1%, then rose **1.43% to $20.62** on **6 October** against a gold trust up **0.72%**, roughly twice the metal and in the same direction, which is the ordinary configuration this card keeps having to restate after the sessions that are not. Then **15 October** handed it its best session since **19 August**, **+6.53% to $18.51** on 1.48 times average volume and off a **$17.00** low, against a gold trust up **1.79%** and the gold miners up **4.34%**: roughly **1.5 times** the miners, which is this card’s documented amplification of a metals move and not an antimony signal. **16 October extended it, and this time the metal cannot take the credit:** **+3.75% to $19.21** on 1.44 times average volume, while the two spot-tracking gold trusts _fell_, **GLD 0.18%** and **IAU 0.17%**, and the gold miners added only **0.79%**. A documented amplification of a metals move cannot account for a session in which the metal did not move, and nothing on the company’s release feed is dated to it, so this one goes down as its own. Its year to date now reads **−20.7%** and the drawdown **48.6%**, and its twelve-month cell **−31.4%** against a 16 October 2025 base of $28.00, from the +23.6% it carried in late September — a large part of that the base rolling onto progressively higher year-ago days rather than price. Antimony went nowhere throughout: **$51.80/kg** at the 18 September assessment, down 5.85% year to date and about 43% from a year earlier. The commodity that earns this name its place on a critical-minerals list is flat to falling either way.

EXIM board approval **$2.9bn** Gold miners **−16.4% 1m** PPTA **−30.5% YTD** Gold **~450koz/yr** US antimony **~35%** First production **2029**

**20 October: −0.27% to $18.72 on 1.02 times average volume, and the gold check passes this time by sitting still.** The gold trust fell 0.54% and the gold miners 0.43%, so a quarter-point fall is roughly half the metal rather than a multiple of it: no antimony signal, and for once no amplification either. Its twelve-month cell fell five and a half points to **−31.4%** on a year-ago base rolled $25.32 to $27.27, which is the calendar and not the metal. On the open item nothing has closed. The **$2.9bn** EXIM facility remains _board-approved_ from 21 May 2026, with definitive documentation outstanding and the company guiding close to the fourth quarter, which has nine weeks left. Approved is still not drawn.

**23 October: −4.75% to $16.44 on 1.48 times average volume, the lowest close since 19 August 2025 and the largest fall on the page for a second consecutive session.** This card’s rule is to price the metal first, and the metal does not cover it. **December gold futures settled $4,007.90, down 1.29%**; the **iShares Gold Trust closed $75.15, down 0.67%**, both their lowest since **16 July 2026**; and the gold miners fell **1.82%**. Perpetua fell **3.7 times** the futures and **2.6 times** the miners, on the only above-average turnover in tier two that day. Two sessions running, gold has moved in the same direction and nowhere near as far, which is the condition under which this card stops attributing the move to the metal. Nothing is dated to the company. The drawdown widens to **56.0%**. Read the 1m cell with care: it _improved_ to −29.5% on a 4.75% fall, because the four-week base rolled $24.76 to **$23.32**.

**Watch**EXIM financial close and the final investment and construction decision, both targeted for Q1 2027. At **Mining Forum Americas** on **28 September 2026** Perpetua put loan closing in the **fourth quarter of 2026**, narrowing the “second half” it carried at the 17 August quarterly, with a **final investment decision after** the close, an approximately three-year build after full sanction, and **initial production in late 2029**. It also sized the project at **more than $3bn** of construction capital, which was its September conference figure and was above the $1.3bn attached to the groundbreaking coverage in the sources below. Approved is still not drawn: the $2.9bn was board-approved in May 2026 and the documentation is unsigned, so headlines describing the loan as received or secured are running a quarter ahead of the facility. With first metal four years out, financing terms matter more than commodity prices here. Cash at 30 June was $574.2m. The tape has run a long way ahead of any of that: **−10.18% to $17.38** on **14 October**, on **2.11m shares** and **2.24 times** average volume, its **lowest close since 29 July** and the worst session anywhere on this page that day, with nothing filed. The surrounding instruments rule the metal out as the cause and confirm what this card has said from the start: **gold fell 1.49%**, the gold miners **4.40%** and the silver miners **5.45%**, so this is roughly **2.3 times** the miners in a rates-driven liquidation, not an antimony move. Check the gold tape before attributing anything here to antimony. **27 October: +10.66% to $18.37 on 1.05 times average volume, and this one was not gold.** December gold futures closed **$4,074.60, up 0.90%**, so the equity moved roughly **11.8 times** the metal. The standing rule on this row is that it prices off gold rather than antimony, and on small sessions the gap is not worth reading; on a session this size it is. Whatever bid tier two caught on 27 October, this row caught it as a critical-minerals name. It is **6.43%** above its 22 October close and the drawdown comes in **4.8** points to **50.8%**, the largest narrowing on the page. The twelve-month cell is the instructive one and it barely moved, **+2.8** points to **−26.0%** on a 10.66% session, because this is the only Western base to rise this run, $23.30 to **$24.84**. **As of 27 October:** the **$2.9bn** facility was board-_approved_ on 21 May 2026, documentation and closing are still guided to the fourth quarter of 2026, a final investment decision follows the close rather than preceding it, and nothing has been drawn.

**2 November is the cleanest attribution this card has had, because its two legs pointed opposite ways.** **−3.00% to $16.15** on **1.20 times** average volume, 1,393,000 shares against 1,165,475, between $15.97 and $17.49, its **lowest close since 1 August 2025**, taking the drawdown to **56.8%** and the year to date to **−33.3%**. Bullion went the other way: the gold trust **+1.65%**, the gold miners **+2.67%** and silver +2.17%. So this is not the gold leg, nothing was published on antimony and nothing was filed by the company. It is the minerals tier selling on its own account, which is the reading this card usually has to argue for and this session hands it. The **$2.9bn** EXIM facility remains board-approved from 21 May and undrawn, with close still guided to the fourth quarter.

Graphite Not a magnet input

### Novonix

NVX · Nasdaq

**20 November in Sydney.** The ASX line settled at A$0.073 (−2.67%). The US ADS closed on 19 November at $1.90 (−1.04%), 88.2% below its 52-week high, on 1.14 times its prior 20-session average volume.

**19 November in Sydney.** The ASX line settled unchanged at A$0.075. The US ADS closed on 18 November at $1.92 (−0.52%), 88.1% below its 52-week high, on 0.50 times its prior 20-session average volume.

**18 November in Sydney.** The [ASX line](<https://stockanalysis.com/quote/asx/NVX/history/>) settled at A$0.075 (−1.32%). The US ADS closed on 17 November at $1.93 (+0.52%), 88.0% below its 52-week high, on 1.31 times its prior 20-session average volume.

**17 November in Sydney.** The [ASX line](<https://www.eoddata.com/stockquote/ASX/NVX.htm>) closed at A$0.076 (+1.33%). The US ADS closed on 16 November at $1.92 (−8.57%), 88.1% below its 52-week high, on 0.94 times its prior 20-session average volume.

**16 November in Sydney.** The [ASX line](<https://stockanalysis.com/quote/asx/NVX/history/>) closed at A$0.075 (−6.25%). The US ADS closed on 13 November at $2.10 (−1.41%), 87.0% below its 52-week high, on 1.01 times its prior 20-session average volume. US history reflects the 27 August ADS ratio change.

The weakest name on either list, and it deserves a plain warning rather than a bullet point. Novonix makes synthetic graphite for battery anodes in Chattanooga, Tennessee — a battery input, not a magnet input, and not part of any robot's bill of materials in the way the tier-one components are.

The financials are severe, and on **31 August 2026** the auditor said so in the accounts. The half-year report for the six months to 30 June shows a loss of **$24.6m** against $20.1m a year earlier, net operating cash outflows of **$23.1m**, and cash down to **$59.5m** from $79.9m at 31 December. Net current liabilities stand at $2.5m where six months earlier there were net current assets of $5.6m. The independent auditor's review report carries a **material uncertainty that may cast significant doubt on the group's ability to continue as a going concern**. The same report has the company backing away from a land purchase beside its Chattanooga plant and stretching the Riverside ramp-up, which is what conserving cash looks like in a capital project. The genuine positive is still the **US$103m of 48C tax credits** certified in April 2026 for the Riverside project — but those vest only when the first 11,000 tpa of capacity is in service before 7 April 2028, and a slower ramp is the one thing that deadline cannot absorb. Mass production for lead customer Panasonic is not expected before 2027.

The going-concern note names the trigger, which is what makes this more than a slow burn. **$35m** of face value in unsecured convertible debentures issued to Yorkville carries an **A$0.12 floor** on the ASX share price: a fall below it may constitute an amortisation event, under which the company could be required to redeem part or all of the outstanding balance in cash. On 31 August, the day that note was published, the ordinary shares closed at **A$0.11**, down 12.0% on 12.3m shares against a recent average nearer 3.5m, and the Nasdaq line fell **12.3%** the same session. They have not recovered: A$0.11 on 1 September, A$0.105 on 2 and 3 September, A$0.11 on 4 September, A$0.105 on 7 September, then −9.5% to A$0.095 on 8 September — the first close below ten cents — A$0.098 on 9 September, a 7.1% bounce to A$0.105 on 10 September on 10.8m shares against a recent average nearer 3.5m, and A$0.105 again on 11 September — ten consecutive closes below the floor, the lowest of them 20.8% under it. Then on 14 September the shares rose 14.3% to A$0.12 on 7.0m shares, their first close back at the threshold since 28 August, with no ASX announcement attached to it. That was the top. Three sessions at A$0.11 followed, then **A$0.10 on 18 September on 23.2m shares** — by far the heaviest volume of this entire run, against a recent average nearer 3.5m — A$0.099 on both 21 and 22 September, **A$0.097 on 23 September**, a fresh low for the run, and A$0.099 on 24 September. The clause turns on a fall _below_ A$0.12: the shares have now closed below it in **eight consecutive sessions** since touching it, the last of them **17.5%** under it, and the heaviest volume of the run came on the way down. They are also **38%** below the A$0.16 the June placement was struck at. Nothing has been announced. This is the position the going-concern note describes, arrived at. The directors rest the going-concern basis on a history of raising capital, including $14.5m from a June 2026 placement; the share purchase plan that followed it raised A$0.96m.

The US listing is a separate problem, already solved once. On 12 August the company agreed with BNY to change its ADS ratio from one ADS per four ordinary shares to **one per forty** — economically a one-for-ten reverse split of the ADSs, effective **27 August 2026**, done to lift the traded price back over Nasdaq's minimum bid threshold. The underlying ASX ordinary shares are untouched; only the US wrapper changed. On the new ratio the US line fell to $2.86 on 3 September, rebounded 5.2% to $3.01 on 4 September, fell 8.3% to $2.76 after Labor Day, eased 0.7% to $2.74 on 9 September — a two-year closing low — then rose **6.2% to $2.91** on 10 September, the only US line on this page to rise that session and a day behind the Australian bounce it was following, added a further 2.4% to $2.98 on 11 September, and then jumped 8.4% to $3.23 on 14 September — its highest close since 28 August and the largest single gain anywhere on this page that session — before handing most of it back, −5.6% to $3.05 on 15 September. Since then it has followed the Australian line down with a session's lag and no news at either end: $2.97 on 16 September, $3.03 on 17 September, **−4.8% to $2.885** on 18 September, −1.6% to $2.84 on 21 September, a near-flat $2.85 on 22 September, −3.4% to $2.801 on 23 September and then **−5.7% to $2.64** on **24 September** — its worst session since 8 September and the **lowest close in its three-year series**, taking out the $2.74 of 9 September that had been the floor. The 52-week low proper is $2.52, an intraday print rather than a settlement. The Sydney line, a session ahead, closed **A$0.098** on 25 September — a ninth consecutive close below the **A$0.12** Yorkville amortisation threshold, **18%** under it, and **39%** below the A$0.16 the June placement was struck at, with no announcement attached to any of it. It then steadied: +1.5% to $2.68 on 25 September and **+0.4% to $2.69** on **28 September**, the only riser in the whole minerals tier on a day the rest of it was sold off. Then **14 October** put it somewhere it has never been: **−4.70% to $2.23**, the **lowest close in its Nasdaq history**, with no close beneath it in **1,179** sessions back to 2 February 2022. There is no prior level to measure that against, which is the point of recording it. **15 October went lower again, −2.24% to $2.18**, a second consecutive record low, on **0.76 times** average volume. With Lynas it was one of only two names in the minerals tier that did not join that session’s reversal, and the only one to set a low in it. **16 October set a third consecutive record low**, **−1.38% to $2.15** on **0.69 times** average volume, with no close beneath it anywhere in the series. That leaves the US line **91.0%** below its high, **89.0%** down over a year against a 16 October 2025 base of $19.60 and **78.7%** lower year to date, still the worst figures in every column on this page. A ratio change buys compliance, not solvency: the US line has gone from $3.51 to **$2.15** since it took effect, still clear of Nasdaq’s $1 threshold but closer to the range where it matters.

H1 loss **$24.6m** Cash **$33.7m** at 30 Sep Auditor **going-concern doubt** Yorkville floor **A$0.12** Off high **−88.2%**

**20 October: −2.31% to $2.11 on 1.00 times average volume, a fresh record closing low on exactly its own average turnover.** The drawdown cell barely moved, 90.0% to **89.8%**, because the 52-week high rolled again, $21.50 to $20.70, and this run the base roll and the price all but cancelled. It is still the deepest drawdown here by twenty points, and the twelve-month cell fell to **−90.6%** on a base rolled $19.80 to $22.50.

**22 October answered this card’s open question, and the answer is cash out of the door.** Novonix told the ASX, and filed on Form 6-K, that it had made a **second amortisation payment of US$7,000,000** to **YA II PN, Ltd**, an affiliate of Yorkville Advisors Global, redeeming convertible debentures of equal face value and reducing the amount outstanding to **US$21,000,000** against a facility whose maximum aggregate face value is US$100m. The trigger is named in the filing and it is the share price itself: an amortisation event subsists while the daily ASX VWAP sits below the **A$0.12** floor, and while it continues further payments fall due on a **rolling 30-day basis at 20% of the amount outstanding**. The payment came from existing cash reserves. Set that against the half-year accounts, which carried **US$59.5m** of cash at 30 June with a material going-concern uncertainty: two payments inside roughly a month have taken **US$14m** out to repay early a facility the tape is forcing it to retire, and the mechanism is self-reinforcing, because the condition that triggers the payments is the condition the payments do nothing to fix. The stock rose **4.14% to $2.09** on 0.71 times average volume the next session, off the record low set the day before.

**30 October answered the cash question with a filing, and the answer has two halves: a number that looks survivable and a contract that got worse.** The September-quarter Appendix 4C puts cash at **US$33.652m** at 30 September against a net operating outflow of **US$9.765m** for the quarter and US$32.820m for the nine months, which the form’s own item 8.5 turns into **3.4 quarters of funding available**. Read what that item actually divides: total available funding by _operating_ cash flow, and item 8.3 records **nothing unused** in finance facilities, so there is no undrawn line behind it. Add the quarter’s **US$9.169m** of investing outflow, which is the Riverside production build and not discretionary while qualification is the whole business case, and the same arithmetic returns **1.8 quarters**. Then take out the second **US$7m** Yorkville amortisation payment, made in October and therefore after the balance date: the December quarter opens nearer **US$26.7m**. Facilities total **US$58.1m and are fully drawn**, a US$30.1m PNC real estate loan and US$28m of Yorkville debentures at 30 September, the latter since reduced to US$21m by that payment. Customer receipts for the quarter were **US$582,000**, against US$2.703m for the nine months. The quarterly carries no going-concern statement of its own, which is normal for a 4C; the half-year’s material uncertainty stands behind it unamended.

**The contract half is the one that changes this card, and it is a straight downgrade.** The **November 2024 binding offtake** with PowerCo, a minimum **32,000 tonnes over five years from 2027** for the St Thomas, Ontario plant, has been replaced by an **Amended and Restated Supply Framework Agreement**. The company’s own words: the parties have “**removed the previous volume, pricing, and delivery commitments**” and will “negotiate in good faith the terms of a new offtake agreement (including volume and price)” once qualification and the applicable industrial certifications are complete. St Thomas itself has had **production deferred from 2027 to 2029**. Riverside achieved **ISO 9001:2015** in October, which the filing is careful to say “does not itself constitute customer product qualification”; the initial C-sample still has specification parameters outstanding; and mass-production sales remain guided to the **second half of 2027**, which is now two years ahead of the customer the tonnage was written for. A business with under two quarters of funding on the fuller measure has exchanged its only binding volume commitment for an agreement to agree. That is a larger change to the position than any price printed on this card, and it is the kind of change the **US$103m of 48C credits** cannot absorb, because those vest only on the first 11,000 tpa being in service before **7 April 2028**.

**Watch**The cash line, now that the amortisation question is answered. The shares touched the A$0.12 floor exactly on 14 September and have closed below it in every session since, ending at **A$0.094** on 5 October, **22%** under it, with the heaviest volume of the run on the way down. The Nasdaq line has gone further since: **−4.70% to $2.23** on **14 October** is the **lowest close in its entire Nasdaq history**, with no close beneath it in 1,179 sessions back to 2 February 2022, and again on no disclosure from the company. That is no longer an open question about whether the threshold gets tested; what is open is the consequence, and a cash redemption against a **$59.5m** balance falling by roughly $23m a half is a large one. Dilution or a raise was already the base case; the auditor has written the reason into the accounts and the tape has now written the trigger. It is the only name on this list where solvency is the question rather than growth. **27 October: +6.67% to $2.24 on 0.86 times average volume, a third consecutive gain.** Nothing was filed. It is **11.61%** above its 22 October close, the largest three-session recovery anywhere on the page in percentage terms, which is what a two-dollar stock does rather than a signal. The drawdown narrows **0.7** points to **89.2%** and remains the deepest on the page by twenty-one points, with the year to date at **−77.8%** and the twelve-month at **−87.1%**. A sector bid lifts this line with the rest of the tier and changes nothing about the question on it, which is financing and cash burn rather than price. **28 October: +9.82% to $2.46 on 14,400 shares, and the drawdown cell moved further on the calendar than on the price.** That is **0.66 times** a twenty-session average of 21,845, which on this ADR is a handful of trades, and it is the **highest close since 9 October**. Nothing was filed. The cell to read carefully is the drawdown: it improves **2.9** points to **86.3%**, of which the session is worth **1.1** and a **52-week high rolling $20.70 to $17.90**, down 13.5% as the 28 October 2025 session left the window, is the other 1.8. It is the only one of the eleven 52-week highs on this page to move this run and the largest base roll of it. The question on this line is unchanged and is not a price question: financing, dilution and a cash balance falling by roughly $23m a half.

**2 November: the deepest drawdown on the page gets deeper, on the lightest turnover of any name here.** **−4.44% to $2.15** on **1.32 times** average volume, 24,600 shares against 18,570, which is the whole difficulty with reading this line: 1.32 times average is twenty-four thousand shares. Its **lowest close since 26 October**, **86.7%** below the $16.15 high, **78.7%** lower year to date and **85.8%** over twelve months. Nothing was filed. The 30 October Appendix 4C is still the live document and the arithmetic in it is unchanged: **1.8 quarters** of funding once the Riverside investing outflow is counted alongside the operating one, against the form’s own 3.4 quarters on operating cash alone.

### Correlation with the robotics basket

The same 103 calendar-aligned weekly returns, measured against an equal-weighted basket of the six tier-one component names. The ordering still inverts — the minerals names lead on REMX, the robotics names lead on the basket — just not as sharply as before the alignment fix.

Robotics component names Rare earth & critical minerals names

06 Tier three — the magnet layer

**This layer explains the correlation result above.** Robotics OEMs almost never buy magnet blocks directly — magnet procurement sits upstream, with the motor, encoder and actuator manufacturers. That is part of the mechanism behind a 0.36 correlation between the robotics basket and REMX: the magnet is real, and it enters the robot through Nidec's and Yaskawa's purchase orders, priced into their cost of goods, but not through anything an assembler or an ETF holder touches directly.

It also explains why the layer is so hard to own. China controls roughly **90% of midstream NdFeB processing**, and most of the Western and Japanese alternatives are private, subsidiaries, or state-funded projects that have not yet produced at scale. Of the seven magnet names below, three are directly investable — and only TDK combines real scale in magnets with a clean listing.

| Company | What it makes | How you would own it | Last | YTD | 1y | Corr robotics |
| --- | --- | --- | --- | --- | --- | --- |
| TDK6762.T · Tokyo | Largest non-Chinese NdFeB maker; ferrite | Direct — listed in Tokyo | ¥2,116.5 | −4.3% | −15.0% | 0.57 |
| Aichi Steel5482.T · Tokyo | Permanent magnets for robotic actuators | Direct — listed in Tokyo | ¥2,452 | −16.1% | −14.4% | 0.15 |
| Arnold Magneticvia CODI · NYSE | Permanent magnets, magnetic assemblies | Only via Compass Diversified, its parent | $12.21 | +154.4% | +115.3% | 0.18 |
| MP MaterialsMP · NYSE | Sintered NdFeB — see tier two | Direct — Mountain Pass feedstock and Independence magnet production | $36.66 | −27.4% | −42.3% | 0.24 |
| Proterialformerly Hitachi Metals | Advanced magnetic materials, RE-free R&D | Not investable — private since 2022 | — | — | — | — |
| Niron Magneticsprivate · Minnesota | Iron nitride, rare-earth-free magnets | Not investable — VC and DoW funded | — | — | — | — |
| Noveon Magneticsprivate · Texas | Sintered NdFeB from recycled feedstock | Not investable — private, Series B | — | — | — | — |Correlations are weekly returns against the equal-weighted tier-one basket over 103 weekly returns, re-bucketed by ISO calendar week. TDK has the highest correlation in this tier at 0.57. Its listed group includes large businesses outside magnets. Aichi Steel, Arnold and MP sit at 0.15–0.24.

Listed Largest ex-China

### TDK

6762.T

**20 November.** TDK closed at ¥2,116.5 (−6.91%) on 0.87 times its prior 20-session average volume. Yahoo’s daily close is null; its 15:30 JST quote metadata supplies the completed auction close.

**19 November.** TDK closed at ¥2,273.5 (+0.42%) on 0.63 times its prior 20-session average volume.

**18 November.** TDK closed at ¥2,264 (+3.36%) on 0.84 times its prior 20-session average volume.

**17 November.** TDK closed at ¥2,190.5 (−0.50%) on 0.86 times its prior 20-session average volume.

**16 November.** TDK closed at ¥2,201.5 (−4.70%).

**13 November.** TDK closed at ¥2,310 (−1.16%). The company separately confirmed about ¥100bn of planned investment in [TMR sensor development and production in Ojiya](<https://www.tdk.com/en/news_center/press/20261030_01.html>), with operations scheduled for 2029.

The most substantial magnet business you can actually buy. TDK is the **largest non-Chinese neodymium magnet manufacturer by revenue** and the most globally diversified, producing sintered NdFeB in Japan and China alongside ferrite and metal-powder products. The company's origins are literally magnetic — it was founded in 1935 to commercialise ferrite.

Two honest caveats. Magnets are a modest slice of the group: TDK is dominated by its energy business, including ATL, one of the world's largest lithium battery makers, plus passive components and HDD heads. Its **0.57 correlation** to the robotics basket is the highest of any tracked company outside tier one. The comparison includes TDK’s entire electronics and battery business. You would be buying a battery-and-components group with a major magnet business attached.

Price **¥2,116.5** YTD **−4.3%** 1y **−15.0%** Corr robotics **0.57**

**30 October: the first-half results landed at the bell with a 23.7% upgrade to full-year operating profit and the annual dividend lifted from ¥40 to ¥50, after the shares had already fallen 7.22% during the session.** Against the forecast of 28 April, net sales go to **¥3,050,000m** (+18.2%), operating income to **¥365,000m** (+23.7%) and net income attributable to owners to **¥280,000m** (+24.4%), taking earnings per share from ¥118.54 to **¥147.50**. The interim dividend is **¥25.00** against a ¥20.00 forecast and ¥16.00 a year earlier, and the year-end projection rises to ¥25.00. The reason given is AI data-centre demand running well above the prior year, with smartphone-led ICT sluggish but rechargeable batteries and sensors expanding on new models and nearline HDD heads up significantly. **None of it is in the price yet:** all three filings are timestamped 15:30 JST, which is the bell, and the fall was complete by 11:00. **What the release still does not do is size magnets.** The upgrade is an AI data-centre story told through batteries, heads and passive components, which is exactly this card’s standing problem: the magnet business stays real, large and unmeasurable from outside. The same day brought a reported **¥100bn** expansion of magnetic-sensor production in Niigata for physical AI, on the former JS Foundry site at Ojiya acquired in June 2026 for operation in 2029. A magnetic sensor is an encoder and actuator input rather than a magnet, so if that figure is confirmed in a TDK release it argues this name sits nearer tier one than the tier it occupies here. The twelve-month cell is now exactly **flat**, 7.0 points lower on the session, and the year to date falls to **+16.3%**.

**21 October: +2.42% to ¥2,902.5 on 0.74 times average volume, recovering most of the previous session’s fall on three quarters of normal turnover.** It is now the only name anywhere on this page whose four-week cell is positive, at **+2.0%**. The twelve-month cell slipped a little further even so, +19.1% to **+18.8%**, on a year-ago base rolled ¥2,379 to ¥2,443.5. Still no segment disclosure sizing magnets.

**Watch**Any segment disclosure sizing magnets against the battery business. Without it, the magnet exposure is real but unmeasurable — the same problem Aichi Steel has, at much larger scale. The September trading history puts that relationship in context: TDK fell **6.9% on 8 September** — the worst session of any name on this page — in a yen-driven selloff that had nothing to do with magnets, recovered 1.4% and a further 1.4% to ¥2,850.5 by 10 September, gave back 1.5% on 11 September as oil and bond yields took the whole Tokyo market down, added 0.3% on 14 September while the index fell 0.81% on its AI complex, rose a further 0.6% on 15 September and added 1.4% to ¥2,872 on 16 September, before giving all of it back and more on 17 September, **−2.3% to ¥2,806** — one of only two Tokyo names on this page to fall on a rising index — then recovering +1.4% to ¥2,845 on the Bank of Japan's hike, the smallest of the five Tokyo gains that session — and then, on the reopen after the three-day holiday, the largest: **+3.9% to ¥2,954.5** on **24 September**, the biggest single gain anywhere on this page that day, on a session led by the semiconductor-equipment names rather than by anything magnetic. It added **+0.8% to ¥2,978** on **25 September**, its highest close since **7 September** — though the index rose 1.30% that day, so this was a lag rather than a lead — and gave a little back on **28 September**, **−0.8% to ¥2,955**, roughly with an index down 0.73%. It moves with Murata and Taiyo Yuden rather than with the gearbox makers, which is the point: this is a Japanese electronics exporter that happens to make magnets. Note what it did _not_ do on 14 September: the AI-pace selloff that took the US semiconductor complex down more than 5% that session left TDK higher on both sides of it. A passive-components maker is not an AI-silicon name, and the tape agrees even when the sector label does not. It eased again on **29 September**, −1.2% to ¥2,919, against an index down 0.60%, then took it back on the quarter's last day, **+2.1% to ¥2,981** on **30 September**, its highest close since **1 September**, and on an index up 1.94% that is a lag again, which is the shape this name keeps making. Then on **1 October** it stopped lagging and led: **+7.35% to ¥3,200** on 15.6m shares, closing exactly on its session high, more than double a Nikkei that rose 3.30%, its highest close since **17 August** and the largest single gain anywhere on this page this run. For once there is a dated disclosure behind it, and it is instructive about what this name is. On **29 September** TDK and **Taiyo Yuden** signed a **memorandum of understanding** to open discussions on a potential business alliance in **multilayer ceramic capacitors and inductors** for AI data centres, AI servers and physical AI systems, with manufacturing and procurement cooperation in scope, a possible future capital alliance left open, and the two naming MLCCs as Specified Critical Products under Japan's Economic Security Promotion Act. It is an agreement to talk rather than an alliance: no terms, no stakes, no joint venture, and nothing that touches NdFeB. So the single best session this page has recorded for its largest non-Chinese magnet maker came out of a capacitor line, which makes the caveat at the top of this card concrete rather than theoretical. And it has now happened twice on the same disclosure: **5 October** added **+7.8% to ¥3,447** on **22.4m shares** against 11.5m the session before, its **highest close since 6 July**, in a Tokyo session led by semiconductors in which TDK outran Tokyo Electron, Advantest, Murata and SoftBank alike. Then it happened a third time and on a third catalyst, and a fourth session undid part of it. **6 October** added **+5.25% to ¥3,628** on **32.4m shares**, its **highest close since 3 July**, on a _Bloomberg_ report that Seagate and Toshiba are bidding for its **HDD magnetic-head business**: a fourth consecutive gain, and **+21.7%** in four sessions. Then **7 October** gave part of it back, **−2.70% to ¥3,530** on 16.6m shares, after Toshiba told the same outlet the report was “inconsistent” with its understanding and denied that any three-way discussion of head supply had taken place. **8 October** gave back a good deal more, **−5.01% to ¥3,353** on 12.7m shares, closing _exactly on its low_ against a high of ¥3,515 — two sessions now worth **−7.6%** from the ¥3,628 peak, which returns about **42%** of the four-session run and leaves it still **+12.5%** on 30 September. The tape has priced the denial rather than the report, and neither TDK nor Seagate has said anything at all. Four catalysts in a fortnight (HDD capacity, a capacitor memorandum, a reported auction for the head business, and a denial of it), and not one of them touches NdFeB. A third down session followed on **9 October**, **−2.54% to ¥3,268** on 16.1m shares, which returns **56%** of the four-session run and leaves it **+9.6%** on 30 September and **24.3%** below its high. Twelve-month return **+43.2%** and YTD **+47.8%**, both of which have swung ten points or more inside a month on where the lookback base lands. Three sessions since have undone the rest of that run and more: **¥3,133** on 13 October, **¥3,152** on 14 October, then **−7.88% to ¥2,903.5** on **15 October** in the CPI selloff, on **14.52m shares** against a twenty-session average of 12.63m and its **lowest close since 18 September**. **16 October recovered part of it, +1.27% to ¥2,940.5** on 0.90 times average volume, and **19 October eased again, −0.94% to ¥2,913** on 0.76 times. That now leaves the twelve-month return **+29.2%**, the year to date **+31.8%** and the drawdown **32.5%**, and once again not one of the four catalysts above was the reason for any of it. Until a segment disclosure sizes magnets against the battery and passives businesses, that is the whole problem with owning this name for magnets: the exposure is real, unmeasurable, and repeatedly outvoted by the rest of the group. **Two sessions have now made that point harder than any argument.** After Friday’s **−5.61% to ¥2,624.50** on **20.1m shares**, roughly 1.8 times average, **26 October took a further 1.49% to ¥2,585.50** on average volume: the **lowest close since 21 April 2026**, and a fall led by the semiconductor complex rather than by anything magnetic. Both columns then moved further than either session. The four-week cell goes **−7.8% to −13.2%** as Tokyo’s base rolls onto 25 September’s **¥2,978** — 4.1 of the 5.4 points is the base — and the twelve-month cell **+9.0% to +2.3%**, the largest column move anywhere on the page this run, five of those points a base rolled ¥2,409 to **¥2,528.5**. A magnet holding inside a battery and passives group gets priced on the group, and the group is being priced as a chip name. **The next dated test is 30 October**, when the second quarter either sizes magnets separately or does not; it has not yet. **27 October added 2.78% to ¥2,657.50** on 0.91 times average volume, its highest close since 22 October, but it is the one Tokyo line on this page that has not recovered the rout: it sits **4.42%** below its 22 October close while the four tier-one Japanese names are all above theirs. The twelve-month cell barely moved, +2.3% to **+2.7%**, because the year-ago base rose ¥2,528.50 to ¥2,589 and absorbed most of the session. Three days to the second quarter. **28 October: +5.00% to ¥2,790.50 on 0.89 times average volume, the largest move in Tokyo and the first real repair of this line since the rout.** Its **highest close since 21 October**, and it closes **0.36%** above its 22 October close having been 4.42% below it a session earlier, so the magnet layer has now joined the recovery it sat out for four days. The twelve-month cell gains **5.8** points to **+8.5%** and the year to date reads **+26.2%**. Below-average turnover on the biggest move in Tokyo is the same pattern this card has recorded for weeks: this row moves with Murata and the Tokyo components complex, not with magnets. **The segment question had a date on it, the date has passed, and the answer is no.** The first-half FY2027 results were published at **30 October 15:30 JST** together with a **23.7%** upgrade to full-year operating profit and a dividend raised from ¥40 to ¥50, and they size nothing magnetic: the upgrade is attributed to AI data-centre demand, rechargeable batteries, sensors and nearline HDD heads. Magnets remain unsized in every quarter this page has watched, which now includes the best quarter the company has reported. TDK confirmed the **¥100bn** Niigata magnetic-sensor expansion in a [separate 30 October release](<https://www.tdk.com/en/news_center/press/20261030_01.html>). It covers TMR sensors and schedules operations for 2029. **4 November: the tape has now taken the whole of the upgrade and more.** **−2.54% to ¥2,451** on **1.18 times** average volume, after opening ¥2,520 and trading as low as ¥2,318.5, its **lowest close since 15 April 2026** and **12.1%** below the ¥2,790.5 it closed on 28 October, two days before the results. Five sessions on from a 23.7% upgrade to full-year operating profit and a dividend lifted ¥40 to ¥50, the line is lower than it was before either. The drawdown reads **43.2%** and the four-week cell **−28.9%**, the widest on the page, though 5.3 of those points are a base rolled 7.8% higher onto 5 October rather than price. The reading does not change: a battery-and-components group whose magnet business is still unsized, trading on AI capital expenditure rather than on magnets.

Listed Toyota group

### Aichi Steel

5482.T

**20 November.** Aichi Steel closed at ¥2,452 (−2.04%) on 0.59 times its prior 20-session average volume. Yahoo’s daily close is null; its 15:30 JST quote metadata supplies the completed auction close.

**19 November.** Aichi Steel closed at ¥2,503 (−0.08%) on 0.43 times its prior 20-session average volume.

**18 November.** Aichi Steel closed at ¥2,505 (unchanged) on 0.48 times its prior 20-session average volume.

**17 November.** Aichi Steel closed at ¥2,505 (−1.73%) on 0.70 times its prior 20-session average volume.

**16 November.** Aichi Steel closed at ¥2,549 (−0.66%).

**13 November.** Aichi Steel closed at ¥2,566 (−0.43%). Its [6 November announcement](<https://www.aichi-steel.co.jp/ENGLISH/news/mt-assets/docs/20261106_news_.pdf>) says it would show flexible MAGFINE bonded magnet samples at electronica; thin motors and actuators are among the potential applications.

The cleanest direct way to own a magnet maker on this list. A Toyota Group company engineering permanent magnets tuned for robotic actuators, with an explicit design goal of **reducing dependence on constrained rare earths** — the same strategic bet Proterial and Niron are making, but in a listed vehicle.

**It was the steadiest chart in this report until 29 October 2026, and that sentence now needs its date on it.** At the time of that note it was **4.9%** lower over a year and **29.0% below its 52-week high**, having lost the shallowest-drawdown title it held for most of this account: **Compass Diversified** at 18.7% and **Nabtesco** at 28.3% are both tighter now. What follows is the record of the quiet period, and it is worth keeping rather than overwriting, because the argument it makes, that a magnet option inside a diversified steelmaker barely moves on the humanoid narrative, survived two months of evidence and was then broken by a dividend cut rather than by anything robotic. Before that: 15 October left it the one Tokyo name that barely moved, −1.40% to ¥2,823 against a tier falling six to twelve per cent, and 16 October took it **+4.00% to ¥2,936** on 1.42 times average volume, the best of the six Tokyo lines that session. **19 October closed it exactly unchanged at ¥2,936**, the only flat close on the page. That drawdown has taken back the title it briefly lost: **Compass Diversified** has widened to **24.5%** after its own 5.74% fall on 16 October, so Aichi is the tightest on the page again, against 27–51% for the six tier-one names. That is partly a virtue and partly a warning: magnets are a modest slice of a diversified steelmaker, so the humanoid narrative barely moves it in either direction, up or down. It sat out the 4 September rally with a 0.8% fall, sat out Monday's 2.1% index rally with a 0.2% fall, and lost 1.8% on 8 September on a day the component names lost 3–7%. What it had not sat out was the drift: four consecutive falls, ending −0.8% to ¥3,115 on 10 September, on a day Harmonic Drive rose 2.7% and Nidec 5.6%. Then it did the rarest thing on the page — on 11 September it closed **unchanged at ¥3,115**, to the yen, while the Nikkei fell 1.93% and every other Tokyo name here lost ground. On 14 September it added 0.3% to ¥3,125, its first rise since 3 September, on a day the index fell again, added 0.8% on 15 September, 1.3% to ¥3,190 on 16 September and a further **2.7% to ¥3,275** on 17 September — four consecutive gains and its highest close since 3 September — before doing the one genuinely contrary thing on the page: **−1.4% to ¥3,230** on 18 September, the _only_ Tokyo name here to fall on the Bank of Japan's hike, on a session when the index rose 1.38% and the gearbox makers rose 0.8–4.8%. Then it did the contrary thing again from the other side: on the **24 September** reopen it closed **unchanged at ¥3,230**, to the yen, while the index rose 0.76% and every other Tokyo name on this page moved 1.2–3.9%. It repeated the trick on **25 September**, **−0.2% to ¥3,225** against an index up **1.30%** — two consecutive sessions of doing almost nothing while Tokyo moved two points. Its four-week column is now **0.0%**, flat to the yen against a 26 August base of ¥3,225. Steadiness cuts both ways. It sat out the drawdown and it has now sat out three rallies, which is exactly what a magnet line inside a diversified steelmaker should be expected to do — and it is why this name will not tell you anything about humanoid demand in either direction.

Price **¥2,452** 1y **−14.4%** Off high **−30.2%** FY3/27 div **¥142, cut ¥8** Corr robotics **0.15**

**30 October: −6.33% to ¥2,338, a third consecutive fall and the lowest close since 5 August 2025.** It opened ¥2,495, traded as low as **¥2,311** and settled ¥27 above it, on **911,700** shares against a twenty-session average of 246,455, so **3.70 times** normal turnover a day after 6.43 times. Measured from the ¥2,830 of 28 October, the session before the half-year filing, the two days since are **17.4%** lower, which is a larger move than the guidance cut itself described. The drawdown widens to **33.5%**, the year to date to **−20.0%** and the twelve-month cell to **−11.7%**. Nothing further has been published since the 29 October filing, so this is the market finishing its repricing of the abandoned second-half margin rather than new information. For a card whose standing complaint is that the magnet business is invisible, the sequence is instructive: the one time this name produced a disclosure large enough to move it, the disclosure was about steel.

One caution attaches to the figures above, because the largest session this card has recorded was not a market judgement. On **29 September 2026** the shares fell **3.9% to ¥3,080** — the worst of the six Tokyo names that day — and they went **ex-dividend** the same morning, **¥75.00** a share, payable 26 November. On the ¥3,205 previous close the distribution is **2.3 points** of that 3.9%, leaving about **1.6%** of actual repricing, and nothing was disclosed by the company. Both the year-to-date cell and the drawdown widen on the ex-date for the same mechanical reason. It is the dividend version of the base-roll problem this page flags elsewhere: a column can move several points because of a corporate action, and a reader comparing it with yesterday's is comparing two different things. The following session removes the excuse, though, and it is the reason the caution does not end there. **30 September** took it a further **1.3% to ¥3,040**, its **lowest close since 31 July**, on a day the Nikkei rose **1.94%** and every other Tokyo name on this page went with it. Three points of relative ground on a session with no dividend in it and nothing disclosed is the only part of the two-day fall that is a market judgement, and it is the part worth watching. It has kept going, and that is now the story on this card: **−0.7% to ¥3,020** on 1 October, **−1.0% to ¥2,990** on 2 October and **−0.7% to ¥2,969** on **5 October**, its **lowest close since 31 July**. Five consecutive falls, and on the last of them it was the _only_ Tokyo name on this page to fall at all, against an index up 2.40%. One session of that is a dividend; five is a name the market is marking down while it marks the rest of Tokyo up, on nothing the company has disclosed. The sixth broke the run, **+1.4% to ¥3,010** on **6 October**, and the seventh gave all of it back and more: **−2.19% to ¥2,944** on **7 October**, its **lowest close since 11 June 2026** and the worst of the six Tokyo names on this page, on an index down 0.92%. Then the eighth went further than any of them: **−4.28% to ¥2,818** on **8 October**, on **321,200** shares against 163,400, its **lowest close since 20 May 2026**. Eight sessions, ¥3,205 to ¥2,818, **12.1%**, with one up day in the middle of them and not a single disclosure behind any of it. The ninth broke it, **+0.71% to ¥2,838** on **9 October**, on roughly half the previous session’s volume. Still nothing disclosed across any of them. Three sessions since have left it the steadiest name on the page by a wide margin: ¥2,898 on 13 October, ¥2,863 on 14 October, and on **15 October** a fall of just **1.40% to ¥2,823** while the rest of Tokyo lost six to twelve per cent. The twelve-month cell now reads **+13.4%** against a base rolled onto 17 October 2025’s ¥2,590 from 16 October’s ¥2,617, which is the whole of that cell’s move on a session the price did not change at all, and the drawdown is the cleaner read at **16.5%** below the ¥3,515 high, which on this run hands the title back: Compass Diversified has widened to 24.5%.

**21 October: +0.42% to ¥2,879 on 0.75 times average volume, the smallest move in Tokyo on a day the index rose 0.85%.** The year-to-date cell stays negative at **−1.4%**, and the drawdown remains the shallowest anywhere on this page at **18.1%**, which is roughly what a name with no humanoid disclosure and no rare-earth leverage looks like through a selloff in both.

**Watch**Any disclosure sizing the magnet business separately from steel. Without it you are buying a steelmaker with a magnet option attached, and the option is not priced because it is not visible. **26 October: −0.63% to ¥2,830** on 0.84 times average volume, its **lowest close since 15 October**, with the four-week cell at **−12.3%** and the twelve-month at **+8.1%** on a base rolled to 24 October 2025. It remains the shallowest drawdown on the page at **18.8%** below its high, which is the one durable fact about this line: a magnet option inside a Toyota-group steelmaker moves far less than anything else here in either direction. **28 October: −0.81% to ¥2,830 on 0.78 times average volume, the only faller anywhere on the page.** It is also the only Tokyo name still below its 22 October close, by **0.95%**, so on a session when five of the six Tokyo lines rose two to five per cent this row went the other way on four fifths of normal turnover. The drawdown widens **0.7** points to **19.5%** and is still the shallowest on the page by eleven points, the year to date reads **−3.1%** and the twelve-month **+8.6%** — the latter gaining 3.4 points on a falling session, entirely because the year-ago base dropped ¥2,713 to **¥2,606**. That is the whole character of this line in one session: a magnet option inside a Toyota-group steelmaker, moving less than anything else here in either direction, with the magnet business still not sized separately. **29 October ended that character in a single session, and the watch line above is still unanswered.** First-half FY3/27 results landed during the Tokyo session and the shares fell **11.80% to ¥2,496** on **6.43 times** average volume, 1,239,200 shares against 192,780, opening ¥2,822, low **¥2,467**, against a Nikkei up 1.10%. That is the **lowest close since 14 August 2025** and the largest fall in two years of daily history bar 27 February 2025’s 13.22%. The half itself _beat_ the company’s own 31 July forecast on all four lines: revenue **¥162,572m** (+8.5%) against ¥151,900m asked, operating profit **¥7,336m** (−16.5% year on year) against ¥6,000m, pre-tax ¥8,071m against ¥7,000m, net **¥5,299m** against ¥4,000m. What was sold is the guidance and the payout. Full-year revenue was **raised ¥310,000m to ¥330,000m**, while operating profit was **cut ¥17,500m to ¥15,000m**, pre-tax ¥18,500m to ¥16,000m and net ¥11,300m to **¥10,000m**, and the annual **dividend was cut ¥150 to ¥142** after being carried at ¥150 with “no revision” in the Q1 tanshin. The implied second half is **¥7,664m** of operating profit on ¥167,428m of revenue, a **4.58%** margin against the **7.27%** the old guidance implied: more revenue, a third less profit from it, and the second quarter standalone already down **30.2%** year on year at ¥2,972m. **None of it sizes the magnet business.** The disclosure this card has asked for since it was written did not arrive, so what a holder now has is a steelmaker whose margin guidance has been cut, whose dividend has been cut, and whose magnet option remains invisible. The one thing the session does settle is the premise: this line did not move on humanoid demand in either direction, and when it finally moved twelve per cent it was on steel margins and a payout. **4 November made it three consecutive gains, and it is the only Tokyo row on this page to rise.** **+1.32% to ¥2,375** on 1.22 times average volume, on a session that took the index 4.84% lower at its low and four tier-one names down between 1.7 and 5.3%. Three sessions have recovered **¥37** of the **¥515** lost on 29–30 October, about a seventh, which leaves the drawdown at **32.4%** and the twelve-month cell at **−13.5%**, the latter two points worse than last run on a year-ago base rolled 3.3% higher. Nothing was filed, and the magnet option is still invisible in the accounts.

Subsidiary Owned, not listed

### Arnold Magnetic Technologies

via CODI

**18 November financing filing.** **Compass Diversified amended its debt covenants.** Its [18 November 8-K](<https://ir.compassdiversified.com/node/14346/html>) reports supplemental indentures executed on 16 November after receiving the required noteholder consents. They change Lugano’s subsidiary treatment and several debt and payment provisions, and add approximately $10.1m of payment-in-kind principal. The note maturities and coupon rates are unchanged. This is a parent financing development for Arnold exposure, rather than a magnet-order disclosure.

**19 November.** Compass Diversified closed at $12.21 (−3.25%) on 0.70 times its prior 20-session average volume.

**18 November.** Compass Diversified closed at $12.62 (+5.96%) on 1.54 times its prior 20-session average volume.

**17 November.** Compass Diversified closed at $11.91 (+3.03%) on 0.94 times its prior 20-session average volume.

**16 November.** Compass Diversified closed at $11.56 (−0.69%) on 0.82 times its prior 20-session average volume.

**13 November.** The listed parent Compass Diversified closed at $11.64 (−1.36%).

**4 November.** Compass Diversified disclosed Arnold Q3 net sales of $47.7m and adjusted EBITDA of $8.9m, versus $37.7m and $4.1m a year earlier. Arnold’s results cover the entire subsidiary. [Issuer results](<https://ir.compassdiversified.com/news-releases/news-release-details/compass-diversified-reports-third-quarter-2026-financial-results>).

A real magnet business you cannot buy directly. Arnold makes permanent magnets, precision thin metals and magnetic assemblies to the tolerances robotics OEMs need — and in **March 2026 signed a distribution agreement with USA Rare Earth**, linking it straight to tier two. But it has been a subsidiary of **Compass Diversified** since a $130.5m acquisition in 2012, and CODI is the only way in.

That is a poor trade for magnet exposure, and the reason is not subtle. CODI is a diversified holding company whose stock **fell 59% in May 2025** when a probe found pervasive accounting fraud at its Lugano Diamonds unit. It restated fiscal 2022–2024, deconsolidated Lugano after a Chapter 11 filing in November 2025, triggered credit-facility defaults requiring forbearance, and suspended its distributions. The stock is up **108.5%** year to date because it is recovering from that hole, not because of magnets — and it fell 2.7% on 9 September and a further 2.3% to $10.96 on 10 September, its lowest close since 6 August, in a session that had nothing to do with either magnets or diamonds. It has since gone nowhere with unusual precision: +0.5%, +2.7% to $11.31, −1.9%, −0.8%, +0.6%, −0.5%, +1.6% to $11.21 on 21 September, +1.1% to $11.33 on 22 September, +0.2% to $11.35 on 23 September — the only US name on this page besides Ambarella to rise on a day the minerals tier was liquidated — −0.9% to $11.25 on 24 September, unchanged at $11.25 on 25 September, −0.4% to $11.20 on 28 September — one of only two US names on this page to fall less than a point on a day the minerals tier was sold off wholesale — and **−1.4% to $11.04** on **29 September**, its largest fall of the sequence and its lowest close since **18 September**, on a session when the metals complex recovered. That is **thirteen** sessions inside thirty-five cents, eight cents above the 10 September low they started from. Nothing about Arnold changed in either direction across any of them, which is the recurring problem with owning a magnet business this way: the only listed route to it is a holding company whose price is set by its own recovery. **16 October made that point at size and with nothing attached to it.** CODI fell **5.74% to $10.01** on **1.07 times** average volume, its **lowest close since 29 July 2026** and the largest single-session fall anywhere on this page that day, on a session with no company disclosure behind it and nothing about the magnet business in it. That takes the year to date to **+108.5%** and the drawdown to **24.5%**, which costs it the tightest-drawdown title it had taken off Aichi Steel a run earlier.

The 17 August Q2 2026 call gave the earlier read on Arnold. Subsidiary adjusted EBITDA came in at $91.5m (+12.6% year over year), helped by the 1 May sale of the Sterno food-service business, whose proceeds (over $280m) went straight to paying down the senior secured term loan — total debt fell to $1.59bn from $1.89bn at year-end 2025. COO Zach Sawtelle, who succeeds retiring CEO Elias Sabo at the end of 2026, called Arnold's segment "a standout, up nearly 50%," with a strong backlog tied to non-China rare-earth magnet sourcing and progress at a Thailand facility. Management said it is still pursuing further divestitures to cut debt, but did not name Arnold as a candidate — if anything, a segment growing 50% is the one they would want to keep. CODI shares still fell 4.8% on the day.

Parent **CODI** YTD **+154.4%** Arnold Q3 adjusted EBITDA **$8.9m** Corr robotics **0.18**

**20 October: Compass Diversified +3.08% to $10.36 on 0.78 times average volume, the second-largest rise on the page, with nothing on the company’s feed dated to it.** It is the mirror of the session before, which put 2.15 times the volume behind four tenths of a point: this one puts three per cent on below-average turnover. Neither is a disclosure, and this is still the only route to Arnold on this list.

**A magnet offtake the size of this whole layer has been cancelled, and it is worth recording here even though no name on this page is a party to it.** Reuters reported on **22 October**, from SEC filings and people familiar with the matter, that **General Motors terminated** its long-term permanent-magnet supply agreement with **e-VAC Magnetics**, the US unit of Germany’s Vacuumschmelze, on **1 July**. VAC opened arbitration days later seeking damages for “wrongful termination”, the dispute having first surfaced in filings in April 2026; both sides decline to comment because the arbitration is confidential. The plant is the $500m South Carolina facility launched late last year at just under **2,000 tonnes a year**, and VAC’s chief executive said in September 2025 that **90%** of its output was committed to GM, the balance to the US government. **Energy Fuels**, which agreed in June to acquire VAC for **$1.9bn**, says its purchase price may be adjusted depending on the outcome. The reason it belongs on this tier: every Western magnet thesis on this page is ultimately an offtake thesis, and the largest one anybody had signed is now in arbitration. It says nothing about MP’s Department of Defense floor or offtake, which is a different counterparty entirely, and it is not evidence about any company listed here. It is evidence that a signed Western magnet contract is a thinner asset than the tape has been treating it as.

**Watch**Whether CODI divests Arnold. Management's 17 Aug commentary points the other way for now — Arnold was singled out as a growth standout, not flagged among the divestiture candidates — but a holding company still repairing its balance sheet post-Lugano remains a seller in waiting. **27 October: Compass Diversified +5.07% to $10.78 on 0.75 times average volume, its highest close since 13 October.** Nothing was filed. The parent is **4.46%** above its 22 October close, with the year to date at **+124.6%** and the twelve-month at **+44.7%**, and a drawdown of **18.6%** that is the second-shallowest on the page. The reading on this row does not change with the price: a 5% session in a diversified holding company still recovering from the Lugano restatement tells you nothing about magnet demand, and Arnold remains reachable only through it. Whether CODI divests Arnold is still the question, and the Q2 2026 call pointed the other way, naming Arnold as a growth standout with adjusted EBITDA up roughly **50%** year on year while the divestiture commentary named other units. **2 November was the best session on the page, and the answer is due the day after it.** Compass Diversified **+4.66% to $11.67** on **1.21 times** average volume, 752,600 shares against 621,730, matching its **4 September** close to the cent and so its highest since, which puts the row **143.1%** higher year to date, **82.9%** over twelve months and **11.9%** below its own 52-week high. Nothing was filed, and it rose on a session when four of the five tier-two minerals names fell between three and eight per cent, which is this row’s point restated: a diversified holding company’s price is not a magnet price. Third-quarter results are dated **4 November 2026**, and the two things to read in them are Arnold’s adjusted EBITDA against that roughly 50% second-quarter gain, and whether the divestiture programme names it for the first time. **3 November added a second consecutive gain and a filing, and the filing is governance rather than magnets.** **+3.00% to $12.02** on **0.76 times** average volume, 469,000 shares against 616,185, its **highest close since 17 August**, which leaves the row **150.4%** higher year to date, **94.5%** over twelve months and **9.3%** below its own high, the shallowest drawdown on the page. The **8-K of 3 November**, period 1 November, reports board-approved amendments to the trust and operating agreements effective **1 November 2026**: the board quorum rises **from 35% to a majority of directors in office**, the board must interview director nominees, the disclosure demanded of nominees and proposing shareholders is expanded, and Rule 14a-19 compliance is added. For a holding company still carrying the Lugano restatement, a quorum floor moving from roughly a third to a majority is a real tightening, and it is the first governance document this row has had to record. It says nothing about Arnold. **Results land after the close tonight**, call at 5:00pm ET.

Private Rare-earth-free

### Niron Magnetics

private · Minnesota

**10 November.** Niron announced a CES Innovation Awards honour and plans to exhibit its iron nitride magnets on **6–9 January 2027**. The company specifies late 2027 for Sartell production and targets 2029 for a second plant with up to 10,000 tons of annual capacity. It announced no new customer qualification milestone. [Issuer release](<https://nironmagnetics.com/niron-magnetics-named-ces-2027-innovation-awards-honoree-in-vehicle-tech-advanced-mobility-for-rare-earth-free-iron-nitride-magnets/>).

Not investable, but the most strategically interesting name in this section — because it is trying to make the bottleneck irrelevant rather than relieve it. Niron's **iron nitride** permanent magnets, out of University of Minnesota research, use no rare earths at all. If they work at scale, the entire NdPr supply argument underpinning tier two weakens.

It just got a serious endorsement: on **7 August 2026** the Department of War's Office of Strategic Capital issued a conditional commitment for a direct loan of up to **$150m over 20 years** for a **287,000 sq ft** plant in Sartell, Minnesota — intended as the world's first full-scale iron nitride production site, targeted at robots, defence systems and electric motors. The same release names a second $150m loan from the Shakopee Mdewakanton Sioux Community for the same site, so the plant carries up to $300m of committed debt against roughly $50m of venture capital raised to date.

Construction reached a visible milestone on **10 September 2026**: Niron raised first steel at Sartell and, in the same release, disclosed an equity investment from **Honda**, made through Honda Xcelerator Ventures on undisclosed terms. That is the part worth marking. Every dollar behind this company so far has come from venture funds or from government — the Department of War loan, the tribal loan. An automaker putting its own balance sheet into rare-earth-free magnets is a customer-side endorsement, and customers are the constituency whose behaviour would actually reprice tier two.

A customer has now put the magnets through qualification, which is that constituency acting. Niron disclosed on **2 October 2026** that **Moog** is testing and qualifying its iron nitride magnets in **electromechanical actuators** aimed at Army land vehicles and guided munitions, with the two exhibiting together at **AUSA 2026** from 12 to 14 October. The collaboration dates to **December 2025**, so the release marks a stage rather than a new deal, and it carries no magnet grade, no test data and no terms: it is a trade-show announcement and worth no more than that. What it does establish is the application. Actuators are precisely what tier one sells, so a rare-earth-free magnet being qualified into them is the threat to tier two arriving through the product rather than through the mine. Niron says it has been producing and qualifying with customers since 2025 and expects commercial production within **18 months**, which is a claim to hold against the 2027 Sartell date rather than alongside it.

The number that matters for tier two is the capacity. Sartell is scoped at up to **1,500 tonnes a year** of rare-earth-free permanent magnets, operational in 2027 — small against MP's planned ~10,000 tonnes, but Niron says a follow-on **10,000 tonne** US plant breaks ground in 2028, with site selection already under way. If the second plant lands on schedule, the rare-earth-free capacity arriving at the end of this decade is the same order of magnitude as the Western NdFeB capacity being subsidised into existence today.

DoW loan **up to $150m** Announced **7 Aug 2026** Sartell **1,500 t/yr, 2027** First steel **10 Sep 2026** Honda **equity, undisclosed** Rare earths **none**

**Watch**Sartell's commissioning, still pointed at 2027 now that the structure is going up, whether the 10,000-tonne follow-on site is actually named, and whether any other original-equipment maker follows Honda in. Track it as a threat to the tier-two thesis rather than as a position — the first credible rare-earth-free magnet at volume reprices every name in that tier downward.

Private Context only

### Proterial, Noveon and the Chinese incumbents

**Proterial** — formerly Hitachi Metals, a genuine heavyweight in advanced magnetic materials and rare-earth-free development — has not been buyable since October 2022, when a Bain Capital-led consortium completed a roughly $7.5bn take-private and the shares were delisted. It was renamed in January 2023. Any list presenting it as a way to invest in magnets is out of date by nearly four years.

**Noveon Magnetics** is a private Texas manufacturer producing sintered NdFeB through its EcoFlux recycling process — a closed-loop route that sidesteps mining entirely. It has raised a $75m Series B and is operational, which makes it a real competitor to MP and USAR, just not a tradeable one.

On the Chinese side, the incumbents are what make the ~90% midstream share real. Two names circulating in research summaries, Ningbo Zhanhao and Mainrich Magnets, could not be verified as listed entities and should not be treated as tickers. The clearest listed Chinese pure-play in rare-earth permanent magnets is **Ningbo Yunsheng** (600366.SS) — A-share access and Chinese policy risk apply, which is much of the reason the West is building alternatives at all.

Proterial **private since 2022** Noveon **$75m Series B** China midstream **~90%**

**Watch**Consolidation. A Western magnet maker being acquired, floated or spun out is the most likely route by which this layer becomes investable at all — and every name here is a candidate.

07 What would move this list

Order books, not forecasts

Harmonic Drive’s humanoid-specific orders and Ambarella’s robotics design wins are two figures to watch. Yaskawa’s latest guidance and order figures give a separate test: Yaskawa revised FY2/27 on **9 October 2026**, raising revenue to **¥600bn** on a weaker yen and cutting operating profit to **¥57.5bn** after the new ERP system hit first-half production and shipments. What it leaves is a back half that has to deliver **+47.8%** operating profit growth against a first half that delivered −5.5%. The same filing did produce a countable number, and it cuts against the premise of this list: second-quarter orders of **¥185.4bn**, a record, **+38%** year on year, with Motion Control up **56%** and attributed by the company to **data centre and semiconductor** demand rather than to robots. An order book can be countable and still be about a different end market. **Schaeffler answered the same question four days later, and answered it the other way.** Its preliminary nine-month release of **13 October** put humanoid-robotics orders on hand at approximately **€400m**, against €350m at the half year, after **two new Greater China customers** in the quarter took its humanoid customer count to **five**. That is an order book attributed to humanoids specifically rather than to an adjacent end market, and it is the only quantified humanoid order book in the current tier-one filings. Schaeffler then answered it a third time on **3 November**, and the third answer subtracts a number rather than adding one: the validated nine-month statement repeats the humanoid subject without the euro figure, “especially in humanoid robotics, we won new customers and continued to expand our order book during the reporting period”, with no order-book amount, no customer count and no robotics or actuator segment in the divisional table. The [20 October Taicang release](<https://www.schaeffler.com/en/media/press-releases/press-releases-detail.jsp?id=88222851>) repeated the order intake figure before the nine-month statement, and the [10 November release](<https://www.schaeffler.co.in/en/news-media/press_releases/press_releases_detail.jsp?id=88226241>) repeated it again. **Nidec’s exchange deadline** returned a filing rather than a resolution; **Lynas’ September quarter** on 29 October gave a realised price and no Seadrift; and **TDK on 30 October** raised full-year operating profit **23.7%** to ¥365,000m and the annual dividend from ¥40 to ¥50, on AI data-centre demand, with magnets still not sized as a segment. Harmonic Drive’s [19 November supplement](<https://contents.xj-storage.jp/xcontents/AS06879/72ea9203/5839/4596/a735/21a7aa1f8a6b/140120261118582977.pdf>) reports first-half sales of ¥1.2bn and orders of approximately ¥1.5bn in its combined AI and humanoid category. The category is broader than humanoids alone. Its gearhead and actuator introduction targets are March and December 2027, with large-volume production envisaged for 2028–2030. Nabtesco [filed on 13 November](<https://www.nabtesco.com/cms/wp-content/uploads/Summary_of_Consolidated_Financial_Statements_for_FY2026_Q3_e.pdf>). Its slides put Q3 precision-reducer orders up 18% year on year and flat quarter on quarter, with Hamamatsu still ramping. Nabtesco’s [results Q&A](<https://www.nabtesco.com/cms/wp-content/uploads/Results_Briefing_QA_for_FY2026_Q3_e.pdf>) says further capacity expansion at Hamamatsu depends on demand. The next earnings date is **24 November** for Ambarella’s Q3 FY2027 ([issuer event](<https://investor.ambarella.com/node/17436>)). Yaskawa’s third quarter follows in January. Perpetua’s Q3 filing puts the Stibnite financing close and final investment decision in Q1 2027. Its separate [17 November tungsten memorandum](<https://perpetuaresources.com/perpetua-resources-announces-non-binding-mou-with-a-major-american-aerospace-defense-company-to-evaluate-domestic-tungsten-opportunities/>) leaves definitive terms, funding and any future development approvals open. Lynas holds its [AGM](<https://announcements.asx.com.au/asxpdf/20261029/pdf/075xzndxqk7000.pdf>) on 25 November at 10:00 AEDT.

1 January 2027

On **1 January 2027**, [DFARS 225.7018](<https://www.acquisition.gov/dfars/subpart-225.70-authorization-acts-appropriations-acts-and-other-statutory-restrictions-foreign-acquisition>) expands covered US defence procurement restrictions to earlier production stages, including from raw inputs through finished NdFeB magnets. The rule also lists samarium-cobalt magnets, tantalum metals and alloys, tungsten powder and tungsten heavy alloy, with exceptions. This applies to covered defence acquisitions.

Nidec, 16 December

Nidec [received approval on 16 November](<https://www.release.tdnet.info/inbs/140120261116581515.pdf>) to file its half-year securities report by **16 December**. It aims to complete consolidated statements and disclosure materials by early December, followed by the auditor’s interim review. The [separate quarterly results summary](<https://www.release.tdnet.info/inbs/140120261116581518.pdf>) will come more than 45 days after quarter-end, with its date to be announced. Nidec has yet to announce a replacement for the cancelled meeting record date. The exchange’s Special Alert review remains open.

Earlier Nidec filings. The FY3/26 annual securities report was filed on **30 September**, the day its twice-set deadline fell: a net loss of **¥564.6bn** on revenue of ¥2,708.7bn, with impairments above **¥632bn**, so the real figure landed between the **~¥250bn** Nidec had under review and the **¥1tn** the press reported, and neither circulating number was right. But **PwC Japan disclaimed its opinion** on both the accounts and the internal control report, the second consecutive year, so the figures are filed and unassured. The date that now matters is the **end of October**, and it is the anniversary of a designation rather than a discretionary deadline: the Tokyo Stock Exchange made Nidec a **Security on Special Alert** effective **28 October 2025**, and its rules require the written confirmation of internal management system one year on, with delisting if the exchange judges the controls inadequate. New president **Michio Kaida** has said he intends to keep the listing. That makes this the only name on the list carrying a listing risk rather than a valuation question, and the April–June quarter is still missing behind it. **The date was 30 October and it has passed**, and the position into it was that Nidec said it aimed to complete the work needed to obtain audit opinions by the end of October, while its auditor’s chief executive said in public on 1 October that evidence from officers and employees who had made false representations could not be trusted. One thing has now been filed, and it is not the confirmation. On **27 October** Nidec disclosed that it had **decided to file** a damages suit against nine named former officers, six former directors and two former corporate auditors under Article 423(1) of the Companies Act plus a former Vice President who was CFO from May 2022 to March 2026, delegating the action to three of the Investigation Committee’s own attorneys. No suit is filed, no court is named and no amount is set, and the release is nonetheless headlined “Nidec Files Suit”, which the claim audit now carries. Checked after Tokyo’s close on **28 October**, the release list ends there and carries nothing dated today: no audit opinion, no amended securities reports for the restated years, no internal control confirmation. A confirmation letter to the exchange need not be a press release, so silence is not evidence either way. The stock rose **3.56% to ¥1,803** into the date on **1.55 times** average volume, the heaviest relative turnover in Tokyo and a reversal from 0.48 times the session before, which is a market that has started trading the question rather than standing away from it. Reporting attributed to _Nikkei_ has an unqualified opinion and a corrected securities report wanted by month end, and two further strands circulate single-sourced and are recorded as such rather than as fact: plans settled this month to discipline about **20** executives and employees, and an exchange decision on the listing expected at **year-end** rather than on the anniversary itself. The stock closed at its lowest since 31 March 2026 on **9 October** with nothing filed. A second question now sits alongside the deadline: on **7 October** **Oasis Management**, holder of about **8%**, asked the board to begin a preliminary review of privatization options, naming the disclaimer and the delisting risk as its reason. Oasis calls it a feasibility check rather than a decision to sell, no board response is on the record, and it cuts both ways here: a take-private removes the listing risk this trigger is written about and removes the disclosure with it. A third strand arrived on **16 October**, and it is no longer pending: Nidec received the report of its **Executive Responsibility Investigation Committee** on whether current and former directors, auditors and vice presidents are legally liable for the improper accounting, and a second release the same day published the Committee’s conclusion. **Nine individuals are named or identified as “believed to be liable in damages”**: founder **Shigenobu Nagamori**, **Hiroshi Kobe**, **Akira Sato**, **Hiroyuki Yoshimoto**, **Jun Seki** and **Mitsuya Kishida** under Article 423(1) of the Companies Act, **Kazuya Murakami** and **Hiroyuki Ochiai** under the same article, and an unnamed former vice president, the CFO from May 2022 to March 2026, under the Civil Code or Nidec’s Vice President Regulations. The redacted Report itself was published on **20 October**, on schedule, and the Japanese digest of it puts the damage at about **¥27.6bn** — investigation costs, audit costs and the exchange penalty — with the apportionment among the nine left to a court. The same release notes that the **appropriation of surplus was excluded** from the Committee’s remit, because the amended securities reports for past years have not been filed, and says Nidec will examine legal action on that separately once they are. What has _not_ happened is the decision: Nidec says it “will consider necessary legal actions such as filing a claim for damages”, which is deliberation, and an aggregator reporting a board resolution to sue is wrong, as the claim audit records. So the only dated event still ahead is an actual damages claim, and this trigger now cuts two ways against the listing question: a company suing its own founder is evidence of the governance reset the exchange is asking for, and a public account of what the directors did is evidence for the exchange to read. **The anniversary has now passed and the answer came on the day.** Nidec’s **28 October** release “Notice Concerning the Status of Efforts to Resolve the Disclaimer of Opinion” confirms the restated reports were _not_ filed by the 28 October target, because its response to the auditor on identifying and evaluating the officers and employees involved is incomplete and the audit procedures are unfinished. But the confirmation letter is not late: Rule 503(2) requires it “promptly after one (1) year has elapsed”, and Nidec says it **plans to submit the Written Confirmation of Internal Management System on 30 October 2026**, after which the exchange examines it **“with the aim of lifting the Security on Special Alert designation”**. So the date on this trigger is **30 October**, and for the first time the company’s stated direction of travel is lifting rather than delisting. The restatement is a separate and still undated question: Nidec’s own remaining procedures, covering **77 people across about 50 locations** plus 132 further candidates, are expected to finish by **end-November 2026**, and the release states explicitly that this is not the audit completion date nor the submission date. Audit contracts were signed **23 October** for the restated FY3/26 report and **24 October** for FY3/27. Read the two apart: the listing question now turns on the exchange’s examination of a letter filed on time, while the accounts remain unassured with no date attached. **The letter was filed on the day.** At **30 October 15:30 JST** Nidec published “Notice Regarding the Status of the Development and Operation of the Internal Management System”, in Japanese and English, which is the submission Rule 503(2) requires. Nothing on 29 or 30 October restates the accounts or lifts PwC Japan’s disclaimer, so this trigger now waits on the exchange’s examination rather than on the company, and press reporting puts that decision at year-end rather than on the anniversary. A separate **29 October** filing, “Notice Regarding Organizational and Personnel Changes”, is the one to read next, because personnel remediation is the outstanding audit item. Checked again after Tokyo’s close on **2 November**, the release list ends at that 30 October notice: nothing further has been filed, the April–June and July–September quarters are both still missing with no date set for either, and the stock has fallen in both sessions since on turnover of 1.55 and **0.30** times average, which is a market that asked the question and has stopped waiting at the door for the answer.

Price floors spreading to Dy/Tb

Chinese NdPr oxide spent the summer below the $110/kg DoD-backed Western floor and has climbed back to roughly **$108.7/kg** on Benchmark Mineral Intelligence's **25 September** Q3 assessment, which puts it within two dollars of the floor rather than comfortably under it. Ex-China heavies have gone further: dysprosium **+56%** in September to $3,500/kg CIF Europe, terbium +23% to $6,745/kg. The 24 August USA Rare Earth SPV extends guaranteed floors to dysprosium and terbium for the first time — watch whether the next deal does too. Floors are bullish REMX and a cost headwind for the component makers.

The 2027 volume ramp

The volume thesis assumes humanoid shipments accelerate from 2027. Agility Robotics began trading as AGLT on 5 November and targets [Digit 5 early access in the first half of 2027](<https://www.sec.gov/Archives/edgar/data/2074973/000121390026118025/ea0311220-s1_agility.htm>). The [13 November amended 8-K](<https://www.sec.gov/Archives/edgar/data/2074973/000121390026119809/0001213900-26-119809-index.htm>) includes the operating company’s September financials and pro forma combined statements. Agility reported [$360,861 of net sales for the first nine months of 2026](<https://investors.agilityrobotics.com/static-files/a19009fc-8955-4e95-bdd9-aa13bcc9ac5f>), down from $1,739,458 a year earlier. Agility’s [16 November amended S-1](<https://investors.agilityrobotics.com/node/7716/html>) describes about $300m of conditional Digit 5 orders tied to 1,000 robots under a three-year service contract as of May 2026. The [GXO and Apptronik letter of intent](<https://investors.gxo.com/news-releases/news-release-details/gxo-and-apptronik-expand-partnership-scale-humanoid-deployment>) offers a separate milestone for late 2027, conditional on the expanded trials meeting agreed targets. Safety standards are not expected until around 2028, and the projected NdPr deficit widens to 21,000 tonnes by then. Slippage hits the tier-one names first.

Segment disclosure

Schaeffler breaking out actuators, or Ambarella breaking out robotics, would let this watchlist be priced on robotics revenue instead of inference. Neither does today.

Western magnet tonnage

USA Rare Earth disclosed September commercial sales of Stillwater block magnets. Watch for finished prototypes by year-end and 1,200 tpa of combined nameplate capacity with Line 1b in Q2 2027. MP Materials completed GM qualification and expects finished-magnet deliveries and revenue in Q4 2026. Its $18.6m Q3 Magnetics revenue was from precursors. REalloys’ [quarterly release](<https://realloys.com/realloys-press-release/realloys-reports-third-quarter-2026-results/>) gives a design target of approximately 3,000 tonnes per year of US sintered NdFeB magnets by 2030. Watch for site selection and financing. The [10-Q](<https://www.sec.gov/Archives/edgar/data/1567900/000162828026073547/aloy-20260930.htm>) says REalloys has stopped pursuing the potential JS Link transaction.

Greenland, from headline to permit

The **18 September** US–Denmark–Greenland security agreement repriced the rare-earth tape in one session without adding a tonne of capacity. Greenland has no scaled rare-earth mine, and the named projects need permits, infrastructure and local consent. Watch whether a permit or an offtake actually follows; until one does, this is a policy premium on the whole tier, and the names here with no Greenland exposure at all caught most of it.

Iron nitride at scale

Niron's Sartell plant, backed by a conditional $150m Department of War loan and a matching $150m tribal loan, raised first steel on 10 September and targets 1,500 t/yr from 2027, with a 10,000 t/yr follow-on breaking ground in 2028 — though as of **22 October** that follow-on still has no site. Niron is running a search with Savills for a **$1.8bn**, 1.6-million-square-foot plant employing more than **700**, and says it will name a state by the end of this year; its chief executive says Minnesota is “still on our list” but is “struggling to be competitive with some of the other states”. Honda has now taken equity alongside the state money, and **Moog** has been qualifying the magnets in electromechanical actuators for Army land vehicles and guided munitions since December 2025, restated on 2 October 2026: the first named customer doing so in tier one’s own product. A working rare-earth-free magnet at volume does not help tier one much, it undercuts tier two. The company targets late 2027 for first Sartell production and 2029 for production at the second plant. Its next announced public exhibit is [CES, 6–9 January 2027](<https://nironmagnetics.com/niron-magnetics-named-ces-2027-innovation-awards-honoree-in-vehicle-tech-advanced-mobility-for-rare-earth-free-iron-nitride-magnets/>).

10 January 2027, China’s extended pause

MOFCOM and China’s customs agency published **Announcement No. 58 of 2026** on 9 November. The extension took effect on **11 November** and continues the pause on six October 2025 notices through **10 January 2027**. The April 2025 rare-earth licensing regime remains in force. The separate U.S.-specific suspension under 2025 Announcement No. 72 has its own 27 November date. [Joint announcement](<https://www.mofcom.gov.cn/zwgk/zcfb/art/2026/art_91466027f465467985942a639a9485f4.html>).

08 Method & caveats

**Prices** are completed local-session closes from Yahoo Finance chart history and the venue sources below. **Tokyo and Sydney use 20 Nov 2026. Xetra, New York and US OTC use 19 Nov 2026.** Tokyo closed at 15:30 JST, Sydney used post-auction closes, Xetra used its closing auction after 17:30 CET, and the US regular session closed at 16:00 EST. Daily change uses the preceding completed session. The 1m base is the last completed close on or before 30 calendar days earlier: 21 October for Tokyo and Sydney and 20 October for Xetra and the US. The 1y bases are 20 November 2025 for Tokyo and Sydney and 19 November 2025 for Xetra and the US. YTD uses the final 2025 session, 30 December in Tokyo and Xetra and 31 December in Sydney and the US. Drawdown uses the vendor’s rolling 52-week intraday high. Volume is measured against the preceding 20 sessions, excluding the current one. Local-currency closes are split-adjusted and returns exclude cash dividends. SOXX history accounts for its 3-for-1 split effective 5 November.

**20 November feed check.** The Tokyo daily OHLC closes are null, with volume present. Quote metadata stamped at the completed 15:30 JST auction supplies the closes; S&P Global’s dated histories corroborate [Harmonic Drive](<https://stockanalysis.com/quote/tyo/6324/history/>), [Nabtesco](<https://stockanalysis.com/quote/tyo/6268/history/>), [Yaskawa](<https://stockanalysis.com/quote/tyo/6506/history/>), [Nidec](<https://stockanalysis.com/quote/tyo/6594/history/>) and [TDK](<https://stockanalysis.com/quote/tyo/6762/history/>). The remaining Tokyo benchmark and Aichi closes use completed-session vendor metadata, pending daily backfill. Sydney’s opening rows are also null; 16:10:06 AEDT auction rows give [Lynas A$12.44](<https://stockanalysis.com/quote/asx/LYC/history/>) and [Novonix A$0.073](<https://stockanalysis.com/quote/asx/NVX/history/>), corroborated by the linked venue quotes. NJDCY’s stale quote metadata is excluded: its 19 November daily close is $2.62, corroborated by [S&P Global’s OTC history](<https://stockanalysis.com/quote/otc/NJDCY/history/>). SFFLY’s 19 November daily close is $5.715 on 1,466 shares. No tracked close is carried forward. Recheck the Tokyo gaps next run.

**Xetra gap log, 20 November.** The 18 November daily bar backfilled at €5.27 on 676,035 shares, matching the reconstruction and its volume. The 19 November bar is present at €5.19 on 714,574 shares. The log stands at 47 holes, 47 reconstructions and 47 confirmed daily backfills, with no German gap outstanding. Frankfurt separately backfilled 18 November at €5.26 on 14,200 shares and has 19 November at €5.39 on 785 shares. The forming 20 November German and US bars are excluded.

**19 November feed check.** Tokyo and Sydney have completed their sessions. Yahoo’s Sydney opening-stamped closes were null; the post-auction rows at 16:10:02 AEDT give [Lynas A$12.79](<https://stockanalysis.com/quote/asx/LYC/history/>) and [Novonix A$0.075](<https://stockanalysis.com/quote/asx/NVX/history/>), corroborated by S&P Global’s post-auction quotes. Nidec’s NJDCY quote metadata remains stale, but its 18 November daily history gives $2.38, corroborated by [Schwab’s 16:00 ET dated record](<https://www.schwab.wallst.com/research/Public/Stocks/Summary?symbol=NJDCY>). SFFLY’s 17 November daily close has backfilled at $5.80; its 18 November daily close and 16:00 EST metadata agree at $5.63. All tracked rows have verified completed closes, with no stale carry-forward.

**Xetra gap log, 19 November.** The 17 November daily bar backfilled at €5.31, matching the published reconstruction. Yahoo now reports 767,268 shares, 3,030 above the earlier Yahoo and S&P figure. The 18 November daily OHLC and volume remain null. Yahoo’s quote stamped 17:35:16 CET, the 17:30 five-minute bar and [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>) agree at €5.27. The five-minute bar has zero volume and repeats the 17:25 price, so the venue history supplies the independent auction confirmation and 676,035 shares. The log stands at 47 holes, 47 reconstructions and 46 confirmed daily backfills, with 18 November pending. Frankfurt’s separate 17 November bar backfilled at €5.36; its 18 November close is €5.26 in the reopened venue’s previousClose, with its daily bar pending. The forming 19 November German bars are excluded.

**18 November feed check.** Yahoo’s Sydney opening-stamped rows had null closes. The post-auction rows at 16:18:40 AEDT for [Lynas](<https://stockanalysis.com/quote/asx/LYC/history/>) and 16:10:15 AEDT for [Novonix](<https://stockanalysis.com/quote/asx/NVX/history/>) carried A$12.37 and A$0.075, respectively, in agreement with S&P Global. On 17 November in US OTC, Yahoo daily history gives NJDCY at $2.52, corroborated by the final recorded [Schwab trade at 15:28 EST](<https://www.schwab.wallst.com/research/Public/Stocks/Summary?symbol=NJDCY>). Yahoo’s quote metadata for NJDCY remains stale. SFFLY’s Yahoo daily close is null. [EODData’s end-of-day history](<https://www.eoddata.com/stockquote/OTCBB/SFFLY.htm>) confirms $5.80, matching Yahoo’s 15:29:25 EST print. The feeds differ on SFFLY volume, so no relative volume is quoted for it. All tracked price rows have verified closes.

**Xetra gap log, 18 November.** Yahoo’s 17 November daily close is null. Its 17:35:01 CET quote, the 17:30 five-minute auction bar and [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>) agree at €5.31. The 17:25 continuous bar closed at €5.32. The long-range request and S&P Global both show 764,238 shares. The log stands at 46 holes, 46 reconstructions and 45 confirmed daily backfills, with 17 November pending. [Frankfurt venue history](<https://www.finanzen.net/historische-kurse/schaeffler?exchange=FSE>) separately shows €5.36, matching Yahoo’s reopened Frankfurt previous close while that venue’s daily bar is also pending.

**17 November feed check.** Yahoo’s opening-stamped Sydney rows were null. Its rows stamped 16:19:50 AEDT for [Lynas](<https://stockanalysis.com/quote/asx/LYC/statistics/>) and 16:10:06 AEDT for [Novonix](<https://www.eoddata.com/stockquote/ASX/NVX.htm>) carried the post-auction closes of A$12.03 and A$0.076, corroborated by the linked vendors. Yahoo’s NJDCY quote metadata is stale. Its daily close of $2.30 is corroborated by [S&P Global’s OTC history](<https://stockanalysis.com/quote/otc/NJDCY/history/>). No tracked line required a stale carry-forward.

**Xetra gap log, 17 November.** Yahoo initially returned null daily OHLC and volume for 16 November. The [17:35:08 CET quote](<https://de.marketscreener.com/kurs/aktie/SCHAEFFLER-AG-176430175/?iSubCodezb=190544066>), Yahoo’s reopened Xetra previous close, and [Investing.com’s Xetra history](<https://www.investing.com/equities/schaeffler-historical-data>) agree at €5.47. The 17:25 continuous bar closed at €5.48 and Yahoo’s five-minute feed has no auction bar for the session. Yahoo has since backfilled the daily bar at €5.47 with 748,222 shares, which is 90 above [S&P Global’s 748,132](<https://stockanalysis.com/quote/etr/SHA0/history/>). Volume comparisons use the Yahoo backfill. The log stands at 45 holes, 45 reconstructions and 45 confirmed daily backfills. [Frankfurt](<https://stockanalysis.com/quote/fra/SHA0/history/>) separately backfilled at €5.46 with 12,546 shares, matching S&P Global.

**16 November feed check.** Xetra was open at collection. Its null 16 November bar was excluded as a forming session, and its 13 November backfill remains verified at €5.54. Yahoo’s Sydney opening-stamped rows were null, but its post-auction rows at 16:11:22 AEDT for Lynas and 16:10:14 AEDT for Novonix carried the closes. S&P Global’s ASX history corroborated both moves. No tracked close was carried forward from an earlier completed session.

**Xetra gap log, 15 November.** The 13 November [Yahoo daily bar](<https://finance.yahoo.com/quote/SHA0.DE/history/>) backfilled at €5.54 with 622,291 shares. Its OHLC matches [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>) to the cent. The two feeds still differ by 1,000 shares. The post-auction quote and the 17:30 bar also confirm €5.54. Frankfurt separately backfilled at €5.58. Xetra totals now stand at 44 recorded holes, 44 reconstructions and 44 confirmed daily backfills.

**Xetra gap log, 14 November.** The 12 November daily bar backfilled at €5.58 with 544,322 shares, 600 above the original Yahoo figure. The 13 November daily OHLC is null. The Yahoo quote stamped 17:35:05 CET and the 17:30 five-minute bar agree with [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>) at €5.54. The 17:25 continuous print was €5.57. Yahoo reports 622,291 shares and S&P Global reports 621,291. Volume comparisons use Yahoo for continuity. The tally is 44 holes and 44 reconstructions, with 43 confirmed daily backfills and 13 November awaiting a Yahoo OHLC backfill. [Frankfurt’s separate close](<https://stockanalysis.com/quote/fra/SHA0/history/>) was €5.58.

**Xetra gap log, 13 November.** The 11 November daily bar backfilled at €5.63. Yahoo now reports 1,022,893 shares, 652 above the original figure. The 12 November daily OHLC is null with 543,722 shares reported. The 17:35:16 CET quote and the 17:30 five-minute bar agree with [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>) at €5.58, below the 17:25 continuous print of €5.61. The tally is 43 holes and 43 reconstructions, with 42 confirmed daily backfills and 12 November awaiting a Yahoo backfill. [Frankfurt’s separate close](<https://stockanalysis.com/quote/fra/SHA0/history/>) was €5.55.

**Xetra gap log, 12 November.** The 10 November bar backfilled at €5.51. Yahoo lists 575,734 shares for that backfill, 200 above the S&P Global figure used on 11 November. Yahoo’s 11 November daily OHLC is null; volume is 1,022,241. The quote stamped 17:35:17 CET and the 17:30 five-minute bar agree with [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>) at €5.63 and differ from the 17:25 print at €5.62. The tally is 42 holes and 42 reconstructions, with 41 confirmed daily backfills and 11 November awaiting a Yahoo daily backfill.

**Xetra gap log, 11 November.** The 9 November daily close backfilled at €5.56. The 10 November daily OHLC and volume were null in Yahoo’s history. The 17:35:25 CET quote and the 17:30 five-minute bar show €5.51, against €5.53 for the 17:25 continuous print. [S&P Global’s Xetra history](<https://stockanalysis.com/quote/etr/SHA0/history/>) confirms €5.51 and supplies volume of 575,534 shares. The tally is 41 holes and 41 reconstructions, with 40 confirmed daily backfills and 10 November awaiting a Yahoo daily backfill.

**Xetra gap log, 10 November.** The 3 November daily close returned at €5.66, matching the reconstruction. The intervening 4–6 November daily bars are present. On 9 November, the daily OHLC is null with 714,528 shares reported. Yahoo quote metadata stamped 17:36:55 CET and S&P Global’s Xetra history independently give €5.56. The 17:30 five-minute bar agrees on zero volume. The 17:25 continuous print was €5.58. The Frankfurt daily series also has a null bar for 9 November. The logged tally is now 40 holes and 40 reconstructions, with 39 backfills confirmed to the cent and 9 November still awaiting a daily backfill.

**4 November feed log and earlier comparisons.** **At that point Xetra’s gap had opened on four of the previous five sessions, and every closed case still agrees to the cent.** The **2 November** bar, published here last run at **€5.97**, has backfilled at exactly that figure, confirming the thirty-eighth reconstruction. The **3 November** bar then returned a null close with **1,564,612** shares intact and a quote-level range of €6.04 to €5.62, and the **Frankfurt** line carries the identical hole, which places the failure upstream rather than in the Xetra ticker and removes the second German daily route in the same stroke. It is closed on two routes at **€5.66**, regularMarketPrice stamped 17:37:33 CET and a five-minute bar stamped exactly 17:30:00, the last continuous tick at 17:25 carrying €5.63 instead, so the tally reads **thirty-nine holes, thirty-nine reconstructions, thirty-eight confirmed to the cent** with 3 November the only one open. The close published here, **€5.97**, stands on three routes: meta.regularMarketPrice stamped **17:35:06 CET** against the 17:30 auction; a five-minute bar stamped exactly **17:30:00** at the same €5.97; and the hourly series ending on it. Two weaknesses are recorded rather than smoothed. That 17:30 bar shows _zero_ volume, so it is the continuous close rather than the auction print; and unlike the last three reconstructions the **17:25** tick does _not_ agree, carrying **€5.98**, a cent above. Both candidates sit inside the bar’s own €5.75 to €6.04 range and a cent moves no cell on this page, but the series has now produced its first route disagreement, and the figure rests on the auction stamp and the hourly close rather than on a tick. That makes it the **thirty-eighth** hole: thirty-seven reconstructed and confirmed to the cent, one open and awaiting its backfill. The pattern holds on a sample of thirty-eight. Every one has been a German line and every one that has backfilled has backfilled at the reconstructed figure. A missing bar in this feed has never once meant an unchanged price. **The clock change is still the live trap, and it keeps producing a wrong figure in the cross-check.** Central Europe left summer time on **Sunday 25 October 2026**, so the feed reports gmtoffset as **3600** rather than 7200, and it applies that single current offset to every timestamp in the array including pre-transition ones. On daily bars this is harmless, because a Xetra daily bar is stamped at the 09:00 open and an hour either way cannot move it across midnight. On _weekly_ bars it is not: those are stamped Monday 00:00 exchange-local, which in CEST is **Sunday 22:00 UTC**, so converting with 3600 lands on Sunday 23:00 and buckets the bar into the previous ISO week. Every Xetra weekly bar from the summer half of the two-year correlation sample sat one week early on that conversion while the winter bars were correct, which on the current window returns **0.42** for Schaeffler against the basket where the published daily-close route gives **0.61**. Converting each timestamp with the offset in force on _its own_ date removes the divergence entirely: both methods now agree to four decimal places on all thirteen names and on the basket. Xetra’s session end moves with the clock too, from 17:30 CEST to **17:30 CET**, so the absolute UTC cutoff shifts from 15:30 to 16:30. The rule stands and has now earned itself twice: convert a stored timestamp with the offset in force on its own date, never with the one the feed reports today, and read every venue’s close against the local clock rather than a fixed UTC hour. **Tokyo settles at 15:30 JST, not 15:00.** The exchange extended its afternoon session by thirty minutes on **5 November 2024**, so continuous trading stops at 15:25 and an itayose closing auction fixes the close at 15:30; every Tokyo figure here is taken at that auction, and a quote read at 15:00 is a pre-auction print rather than a close. The claim audit carries the consequence, including one instance of this page stating the old time itself. **The 1d cells are not comparable this run, and the weekend is why.** Tokyo’s four tier-one cells and its two tier-three cells span **30 October to 2 November**, three calendar days with a weekend inside them; the Western ones span **29 to 30 October**, a clean overnight. So the Western half of the column holds Friday’s session itself, the largest on this page in a month, while the Tokyo half holds Monday’s response to it. Both are correct and they measure different windows, which is the one thing this column should never be read as doing silently. **Both four-week bases rolled this run, and they split by venue as they usually do.** Tokyo rolled off 30 September onto **2 October**; the Western venues off 29 September onto **30 September**. One roll does most of the work inside the column. **Nidec’s 1m cell goes −23.5% to −16.8%**, **6.7 points** better on a session that _fell_ 2.17%, because its base dropped **¥2,355 to ¥2,117**: 1 October 2026 was the 6.66-times-volume rout, and it has now rolled into the denominator. The base is worth 8.4 of those points and the price takes 1.7 back. Three bases went the other way and cost their cells real points: **Yaskawa ¥4,633 to ¥4,806** (3.1 points), **Harmonic Drive ¥6,420 to ¥6,570** (1.9) and **Ambarella $68.73 to $69.51** (1.0). Nabtesco (¥4,919 to ¥4,926) and Schaeffler (€6.40 to €6.39) moved a tenth of a point each, and REMX’s own base eased **$64.79 to $64.26** for 0.6 of improvement against 4.2 points of price. Net across the six tier-one names the calendar improved the basket cell by **0.4** and the tape worsened it by **1.5**. Read the 1m column against its own date. One column does not follow the completed-close rule at all, and it is worth restating: fiftyTwoWeekHigh is anchored on the **request date** rather than on the venue’s last completed close, so the drawdown lookback can roll on a day the price behind it did not. **This run exactly one of the eleven highs moved**, and it is Novonix for a third consecutive run: **$16.40 to $16.15**, down 1.5% as the request date rolled two days across a weekend. Every other change in the column is therefore price, and ten of the eleven widened. The minerals tier took the four largest: **Lynas 4.5 points to 47.0%**, **Perpetua 3.7 to 55.4%**, **USA Rare Earth 2.8 to 70.7%** and MP 2.7 to **48.5%**. In tier one, **Schaeffler widens 1.8 to 51.5%** and still holds the tier record, Ambarella 1.5 to **34.7%**, Nidec 1.3 to **40.1%**, Yaskawa 0.9 to **50.5%** and Nabtesco 0.4 to 29.9%. **Harmonic Drive is the only one of the eleven to come in**, 0.5 points to **43.1%**, because it is the only one of the eleven that rose. Novonix holds the page record at **86.1%**, fifteen points clear of USA Rare Earth, and of its 0.7-point widening 0.9 is price against 0.2 of base pulling the other way. **The one-year column again moved further than any price, and this run the Tokyo base stepped three calendar days because of a holiday that has not happened yet.** Tokyo rolled from 28 onto **31 October 2025**, since 1 and 2 November 2025 were a weekend and 3 November is Culture Day; the Western ones stepped one session, from 29 onto **30 October 2025**. **The largest move in the column is not on the robotics tier at all.** CODI’s 1y goes **+65.6% to +75.3%**, a **9.7-point** gain on a session that _fell_ 1.15%, because its base dropped **$6.81 to $6.36**: **11.6** points of that cell is the calendar and **−1.9** is the tape. **Harmonic Drive gains 3.2 points to +82.5%** on a base cut **¥2,992 to ¥2,852**, worth 8.5 points against 5.4 taken back by the week’s price. **Nidec’s improves 1.1 points to −6.4%** on the same mechanism, a base cut **¥2,070.5 to ¥1,883** as the post-designation slide of late October 2025 rolls into the denominator: 8.5 points of base against 7.4 of price. Yaskawa loses 2.7 to −7.7% despite a base cut ¥4,354 to ¥4,245, Nabtesco 0.6 to +10.6%, and **TDK crosses zero, +0.0% to −5.9%**, of which 3.7 points is a base that _rose_ ¥2,572.5 to ¥2,673. Aichi Steel is the one cell where price and calendar fully offset, 0.1 point of net change on 1.7 of price against 1.8 of base. **Perpetua is the largest deterioration anywhere in the column**: −23.2% to **−33.9%**, **10.7 points** worse, of which 6.0 is Friday’s 7.81% session and **4.8** is a base that rose **$23.51 to $25.20**. Lynas loses 3.4 to −15.1% on a base cut $10.25 to $9.82, MP 3.8 to −39.0% on a base that rose $64.29 to $64.92, USA Rare Earth 3.4 to −51.1%, Schaeffler 2.9 to −17.5% and Novonix 0.4 to −85.7%. The rule is unchanged: read the 1y column with the base in hand, and never read a flat cell as a flat week. **Both documented feed traps appeared in the same ticker on consecutive runs, pulling opposite ways, which is the clearest demonstration of why the rule is split in two.** Last run the German daily array carried a **12 October** bar at **€5.97** stamped 09:04:22 CEST on 57,043 shares, an intraday quote masquerading as a close, and it was discarded because the venue was open. This run that same session has become a _hole_: the 12 October daily bar is **null**, and the close had to be recovered rather than read. Two independent routes agree on **€5.95**: regularMarketPrice carrying it stamped after the 17:30 auction, captured in the narrow window before that field rolled forward to a live 13 October print of €5.93 on 5,823 shares, and previousClose returning the same **€5.95** on a five-minute request made after Xetra reopened. The continuous five-minute series is the trap inside the trap: it stops at **17:25** on **€5.89**, and taking that last tick would have published a **flat** session rather than the **+1.19%** one that happened. So the rule holds in both halves. **If the venue is open, the last bar is discarded regardless of what it says**; and when a bar is missing, prefer the auction print over the last continuous tick. That was the **twenty-fifth** hole in this series. The gap has recurred on both sessions since, at **13** and **14 October**, while 12 and 13 October have each backfilled at exactly the figure published here, **€5.95** and **€5.85**. The **14 October** bar backfilled at exactly **€5.94** and the **15 October** bar has now backfilled at exactly **€5.75**, both the figures published here from three routes, so the reconstruction that stood open last run is confirmed. **16 October** is null in its place, and this one is a genuine hole rather than the ordinary open-venue case, because Xetra had been closed for thirteen hours when this page compiled. **16 October has since backfilled at exactly €5.83**, the figure published here from two routes, and a thirtieth hole opened in its place on **20 October**, closed here at **€5.89** on two routes: regularMarketPrice stamped **17:35:05 CEST** against the 17:30 close, and previousClose returning the same €5.89 on a request made after Xetra had reopened. **The tally itself was one short when it was published, and the correction is owed before the next figure is added.** Last run this sentence read “thirty holes… with 20 October the only one open”, while the venue note a few lines above it reported the **21 October** bar null as well and gave the close it was recovered at, €5.82. That gap was the **thirty-first** and was never counted here; the claim audit now carries the miscount as this page’s own. Both of those reconstructions have since been confirmed by the vendor to the cent, **20 October at €5.89** and **21 October at €5.82**, so nothing published ever depended on the difference. A **thirty-second** hole opened immediately behind them on **22 October**, and it is among the best-closed cases in the series: three routes all carry **€5.55**, and this time the auction bar and the last continuous tick carry the same figure, so nothing hung on choosing between them. Three more opened at **23**, **26** and **27 October**, and all three have backfilled at exactly the figures published here, €5.39, €5.69 and €5.82, which closes the thirty-fifth case. A **thirty-sixth** opened immediately behind them at **28 October**, closed on three routes at **€5.93** and carrying the identical hole on the Frankfurt line. So the standing tally is **thirty-six holes, thirty-six reconstructions, thirty-five confirmed to the cent**, with **28 October** the only one open. This one is a deeper hole than its predecessors. The 16 October bar had made the shape of these gaps unusually clear, its _volume_ field populated at **877,225** shares while its close was null, so the session was in the feed and only the price was missing; the 20 October bar drops the volume too, which is a different and worse failure, and the twenty-session comparison had to be made against the quote-level volume field instead. The trap inside the trap also pulled the opposite way from last time: the continuous five-minute series stops at 17:25 on **€5.94**, so taking the last tick rather than the auction would have published a +2.95% session instead of the **+2.08%** one that happened. That session also produced this series’ first corroboration from outside the vendor altogether: a German market report stamped 15 October 13:20 CEST quotes the preceding “Xetra close of EUR 5.94 on October 14, 2026”, matching the reconstruction rather than the feed that lost it. One route was weaker than last run’s, and it should be recorded as such: the 17:30 five-minute bar carried €5.75 on _zero_ volume, so it is the continuous close rather than the auction print, where 14 October had shown the auction itself on 1,221,280 shares. **The biggest base roll this series has recorded fell on the 13 October 2026 run, when the largest moves anywhere on the page were in a column rather than in a price.** Every twelve-month base stepped from 9 onto **10 October 2025**, and that single day did something none of the others had: it split the page. 9 October 2025 is the session **MOFCOM published Announcements 61 and 62**, the extraterritorial rare-earth export controls; 10 October 2025 is the day the market traded them, and the Western producers re-rated on scarcity while the broad complex sold off on the tariff threat that followed. So the bases moved in _opposite directions_ and the twelve-month cells followed them. **Novonix**’s base rose **35.8%** in one step, from $21.65 to **$29.40**, which is the largest single base movement this page has seen in any column, leaving its cell at **−91.8%**. **Compass Diversified**’s rose **13.6%**, from $7.35 to $8.35, and its cell fell **18.5 points**, from +47.6% to **+29.1%**, on a **−0.65%** session. **MP**’s rose **8.4%**, from $72.29 to $78.34, for an **8.5-point** fall to **−44.7%**, and **USA Rare Earth**’s **5.0%** for a 4.2-point fall to **−63.8%**. **Ambarella** went the other way and furthest: its base _fell_ **13.5%**, from $84.28 to **$72.88**, so a 3.89% loss left its cell **8.9 points better**, from −20.0% to **−11.1%**. The instructive one is **REMX**, whose own base fell **7.1%**, from $74.78 to $69.47: on the session that priced China restricting rare-earth exports, the rare-earth ETF fell while three of the five producers it is meant to proxy rose, which is this page’s third finding stated by a single day of data. **The four-week bases did not move at all on that run**, all seven sitting on **11 September** for a fourth consecutive run, because a 30-day lookback from 12 and from 13 October both landed back on the same Friday close. That alignment has since broken in the ordinary way, and the bases now sit where the two calendars put them: **18 September** for the four Tokyo lines, **22 September** for the Western ones. **That run of still bases has now ended, and exactly one of the eleven 52-week highs moved.** **Novonix**’s fell **$20.70 to $17.90**, down 13.5%, as the 28 October 2025 session left the window, improving its drawdown cell **2.9** points to 86.3% on a session worth only 1.1 of them. The other ten did not move, so every other change in the drawdown column is price. The two widest moves in it are **Perpetua**’s cell out **3.9** points to **53.8%** and **Ambarella**’s out **3.0** to 34.7%, which hands the tightest drawdown on the page back to **Nabtesco** at 32.6% one run after taking it; Schaeffler’s out 2.3 to **53.7%** keeps it the widest in tier one, four and a half points clear of Yaskawa’s 49.2%. **This run the roll is in the twelve-month column and it is a large one on two names that barely moved.** Tokyo’s 1y base stepped from 22 onto **23 October 2025** and the Western names’ from 21 onto **22 October 2025**. **Harmonic Drive’s base fell 8.5%** in that one step, ¥3,270 to **¥2,992**, so a **+0.41%** session lifted its 1y cell **14.5 points**, from +49.1% to **+63.6%**: **13.9** of those points are the calendar and seven tenths of one is the price. **Compass Diversified’s fell 9.5%**, $8.04 to **$7.28**, for a **13.9-point** gain, +27.9% to **+41.8%**, on a +0.39% session. Two more gain three and a half points on the base and give part of it back to the tape: **USA Rare Earth** −61.7% to −60.6%, where a base stepping $26.73 to $24.46 is worth +3.6 points against a −5.87% session, and **MP** −46.7% to −44.8% on a base $74.11 to $69.80, +3.3 points against −2.38%. **Perpetua** is the one name where the calendar does nothing: its base moved $23.71 to $23.65, two tenths of a point, so the whole of −21.0% to **−27.0%** is the **−7.90%** session. Read every column against its own base date, and treat a large move in one on a small move in the price as a date problem until shown otherwise. A venue date is still not a page date; read each column against its own. One trap, recorded on 19 Aug 2026: Yahoo’s daily bar array can lag a venue’s latest close, and that afternoon it still ended at the 18 August Tokyo session while the quote metadata already carried the 19 August close. Taking the bars at face value would have shown Harmonic Drive at ¥5,960 rather than ¥5,670, understating the fall by about five percentage points. The German feed keeps swallowing sessions, and the running tally is the argument for the method rather than against it. The deepest case in the series stays closed and confirmed: **24 September**, the largest session of the series at −6.2% and the only one closed on four independent routes rather than two, backfilled at exactly **€6.54**, the figure the auction bar, previousClose, chartPreviousClose and an independent German quote service all carried at the time. That took the tally to **sixteen holes, sixteen reconstructions, sixteen confirmed to the cent** through the end of that week. It has not stayed there for a single run since. The **nineteenth** hole, at **1 October**, has now healed, and it is the strongest confirmation in the series because it was the one closed on _three_ independent routes rather than two. That bar now reads exactly **€6.23**, which is what all three carried when it was absent: regularMarketPrice stamped **17:37:57 CEST** against a 17:30 close, previousClose on a five-minute request made after Xetra reopened, and a five-minute bar stamped exactly **17:30:00**, the auction itself rather than a proxy for it. Three routes, one figure, and the vendor has now agreed with all three. The **twentieth** hole, at **5 October**, was the first in the series that could not be closed on a second daily route, and it has now healed on the vendor's own bar at exactly **€6.24** — the figure regularMarketPrice carried stamped **17:39:37 CEST** against a 17:30 close, in the single narrow window it was available, since on the next request the same field had already rolled to a live print. That capture window is the whole of the method here, and the single-route reading has been vindicated rather than merely left standing. A **twenty-first** hole opened immediately behind it, at **6 October**, and that one closed on two routes: regularMarketPrice at **€6.36** stamped **17:35:22 CEST**, and a five-minute bar stamped exactly **17:30:00** carrying the same **€6.36**, which is the auction itself rather than a proxy for it. The **Frankfurt line carries the identical hole** for a second consecutive session, which again confirms the gap is one upstream failure rather than a broken Xetra ticker, and again removes the second German daily route in the same stroke. So the tally reads **twenty-one holes, twenty-one reconstructions, twenty-one confirmed to the cent**, and that sentence is only true as of this run: when it was first published the **6 October** bar was still null, so the twenty-first case was reconstructed rather than confirmed, and this note said “none open” while the entry above it correctly reported the bar as missing. The vendor has since backfilled 6 October at exactly **€6.36**, which is what both routes carried, so the claim has caught up with itself and nothing published ever depended on the difference. The **twenty-fourth** hole, at **9 October**, has now healed as well, and it was one of the better-closed cases in the series rather than a single-route reading. The vendor's bar now reads exactly **€5.88** on **932,268 shares**, which is what both routes carried when it was absent: regularMarketPrice stamped **17:39:43 CEST** against a 17:30 close, and a five-minute bar stamped **17:35:00** at the same **€5.88** on **435,144 shares**, against 2,745 to 18,539 in each of the six bars before it. That took the tally to **twenty-four holes, twenty-four reconstructions, twenty-four confirmed to the cent, and none open**, on the session that took Schaeffler to its lowest close since 2 October 2025. Three further holes have opened since, at **12, 13 and 14 October**, and the first two have already backfilled at exactly the figures published here, so the tally at that point read **twenty-seven holes, twenty-seven reconstructions, twenty-six confirmed to the cent** with the newest open. The procedure has been contradicted on none of the thirty-four closed cases, which is as settled as a workaround gets. It stays in place, because the gap has recurred often enough that its absence for one session would not be evidence it had stopped. The rule stands with its clause: when a German bar is missing, prefer the auction print, and where the intraday series stops short of the auction, previousClose on a request made after the venue reopens is the auction, not the last tick before it. The US feed's own hole has healed too. The **22 September** daily bar was missing for four days for Ambarella, Lynas's ADR, Perpetua, Novonix, REMX, Compass Diversified and the iShares gold trust, while MP, USA Rare Earth and the SPDR gold trust carried it throughout — the same split, and mostly the same names, as the 28 August attrition recorded below. All seven were closed from previousClose and all seven have now returned, matching the reconstruction on every one: Ambarella at **$67.06**, REMX at **$70.97**, Perpetua at **$24.76**, Lynas at **$10.33**, Novonix at **$2.85**. That last one is the figure the claim audit's newest row turns on, and it is worth noting that the page's _table_ had it right on the day and only the prose did not. The same gap hit an earlier run from the opposite direction, when the daily bars for _every_ US name ended at 2 September while the metadata already carried the completed 3 September close. The metadata timestamp, converted to exchange-local time, is what decides whether a session is complete. Returns are computed on the local primary line, so no currency translation is embedded. This is a dated snapshot — the page cannot fetch live quotes and does not pretend to. It is refreshed daily at 07:00 UTC, and the compiled date in the masthead moves every time any content on this page changes, so that stamp is always the age of what you are reading.

**28 Aug 2026: the corporate-action trap.** Novonix changed its ADS ratio on 27 August, effectively a one-for-ten reverse split of the US line. Yahoo adjusted its quote metadata immediately — the 52-week high became $38.60 — but had not yet restated the daily close history, which still ran to $0.435. Read together, the two would have put Novonix 98.9% below its 52-week high, an invented 10× drawdown. Yahoo restated the history in stages, and for a fortnight put the seam in the wrong place: the 10× step sat between **21 and 24 August**, three sessions before the change took effect on 27 August, so 24–26 August were carried at post-change prices they never traded at. **That restatement has now completed.** The whole series is on the post-change ratio, the misplaced step is gone, and a scan of two years of daily bars finds no jump larger than the two genuine ones from November 2024 and October 2025. The figures on this page were computed throughout with both sides on the post-change ratio, scaling closes at the seam the data actually contained rather than the one the corporate action implied — and **they did not move when the vendor caught up**, which is the only real test such a patch gets. Taken raw at the time, the same mismatch would have reported Novonix _up 618% in a month_, the earlier error wearing the opposite sign. Any screen mixing adjusted metadata with unadjusted bars across a corporate action produces the same class of error, silently.

**A hole in the German feed, since filled.** Yahoo's daily bars carried no 28 August close for any German single stock — Schaeffler, SAP, BMW and Continental all return null for that session, as do the Frankfurt, Munich, Düsseldorf and Hamburg lines, so it is one upstream gap rather than several — while the DAX index itself printed 26,570.0 that day. Taking the bars at face value would have held Schaeffler at its 27 August close of €7.35 and reported a 2.4% rise as a flat session. The €7.53 on this page is reconstructed from the intraday series for 28 August and cross-checked against an independent price history that agrees to the cent and matches Yahoo's own daily bars for 24–27 August exactly. A missing bar is not the same as an unchanged price. Yahoo backfilled the session later the same day at **€7.53**, matching the reconstruction to the cent.

**The German hole is now a standing feature — and it heals itself, which is the useful part.** It has recurred on eleven sessions since late August, and it is not Schaeffler-specific: SAP and Siemens return null for exactly the same days, so it is one upstream gap rather than a broken ticker. What has changed is that the earlier gaps have since _backfilled_, which turns them into a test of the patch. Yahoo now carries daily bars for 28 August and 1–4 September, and every one of them matches the figure this page reconstructed at the time to the cent: **€7.53, €7.22, €7.02, €7.30 and €7.43**. Five reconstructions, five exact confirmations, using two different routes — the neighbouring session's quote metadata, and the intraday series. The 7 and 8 September holes needed a third route, because both were still missing while Xetra had already opened on 9 September, so the metadata carried a live quote rather than either settlement and the standard 5-minute series stops at 17:25, before the closing auction. Requesting the 5-minute series with pre- and post-market bars included returns the auction print itself, stamped 17:35 CEST: **€7.34** on 7 September and **€7.43** on 8 September, the latter on 229,007 shares against a few thousand in each of the preceding bars — the volume signature of a closing auction rather than a stray tick. The 8 September bar has since backfilled at exactly **€7.43**, a sixth reconstruction confirmed; the 7 September bar has still not returned at all. The **9 September** close, €7.40, came back a fourth way and the simplest one yet: the daily request, made before Xetra's 10 September auction, still carried 9 September's settlement in regularMarketPrice stamped 17:39 CEST, and the 5-minute request made at the same moment returned the same €7.40 in its previousClose field. Two independent fields, one figure — and the vendor has since backfilled that session at exactly **€7.40**, a seventh reconstruction confirmed. Taken from the bars alone the page would have carried Schaeffler at €7.43 and called a live session unchanged. **11 September** behaved the same way: €7.19 in regularMarketPrice, stamped 17:35:12 CEST against a 17:30 close, while the daily bars stopped at 10 September — and that session has since backfilled at exactly **€7.19**, an eighth reconstruction confirmed. **14 September** has since backfilled at exactly **€6.95**, a ninth reconstruction confirmed, and **15 September** at exactly **€6.78**, a tenth — that one having been closed by two independent routes on the day it opened, the metadata auction stamp and the 17:35 five-minute bar on 575,656 shares, rather than after the fact. **16 September** took a different second route again: €6.86 in regularMarketPrice stamped 17:38:13 CEST, with the five-minute series stopping at 17:25 and never printing the auction, so the corroboration came from previousClose on a request made after Xetra reopened — and that session has since backfilled at exactly **€6.86**, an eleventh reconstruction confirmed. So has **7 September**, at exactly **€7.34**, a twelfth — the hole that stayed open for a fortnight, and the only one that needed the extended-hours request to close at all. **21 September** closed the quickest of that group: the five-minute series printed the 17:30 auction at €6.99 on the day, matching regularMarketPrice at 17:35:12 CEST, and it has since backfilled at exactly **€6.99**, a thirteenth reconstruction confirmed. **22 September** followed at exactly **€7.13**, a fourteenth — the most instructive of them, because the intraday series stopped at 17:25 showing €7.16 and the page took previousClose over it; had it taken the continuous print it would have published a 2.4% session rather than a 2.0% one. **23 September** has now backfilled at exactly **€6.97**, a fifteenth. Sixteen holes, sixteen reconstructions, fifteen of them confirmed to the cent; only **24 September**, the largest session of the lot at −6.2%, is outstanding — and it is the one closed on four routes rather than two. That is settled enough to be a procedure rather than a surprise — when a German bar is missing, take the auction print from the metadata timestamp and check it against a second field before believing it.

**Bars that come and go.** Novonix's daily history around its 27 August ADS ratio change is not stable between requests: 28 August read $3.51 at one run and had disappeared again, along with 26 and 27 August, by the next. Left alone, that would have moved the name's correlation and shown a four-session fall as a one-day move. The 28 August close is pinned to the verified figure so the series does not wobble with the feed. By 3 September the instability had spread well beyond Novonix: the whole **28 August US session** had gone missing for Ambarella, REMX, Perpetua, Lynas and Compass Diversified while surviving for MP and USA Rare Earth — per-ticker attrition, not one dropped day. **That has since healed.** All five carry 28 August again, and Novonix's bar for it has returned at exactly **$3.51**, the figure this page pinned when it kept vanishing. The weekly series takes each ISO week's _last available_ close, which absorbs this without shifting a week; where it removes a Friday, that week ends on the Thursday instead. Where a vendor's history changes underneath you, the fix is to record what you verified, not to re-derive it each run.

**4 Sep 2026: a continuous futures series that is not continuous.** This page has priced gold off GC=F, Yahoo's front-month gold contract, to sanity-check Perpetua's moves. That series silently rolled: GC=F now resolves to _Gold Dec 26_, and its own bar array still carries $4,366.3 for 2 September — the expiring contract — while the December contract closed that session at **$4,414.6**. Differencing across the roll would have reported a **+4.0%** gold session on 3 September against **+1.9%** for the bullion ETFs, which agree with each other to four hundredths of a point (SPDR Gold Shares +1.85%, iShares Gold Trust +1.88%). A futures move measured across a contract change is a spread, not a return. Gold figures on this page are now taken from the spot-tracking ETFs, which also settle on the same 16:00 ET clock as Perpetua and the miners they are being compared with — the earlier $4,366/oz reference has been dropped for the same reason.

**Correlations** use 103 completed ISO-week returns from 104 aligned weekly closes, from ISO week 2024-47 to 2026-46, ending on 13 November 2026. Week 46 is included because its last trading session has settled in every tracked venue. The weekly-rebalanced basket averages the six tier-one returns each week. On 20 November, with the same completed window, the exchange-local daily-close route and a separate aggregation of Yahoo’s native weekly bars agreed to six decimal places on every tracked correlation, including the basket’s 0.3576 against REMX. The Lynas chart uses LYSDY to match the price table. LYC.AX was checked separately.

**Weekly timestamps use exchange-local time.** Xetra’s summer Monday 00:00 weekly anchor falls on Sunday 22:00 UTC. Applying the current winter offset to it yields Sunday 23:00 and assigns the close to the preceding ISO week. Each timestamp uses the offset in effect on its date, and a residual local Sunday anchor is assigned to the following Monday. The same check covers the US fall clock change. The 30 October window reproduced at 0.3918 before it was rolled forward.

**Earlier correlation checks.** **The window did not move on the 20 October run, and three cells carried on it were still slightly wrong.** ISO week 43 opened on Monday 19 October and does not close until **Sunday 25 October**, so the sample holds at **104 aligned weekly closes and 103 returns**, week **2024-43** to week **2026-42**, exactly where the 17 October run set it, and basket-to-REMX holds at **0.3885** on both methods. On an unchanged window a cell cannot move, so a cell that recomputes differently was carried wrong rather than overtaken. Three did: **Lynas to REMX at 0.6471** against a published 0.64, **Nidec to the basket at 0.6229** against 0.61, and **Harmonic Drive to the basket at 0.8152** against 0.81. Each rounds a point higher than the figure on the page, each was published on a window identical to this one, and all three are corrected here to **0.65**, **0.62** and **0.82** in both bar charts and in the tier-two table. The claim audit carries it. None of them changed a reading, and on the window used for that October note the ranges were **0.19** to **0.47** against REMX, **0.54** to **0.82** against the basket, and **0.37** to **0.65** for the minerals names. **The account that follows belongs to the run on which the window last advanced, and it did so on a judgement rather than a rule, so the judgement is kept on the record rather than left for a reader to infer from a moved number.** ISO **week 42** runs to Sunday 18 October, so on a weekday run it would still be in progress and excluded, which is what the last three runs did. This run is a **Saturday**: every venue has settled Friday 16 October and none reopens before Monday, so week 42’s last close is _final_ even though its calendar week is not over. The week is therefore counted as completed, the sample now ends at the week ended **16 October**. The count published with it, **104** returns against 103 on the preceding runs, was wrong: a two-year request rolls forward at both ends, so the week gained at the end is paid for by one dropped at the start, and the aligned sample held at 104 closes and **103** returns. The judgement about which week ends the sample stands; only the count was overstated, and it is corrected throughout on the 19 October run. That is a shift from “the calendar week has ended” to “the week’s trading has ended”; it can only ever differ on a weekend run, and it is the kind of definitional change this page has an audit row about making silently. Week 42 was heavy on both sides and they fell together: the basket **−8.31%** against REMX **−11.05%**, with Harmonic Drive −13.99%, Nidec −10.45%, Ambarella −8.50%, Yaskawa −8.47% and Nabtesco −7.64%, against Schaeffler nearly flat at −0.85%, and in the minerals tier USA Rare Earth −16.08%, Novonix −13.93%, MP −12.94%, Lynas −12.60% and Perpetua −6.93%. Two tiers falling hard in the same week sits close to the existing relationship rather than against it, so the headline figure rose slightly and no conclusion in section 04 moves: basket-to-REMX goes from **0.3687** to **0.3877**. Both bucketing methods were recomputed from freshly fetched data and agreed on every series, 0.388 each on the headline figure. **The one name the new week moved against the grain is Schaeffler.** It was the only line in the basket not to fall materially in a week the basket fell eight per cent, and that pulled its correlation to the basket _down_, from **0.6070** to **0.6016** — the exact mirror of the mechanism recorded below, where being the worst line in a falling week had pushed the same figure up. **The account that follows is the record of the last run in which the window actually advanced.** Week 41 closed on Sunday **11 October**, so the end advanced from the week ended 2 October to the week ended **9 October**, and the whole of a busy week entered the sample in one step: TDK's four-session run and the three losses that followed it, Monday's Tokyo rally, Thursday's de-rating of the gearbox layer, Friday's partial reversal, Yaskawa's first-half filing, two sessions of rare-earth-specific selling and Friday's seven-point move in Ambarella. The headline figure moved **seven ten-thousandths**, from **0.3681** to **0.3687**. The largest change anywhere on the page is Schaeffler, **0.5997** to **0.6070** against the basket and 0.4564 to **0.4645** against REMX; Lynas goes to **0.6333**, MP to **0.5524**, Perpetua to **0.5248**, USA Rare Earth to **0.3678** and Novonix to **0.3603**. Both bucketing methods were recomputed from scratch on fresh data and agreed to **four decimal places on every one of the fifteen series**, basket included, which is the stricter of the two outcomes this check produces. **One prediction in this note was right in its number and wrong in its reason, and the correction matters more than the figure.** It said the week would “substantially cancel”. It did not cancel: week 41 was a broadly down week on both sides, the basket **3.27%** lower and REMX **3.54%** lower, with nine of the eleven charted names falling. The figure barely moved because the two sides fell by nearly the same amount, which places the new observation almost on the existing line and so confirms the relationship instead of altering it. Equal-sized falls on both sides and mutual cancellation inside one side are different mechanisms that happen to produce the same small change, and only one of them was what occurred. The mechanical rules all held: a window ending at the last completed ISO week cannot move inside an open one, so neither a Friday close nor a Saturday refresh advances the end, which advances on the first refresh _after_ a Sunday — the 12 October run was that refresh, and no run inside the same week can repeat it, which is why this run still holds that window. The start, separately, rolls forward on its own, which was the lesson of 16 Sep 2026 and is visible here: the sample start stepped from ISO week 2024-41 to **2024-42** as the end advanced, holding the length at 103. And a standing claim in this note has to be withdrawn, because testing it is what this check is for. The note has said that a _two-year_ daily request yields **103** weekly returns rather than 104, that a true 104-return window needs a three-year request, and that the figures here are built from one. The first two are right; the third is not. Recomputing from a three-year request at a genuine **104** returns gives, on the window as it now stands, basket-to-REMX **0.3710**, Lynas **0.6336**, Schaeffler-to-basket **0.6022** and Harmonic Drive-to-REMX **0.1629**, none of which is what this page publishes; the same test on the window ending 2 October gave 0.3670, 0.6322, 0.5991 and 0.1581. Every figure here reproduces exactly on **103** returns and on nothing else, so 103 is what the window is, and the counts in the charts and captions have been corrected to say so. The difference is about a thousandth and changes no conclusion, which is precisely why it survived: a discrepancy too small to notice in the output is only ever caught by rebuilding the claim rather than re-reading it. **The four-week comparison between the tiers uses a different construction again, and it is now stated:** the basket's 1m figure is the mean of the six tier-one names' own 1m cells — equal-weighted, not rebalanced, each name on its own venue's 30-calendar-day lookback — so a reader can rebuild it from the table above. This page previously quoted 5.2% for that figure and it cannot be reproduced on any construction tried; the claim audit carries that too. Correlation measures co-movement, not causation — treat 0.37 as "these travel together loosely," not as a precise constant.

**Schaeffler’s trading line matters.** The originally published SFFLY ADR correlation to the basket was 0.16. On the window ending 30 October it was 0.29 for the ADR and 0.61 for Xetra. The current Xetra figure is 0.61. The Xetra line is used for the basket. The Lynas ADR remains the tracked US line and its Australian listing is checked separately.

**18 Aug 2026 correction: the alignment bug.** Every correlation figure this page had published through 17 Aug 2026 was computed from Yahoo's native weekly bars (interval=1wk). Those bars anchor Tokyo and Xetra weeks roughly one calendar week off from New York's — confirmed directly by comparing native bars against daily closes re-bucketed into true ISO weeks, which showed a consistent one-week offset for every Tokyo and Xetra ticker tested but not for US tickers. Because the tier-one basket is four-sixths Japanese names, this silently corrupted every cross-region figure: the robotics basket's correlation to REMX read 0.19 on the misaligned data and about 0.40 once daily closes were re-bucketed by calendar week (0.39 on the current window, weekly-rebalanced) — confirmed by two independent methods (ISO-week bucketing and Friday-anchored resampling) that agree to three decimal places. Correlations computed entirely within one region (e.g. TDK, a Tokyo name, against the mostly-Tokyo basket) barely moved, which is exactly what a same-direction shift on both sides of a comparison would produce, and is why the bug went unnoticed until a cross-region figure was checked against an independent method. Every correlation on this page, in every table, chart and paragraph, is now computed from daily closes re-bucketed into ISO calendar weeks.

**Use the primary listings.** OTC quote metadata can sit stale for months. NJDCY still carries a February quote, while its 19 November daily history and S&P Global’s dated OTC history agree on $2.62. SFFLY’s 19 November daily bar gives $5.715. The Tokyo and Xetra primary listings remain the basket inputs.

**Not investment advice.** This is a research snapshot assembled from public filings, company releases and press reporting, built to check a social-media thesis rather than to endorse it.

09 Sources

[S&P Global via Stock Analysis, 20 November Tokyo closing histories, cross-checked against Yahoo auction-stamped quote metadata; other Tokyo links are in the feed log.](<https://stockanalysis.com/quote/tyo/6324/history/>)  
[Harmonic Drive Systems, 19 November Japanese half-year presentation, combined AI and humanoid sales, orders and product targets.](<https://contents.xj-storage.jp/xcontents/AS06879/72ea9203/5839/4596/a735/21a7aa1f8a6b/140120261118582977.pdf>)  
[Compass Diversified, 18 November Form 8-K, supplemental indentures and additional PIK principal.](<https://ir.compassdiversified.com/node/14346/html>)  
[Perpetua Resources, 17 November non-binding memorandum on potential tungsten evaluation, funding and supply.](<https://perpetuaresources.com/perpetua-resources-announces-non-binding-mou-with-a-major-american-aerospace-defense-company-to-evaluate-domestic-tungsten-opportunities/>)  
[Lynas via ASX, March-quarter report describing the US rare-earth oxide letter of intent and intended Seadrift funding reallocation.](<https://announcements.asx.com.au/asxpdf/20260421/pdf/06ypfwgh3p6xmg.pdf>)  
[MarketScreener, Schaeffler’s €5.31 Xetra quote at 17:35:01 CET on 17 November.](<https://es.marketscreener.com/cotizacion/accion/SCHAEFFLER-AG-176430175/cotizaciones/>)  
[finanzen.net, Frankfurt venue history, Schaeffler at €5.36 on 17 November.](<https://www.finanzen.net/historische-kurse/schaeffler?exchange=FSE>)  
[S&P Global, Lynas Sydney post-auction history.](<https://stockanalysis.com/quote/asx/LYC/history/>)  
[EODData, Schaeffler OTC end-of-day history.](<https://www.eoddata.com/stockquote/OTCBB/SFFLY.htm>)  
[Agility Robotics, 16 November amended S-1 covering resale of PIPE shares, conditional Digit 5 orders and the early-access schedule.](<https://investors.agilityrobotics.com/node/7716/html>)  
[Agility Robotics, separate 16 November amended S-1 covering other resale securities and shares issuable on warrant exercise.](<https://investors.agilityrobotics.com/sec-filings/sec-filing/s-1a/0001213900-26-120832>)  
[Perpetua Resources, 16 November Stibnite construction and road update.](<https://perpetuaresources.com/project/local-notices/>)  
[Lynas via ASX, annual general meeting notice for 25 November.](<https://announcements.asx.com.au/asxpdf/20261029/pdf/075xzndxqk7000.pdf>)  
[MarketScreener, Schaeffler’s €5.47 Xetra auction quote on 16 November, stamped 17:35:08 CET.](<https://de.marketscreener.com/kurs/aktie/SCHAEFFLER-AG-176430175/?iSubCodezb=190544066>)  
[Investing.com Xetra history, independent 16 November Schaeffler OHLC and volume.](<https://www.investing.com/equities/schaeffler-historical-data>)  
[Stock Analysis, Lynas 17 November Sydney close of A$12.03.](<https://stockanalysis.com/quote/asx/LYC/statistics/>)  
[EODData, Novonix 17 November Sydney close of A$0.076.](<https://www.eoddata.com/stockquote/ASX/NVX.htm>)  
[Nidec, 16 November application for a filing extension and explanation of outstanding review work.](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1116-01/261116-01en.pdf>)  
[Nidec via TDnet, 16 November approval of the extension to 16 December.](<https://www.release.tdnet.info/inbs/140120261116581515.pdf>)  
[Nidec via TDnet, 16 November notice of delayed quarterly results summary.](<https://www.release.tdnet.info/inbs/140120261116581518.pdf>)  
[Nikkei index provider, 16 November closing summary.](<https://indexes.nikkei.co.jp/nkave/archives/summary?dt=20261116&idx=nk225>)  
[S&P Global via Stock Analysis, Lynas 16 November Sydney close A$13.11.](<https://stockanalysis.com/quote/asx/LYC/history/>)  
[S&P Global via Stock Analysis, Novonix 16 November Sydney move of −6.25%. The table rounds the A$0.075 close to two decimals.](<https://stockanalysis.com/quote/asx/NVX/history/>)  
[Nabtesco, 13 November Q3 results Q&A. Hamamatsu capacity and RVmini.](<https://www.nabtesco.com/cms/wp-content/uploads/Results_Briefing_QA_for_FY2026_Q3_e.pdf>)  
[GXO and Apptronik, 12 November letter of intent and trial milestones.](<https://investors.gxo.com/news-releases/news-release-details/gxo-and-apptronik-expand-partnership-scale-humanoid-deployment>)  
[REalloys, 13 November Form 8-K furnishing quarterly results.](<https://www.sec.gov/Archives/edgar/data/1567900/000118518526005348/realloys8k111126.htm>)  
[REalloys quarterly release. US sintered NdFeB design target and development conditions.](<https://realloys.com/realloys-press-release/realloys-reports-third-quarter-2026-results/>)  
[REalloys, 13 November Form 10-Q. Status of the potential JS Link transaction.](<https://www.sec.gov/Archives/edgar/data/1567900/000162828026073547/aloy-20260930.htm>)  
[Agility Robotics, 13 November Form 8-K/A. Operating-company financial statements.](<https://investors.agilityrobotics.com/static-files/a19009fc-8955-4e95-bdd9-aa13bcc9ac5f>)  
[Yahoo Finance, Schaeffler Xetra 13 November daily OHLC backfill checked 15 November.](<https://finance.yahoo.com/quote/SHA0.DE/history/>)  
[Perpetua Resources, 13 November project update and 17 November webinar announcement.](<https://perpetuaresources.com/perpetua-resources-announces-third-quarter-2026-financial-results-and-provides-project-update/>)  
[Nidec, 14 November notice on a possible semiannual filing extension.](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1114-01/261114-01en.pdf>)  
[Nidec, 13 November cancellation of the extraordinary meeting record date.](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1113-01/261113-01en.pdf>)  
[Nidec, 13 November resolution on the interim dividend.](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1113-02/261113-02en.pdf>)  
[Perpetua Resources, 13 November Form 10-Q. EXIM closing conditions and updated Stibnite timetable.](<https://www.sec.gov/Archives/edgar/data/1526243/000110465926127248/ppta-20260930x10q.htm>)  
[S&P Global via Stock Analysis, 13 November Xetra close and volume.](<https://stockanalysis.com/quote/etr/SHA0/history/>)  
[S&P Global via Stock Analysis, 13 November Nidec OTC close.](<https://stockanalysis.com/quote/otc/NJDCY/history/>)  
[S&P Global via Stock Analysis, 13 November Frankfurt close.](<https://stockanalysis.com/quote/fra/SHA0/history/>)  
[Nabtesco, 13 November Q3 financial statements and KPMG AZSA review.](<https://www.nabtesco.com/cms/wp-content/uploads/Summary_of_Consolidated_Financial_Statements_for_FY2026_Q3_e.pdf>)  
[Nabtesco, 13 November Q3 briefing. Precision-reducer orders and plant utilisation.](<https://www.nabtesco.com/cms/wp-content/uploads/Results_Briefing_Material_for_FY2026_Q3e.pdf>)  
[Nabtesco, August Q2 briefing. Prior plant capacity and utilisation disclosure.](<https://www.nabtesco.com/cms/wp-content/uploads/Results_Briefing_Material_for_FY2026_Q2e.pdf>)  
[TDnet, Nabtesco Q3 release at 16:00 JST on 13 November.](<https://www.release.tdnet.info/inbs/140120261113578060.pdf>)  
[Timken, 31 May 2022 completion of the Spinea acquisition.](<https://news.timken.com/Timken-Completes-Acquisition-of-Spinea>)  
[USA Rare Earth, 12 November Form 10-Q. Serra Verde Phase 1 offtake.](<https://www.sec.gov/Archives/edgar/data/1970622/000197062226000072/usar-20260930.htm>)  
[Agility Robotics, 4 November business combination closing announcement.](<https://www.sec.gov/Archives/edgar/data/2074973/000121390026116898/ea031120401ex99-2.htm>)  
[Agility Robotics, 10 November Form S-1. Nasdaq listing and Digit 5 early-access target.](<https://www.sec.gov/Archives/edgar/data/2074973/000121390026118025/ea0311220-s1_agility.htm>)  
[S&P Global via Stock Analysis, 12 November Xetra close and volume.](<https://stockanalysis.com/quote/etr/SHA0/history/>)  
[S&P Global via Stock Analysis, 12 November Schaeffler OTC close.](<https://stockanalysis.com/quote/otc/SFFLY/history/>)  
[FT market data, 13 November Lynas Sydney close A$13.19.](<https://markets.ft.markitdigital.com/data/equities/tearsheet/summary?s=LYC%3AASX>)  
[S&P Global via Stock Analysis, 13 November Novonix Sydney close A$0.080.](<https://stockanalysis.com/quote/asx/NVX/history/>)  
[USA Rare Earth, 9 November Serra Verde release. October shipments and first November offtake sales.](<https://investors.usare.com/news-releases/news-release-details/usa-rare-earth-announces-start-commercial-scale-shipments-and>)  
[Niron Magnetics, 10 November announcement of CES recognition, a 6–9 January 2027 exhibit and target production dates.](<https://nironmagnetics.com/niron-magnetics-named-ces-2027-innovation-awards-honoree-in-vehicle-tech-advanced-mobility-for-rare-earth-free-iron-nitride-magnets/>)  
[DFARS 225.7018, covered materials, 1 January 2027 sourcing expansion and exceptions.](<https://www.acquisition.gov/dfars/subpart-225.70-authorization-acts-appropriations-acts-and-other-statutory-restrictions-foreign-acquisition>)  
[Nasdaq, USA Rare Earth history. The 11 November close was $8.265 before rounding.](<https://www.nasdaq.com/market-activity/stocks/usar/historical>)  
[S&P Global via Stock Analysis, Xetra history. 11 November close €5.63 and 1,022,241 shares.](<https://stockanalysis.com/quote/etr/SHA0/history/>)  
[FT market data, 12 November Lynas Sydney close A$13.34.](<https://markets.ft.markitdigital.com/data/equities/tearsheet/summary?s=LYC%3AASX>)  
[S&P Global via Stock Analysis, 11 November Nidec OTC close $2.70.](<https://stockanalysis.com/quote/otc/NJDCY/history/>)  
[Harmonic Drive Systems via TDnet, 11 November FY3/27 half-year results. Includes orders by region and product and the planned 19 November posting of supplementary material.](<https://www.release.tdnet.info/inbs/140120261106567186.pdf>)  
[Harmonic Drive Systems via TDnet, 11 November revision to full-year results and dividend forecasts.](<https://www.release.tdnet.info/inbs/140120261106567176.pdf>)  
[Schaeffler, 10 November Taicang release for Yokohama. Repeats the approximately €400m humanoid order intake disclosed in October.](<https://www.schaeffler.co.in/en/news-media/press_releases/press_releases_detail.jsp?id=88226241>)  
[Nasdaq, Ambarella historical quotes. The 10 November close was $60.355 before display rounding.](<https://www.nasdaq.com/market-activity/stocks/amba/historical>)  
[S&P Global via Stock Analysis, Ambarella daily history. Shows the 10 November close rounded to $60.36.](<https://stockanalysis.com/stocks/amba/history/>)  
[FT market data, Lynas Rare Earths on ASX. Confirms the 11 November A$13.92 close.](<https://markets.ft.markitdigital.com/data/equities/tearsheet/summary?s=LYC%3AASX>)  
[S&P Global via Stock Analysis, Nidec OTC daily history. Confirms the 10 November $2.50 close.](<https://stockanalysis.com/quote/otc/NJDCY/history/>)  
[TDK, 30 October release on the Ojiya TMR sensor project, expected investment of about ¥100bn and scheduled 2029 operations.](<https://www.tdk.com/en/news_center/press/20261030_01.html>)  
[Aichi Steel, 6 November announcement of flexible MAGFINE bonded magnet sheet samples at electronica.](<https://www.aichi-steel.co.jp/ENGLISH/news/mt-assets/docs/20261106_news_.pdf>)  
[Japan Exchange Group, Securities on Special Alert. Nidec is listed under code 6594.](<https://www.jpx.co.jp/english/listing/measures/alert/index.html>)  
[MarketScreener OTC quote history for Nidec’s NJDCY ADR. Cross-check for recent US trades.](<https://www.marketscreener.com/quote/stock/NIDEC-CORPORATION-120787337/quotes/>)  
[MOFCOM and China’s General Administration of Customs, joint Announcement No. 58 of 2026, published 9 November. Extends the suspension of six October 2025 notices from 11 November 2026 through 10 January 2027.](<https://www.mofcom.gov.cn/zwgk/zcfb/art/2026/art_91466027f465467985942a639a9485f4.html>)  
[USA Rare Earth, 9 November Q3 results at 16:05 ET. Discloses September block-magnet sales, October Serra Verde shipments and updated 2027 milestones.](<https://investors.usare.com/news-releases/news-release-details/usa-rare-earth-reports-third-quarter-2026-financial-results>)  
[USA Rare Earth, SEC Exhibit 99.1 to the 9 November 8-K. Filed copy of the Q3 earnings release.](<https://www.sec.gov/Archives/edgar/data/1970622/000197062226000066/exhibit991-earningsrelease.htm>)  
[MP Materials, 5 November Q3 results. GM magnet qualification completed and Magnetics revenue of $18.6m from precursors.](<https://investors.mpmaterials.com/investor-news/news-details/2026/MP-Materials-Reports-Third-Quarter-2026-Results/default.aspx>)  
[MP Materials, 5 November Form 10-Q. First saleable finished magnet deliveries to GM are expected in Q4 2026.](<https://www.sec.gov/Archives/edgar/data/1801368/000180136826000057/mp-20260930.htm>)  
[Compass Diversified, 4 November Q3 results. Arnold’s segment sales and adjusted EBITDA are disclosed separately.](<https://ir.compassdiversified.com/news-releases/news-release-details/compass-diversified-reports-third-quarter-2026-financial-results>)  
[Nidec, 5 November notice renewing its two bilateral commitment lines, ¥300bn each.](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1105-01/261105-01en.pdf>)  
[Perpetua Resources, 4 November Stibnite drill results, including gold and tungsten at SB597.](<https://perpetuaresources.com/perpetua-announces-new-gold-discovery-at-stibnite-gold-project-continued-high-grades-from-clark-tunnel-fault-zone/>)  
[Ambarella, issuer calendar for the 24 November Q3 FY2027 call at 13:30 PST.](<https://investor.ambarella.com/node/17436>)  
[Nabtesco, issuer calendar for the planned 13 November FY2026 third-quarter results.](<https://www.nabtesco.com/en/about/ir/calender/>)  
[Harmonic Drive Systems, issuer IR calendar. See also the separately sourced market calendar below.](<https://www.hds.co.jp/english/ir/event/calendar/>)  
[Lynas, issuer September-quarter briefing on 29 October 2026.](<https://lynasrareearths.com/investors-media/briefings/>)  
[S&P Global Market Intelligence via Stock Analysis, Xetra SHA0 daily history. Independently confirms the 9 November €5.56 close.](<https://stockanalysis.com/quote/etr/SHA0/history/>)  
[Yahoo Finance chart, SHA0.DE five-minute history. The 9 November close was captured from the 17:36:55 CET quote timestamp while the daily OHLC bar was null. This endpoint rolls forward.](<https://query2.finance.yahoo.com/v8/finance/chart/SHA0.DE?range=5d&interval=5m&includePrePost=true>)  
[Schaeffler AG — “Schaeffler with solid nine-months results”, company release, Herzogenaurach, **3 November 2026**, published 07:45 CET, read the same morning. The validated nine-month figures: revenue **€17,481m**, EBIT before special items **€794m** at a **4.5%** margin, free cash flow before M&A **−€40m** with the quarter at +€260m, net income attributable to shareholders **€11m** after €201m of special items, net financial debt €5,296m at 2.3x. Guidance reconfirmed by the board on 26 October. E-Mobility EBIT before special items **−€571m** with about €240m of charges, €220m booked in the quarter. On humanoids it says only that the company “won new customers and continued to expand our order book”, with **no euro figure and no robotics segment**, which is why the €400m of 13 October is still carried as a preliminary number.](<https://www.schaeffler.de/de/news_medien/pressemitteilungen/pressemitteilungen_detail.jsp?id=88224327>) [USA Rare Earth, Inc. — “USA Rare Earth Announces Date for Release of Third Quarter 2026 Results and Conference Call”, company release via GLOBE NEWSWIRE, datelined Stillwater, Okla., **2 November 2026**, read 3 November. Results come **after US markets close on Monday 9 November 2026** with the live call at **4:30 p.m. ET**. Three trackers carried 5, 6 and 13 November against it on the morning of 3 November, one of them under the label “verified”; the claim audit carries that row.](<https://finance.yahoo.com/markets/stocks/articles/usa-rare-earth-announces-date-213000666.html>) [Ambarella Inc — SEC filing index, read **31 October 2026**. The company’s only filing dated **30 October 2026** is a **Schedule 13G/A** (accession 0002100121-26-001058), a passive-holder ownership amendment lodged by a third party. There is no 8-K, no merger agreement and no transaction filing of any kind behind the exclusive-talks report; the most recent company-filed 8-K is dated **3 September 2026**.](<https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001280263&type=&dateb=&owner=include&count=40>) [NXP Semiconductors N.V. — SEC filing index, read **31 October 2026**. Its **30 October 2026** filings are six **Form 4** insider reports, and its most recent 8-K and 10-Q are both dated **27 October 2026**, the third-quarter results. Nothing on the reported acquirer’s side discloses a transaction either, which is the point of checking both.](<https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001413447&type=&dateb=&owner=include&count=40>) [Seeking Alpha, **30 October 2026** at 2:47 PM ET — the exclusivity report at its most careful: Ambarella “is _speculated_ to be in exclusive talks with a possible buyer”, with the board said to have met that week to discuss multiple offers. No originating outlet is named and no party is quoted. It puts the shares “down about 2%”, which is the figure that reconciles to the tape.](<https://seekingalpha.com/news/4661636-ambarella-in-exclusive-talks-with-one-bidder-report>) [Crypto Briefing, **30 October 2026** — the fullest version of the same report: talks with NXP said to have been entered after the board weighed multiple bids including Qualcomm and Nvidia, described explicitly as **non-binding**, with neither company confirming. Puts the share move at “between 2% and 3.6%”.](<https://cryptobriefing.com/ambarella-exclusive-talks-nxp-acquisition/>) [GuruFocus, **30 October 2026** — the audited claim: “Ambarella (AMBA) in Exclusive Talks with NXP Semiconductors, Shares Drop 5.5%”. The close was **$63.12**, down **2.26%** from $64.58. The same outlet filed the 23 October GM magnet claim this page already carries.](<https://www.gurufocus.com/news/9159728/ambarella-amba-in-exclusive-talks-with-nxp-semiconductors-shares-drop-55>) [Nidec — “Nidec Files Suit for Damages against Former Executives”, released **27 October 2026**, Kyoto. The primary document for the nine named former officers: six former directors (Shigenobu Nagamori, Hiroshi Kobe, Akira Sato, Hiroyuki Yoshimoto, Jun Seki, Mitsuya Kishida) and two former corporate auditors (Kazuya Murakami, Hiroyuki Ochiai) under Article 423(1) of the Companies Act, plus a former Vice President who was CFO from May 2022 to March 2026 under Article 644 and Article 415(1) of the Civil Code or Nidec’s Vice President Regulations. The release states the company “has decided today to file” the suit, “plans to examine… the amounts of compensation to be sought” and will file “as soon as the Company is ready to do so”, so the headline is ahead of the text; the claim audit carries the mismatch. It also records that appropriation of surplus was excluded from the Committee’s investigation because amended securities reports for prior years had not been submitted (28 Oct 2026)](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/1027-01/261027-01en.pdf>)  
[Nidec — corporate release list, read after Tokyo’s close on **28 October 2026**. Ends at the 27 October damages disclosure, with nothing dated 28 October: no audit opinion, no amended securities reports, no internal control system confirmation. The securities report for the year ended 31 March 2026 was filed on **30 September 2026**, alongside notices of material weaknesses in internal control over financial reporting and a disclaimer of opinion in the internal control audit report. A confirmation letter to the exchange need not be issued as a press release, so the absence is not evidence either way (28 Oct 2026)](<https://www.nidec.com/en/corporate/news/>)  
[USA Rare Earth — “USA Rare Earth Breaks Ground on Advanced Rare Earth Metal and Magnet Manufacturing Facility in Blacksburg, South Carolina”, company release dated **9 September 2026**: approximately **$1.2bn** of investment, roughly **490** jobs, an 800,000 sq ft plant on a 124-acre site in Bailey Industrial Park, Cherokee County, targeting **6,400 t** a year of sintered NdFeB and **5,000 t** of strip-cast metal and alloy, commissioning targeted for **2028**. This is the date behind the claim audit row on the 27 October “Just Broke Ground” headline (28 Oct 2026)](<https://investors.usare.com/news-releases/news-release-details/usa-rare-earth-breaks-ground-advanced-rare-earth-metal-and>)  
[TDK Corporation — “Notice regarding Revision of Projections for Consolidated Financial Results, Dividend of Surplus (Interim) and Year-End Dividend Projections”, filed to TDnet at **30 October 2026, 15:30 JST**. The primary document for the upgrade: net sales ¥2,580,000m to **¥3,050,000m** (+18.2%), operating income ¥295,000m to **¥365,000m** (+23.7%), profit before taxes ¥300,000m to ¥370,000m (+23.3%), net income attributable to owners ¥225,000m to **¥280,000m** (+24.4%), EPS ¥118.54 to ¥147.50; interim dividend **¥25.00** against a ¥20.00 forecast and ¥16.00 a year earlier, ¥47,458m in total, effective 2 December 2026; year-end projection ¥20.00 to ¥25.00, full year **¥40.00 to ¥50.00**. Reason given: AI data-centre demand significantly above the prior year, sluggish smartphone-led ICT demand, rechargeable batteries and sensors expanding on new models, nearline HDD heads and suspension assemblies up significantly. Filed with the first-half results and a notice of transition to a company with three committees, all three timestamped at the bell, after the 7.22% session (30 Oct 2026)](<https://finance-frontend-pc-dist.west.edge.storage-yahoo.jp/disclosure/20261030/20261030560760.pdf>)  
[Nidec Corporation — timely-disclosure record, read after Tokyo’s close on **30 October 2026**. Carries “Notice Regarding the Status of the Development and Operation of the Internal Management System” in Japanese and English at **30 October 15:30 JST**, the submission Rule 503(2) of the Listing Regulations requires one year after the Security on Special Alert designation took effect, and “Notice Regarding Organizational and Personnel Changes” at **29 October 15:30**. Nothing on either date restates the FY3/26 accounts or lifts PwC Japan’s disclaimer of opinion (30 Oct 2026)](<https://finance.yahoo.co.jp/quote/6594.T/disclosure>)  
[Lynas Rare Earths — September 2026 quarter activities report scheduled for **29 October 2026**, with an analyst and shareholder briefing at 10am Sydney time, per a company announcement reported 19 October 2026. The date rests on a single secondary report and is carried as such; it is where heavy rare-earth output and any movement on the **Seadrift** decision would appear (28 Oct 2026)](<https://kalkine.com.au/news/announcements/lynas-rare-earths-to-release-september-2026-quarterly-results-on-29-october>)  
[TDK — second-quarter FY2027 results expected **30 October 2026**. Sources disagree on whether the release is before the open or after the close, so the date is carried and the time is not. This is the publication at which any segment disclosure sizing magnets against the battery and components businesses would appear; none has appeared in any quarter this page has watched (28 Oct 2026)](<https://www.marketbeat.com/earnings/reports/2026-10-30-tdk-co-stock/>)  
[Shanghai Metals Market — Pr-Nd oxide average of roughly **720,500 yuan/mt** on **27 October 2026**, about **−0.14%** month on month, which is a little under the **$110/kg** DoD-backed Western floor at prevailing rates while ex-China assessments sit well above it. The two-price market tier two is built on is intact (28 Oct 2026)](<https://www.metal.com/rare-earth-oxides/201102250162>)  
[Reuters market wrap, 26 October 2026 — the Nikkei 225 fell for a third consecutive session on chip weakness imported from Friday’s Wall Street session, with energy-linked shares higher on a stronger oil price; morning-trade breadth put **166** of the 225 constituents up against 59 down, so the index fell on heavyweights rather than on the market. The index close itself, **62,189.65, −0.80%**, is taken from ^N225 rather than from any wrap (26 Oct 2026)](<https://www.businesstoday.com.my/2026/10/26/nikkei-falls-for-third-session-as-chip-stocks-weigh/>)  
[Harmonic Drive Systems, 13 May FY3/26 results. Operating profit was ¥2.567bn against ¥6m in FY3/25.](<https://finance-frontend-pc-dist.west.edge.storage-yahoo.jp/disclosure/20260513/20260513529667.pdf>)  
[Nidec — Tokyo Stock Exchange designation as a Security on Special Alert effective **28 October 2025**, requiring an Internal Control System Confirmation Letter one year on, reviewed by Japan Exchange Regulation, with delisting if the internal management system is judged inadequate. Reporting attributed to _Nikkei_ adds that an unqualified audit opinion and a corrected securities report are sought by the end of October 2026. Two further strands — discipline of about **20** executives and employees this month, and an exchange decision on the listing at **year-end** — trace to a single secondary report citing unnamed sources and are carried on this page as single-sourced rather than as fact (26 Oct 2026)](<https://www.investing.com/news/stock-market-news/nidec-urged-to-fix-internal-controls-amid-delisting-risk-93CH-4390128>)  
[MP Materials — third-quarter 2026 results scheduled for **5 November 2026**, the next disclosure of magnetics segment revenue and of the Fort Worth commercial shipments to General Motors guided for the fourth quarter (24 Oct 2026)](<https://fool.com/investing/2026/10/24/is-mp-materials-stock-a-millionaire-maker>)  
[GuruFocus, 23 October 2026 — the audited claim: “On October 23, 2026, General Motors Co (NYSE: GM) announced the termination of its long-standing supply agreement” with “a U.S. manufacturer of rare-earth magnets”, counterparty unnamed. Wrong on the date (notice delivered 1 July 2026), wrong on the mechanism (reported by Reuters from SEC filings on 22 October; GM declines to comment because the arbitration is confidential) and missing the counterparty (e-VAC Magnetics). See the claim audit (23 Oct 2026)](<https://www.gurufocus.com/news/9139382/general-motors-gm-ends-rareearth-magnet-supply-deal-amid-domestic-sourcing-challenges>)  
[CBT News, 23 October 2026 — the same story published the same day with the provenance intact: “According to regulatory documents filed with the Securities and Exchange Commission”, a customer terminated its agreement effective **1 July** per Energy Fuels’ filing, the customer identified as GM by a person familiar with the situation, counterparty named as e-VAC Magnetics. Cited in the claim audit as evidence the facts were available (23 Oct 2026)](<https://cbtnews.com/gm-cancels-magnet-supply>)  
[NOVONIX Limited — Form 6-K, ASX announcement dated 22 October 2026 (Chattanooga, TN): a **second amortisation payment of US$7,000,000** to YA II PN, Ltd, an affiliate of Yorkville Advisors Global; convertible debentures of equal face value redeemed; amount outstanding reduced to **US$21,000,000** against a US$100m maximum aggregate face value; the amortisation event triggered by the daily ASX VWAP sitting below the **A$0.12** floor price; further payments due on a rolling 30-day basis at 20% of the amount outstanding while the event continues; payment funded from existing cash reserves. The primary source for the Novonix card’s cash-burn read (22 Oct 2026)](<https://www.sec.gov/Archives/edgar/data/0001859795/000119312526428257/nvx-ex99_1.htm>)  
[Reuters via Investing.com — “GM cancels deal with new US rare-earth magnet plant, documents say”: General Motors terminated its long-term permanent-magnet supply agreement with e-VAC Magnetics, the US unit of Vacuumschmelze, on **1 July**; VAC opened arbitration days later seeking damages for wrongful termination, the dispute having first surfaced in SEC filings in April 2026; the $500m South Carolina plant launched late last year at just under 2,000 t/yr, with VAC’s chief executive saying in September 2025 that 90% of output was committed to GM; Energy Fuels, acquiring VAC for $1.9bn, flags a possible purchase-price adjustment. No name on this page is a party to it (22 Oct 2026)](<https://www.investing.com/news/stock-market-news/exclusivegm-cancels-deal-with-new-us-rareearth-magnet-plant-documents-say-4967695>)  
[Axios Twin Cities — Niron Magnetics is running a site search with Savills for its second plant, a **$1.8bn**, 1.6-million-square-foot facility employing more than **700** with construction from 2028, and will name a state by the end of 2026; chief executive Jonathan Rowntree says Minnesota “is still on our list” but is “struggling to be competitive with some of the other states”, the state development agency having put $10m into the Sartell plant. The 10,000 t/yr follow-on therefore has no site yet (22 Oct 2026)](<https://www.axios.com/local/twin-cities/2026/10/22/niron-magnetics-plant-selection-state>)  
[InvestingLive — 23 October 2026 US session recap, used to cross-check the close: S&P 500 −0.52% to 7,623.68, Nasdaq Composite −0.21% to 25,955.34, Russell 2000 −0.29% to 2,775.90 and Ambarella **+5.35%** all agree with the chart data to the figure. Its Dow line does not: it gives 52,220.21 and −653.07 points against an actual 52,215.16 and −652.82, and its own two numbers do not reconcile with each other. Immaterial at 0.01 of a percentage point, so it is noted here rather than carried as an audit row (23 Oct 2026)](<https://investinglive.com/stocks/dow-leads-stocks-lower-small-caps-win-the-week>)  
[Schaeffler AG — “Ready for the mass market: Schaeffler presents formed strain wave gearboxes for humanoid robots”: validation testing of the formed strain wave gearboxes concluded, laying the foundation for the start of mass manufacturing in 2027, beginning in Germany and rolling out to other regions; forming rather than cutting claimed to cut manufacturing cost by more than 25% and material consumption by more than 75%. Company claims, not independently verified here. The disclosure that put a second tier-one name on the strain-wave row of the parts table (13 Aug 2026)](<https://www.schaeffler.com/en/media/press-releases/press-releases-detail.jsp?id=88210241>)  
[China export-control timeline — MOFCOM Announcement No. 70 of 2025, dated 7 November 2025, suspending Announcements 55, 56, 57, 58, 61 and 62 until 10 November 2026; the April 2025 licensing regime for samarium, gadolinium, terbium, dysprosium, lutetium, scandium and yttrium never suspended; Announcement No. 72 suspending Article 2 of Announcement 46 of 2024 until 27 November 2026. Last verified 1 October 2026 with no extension published, and a second tracker reading the ministry’s 2026 announcement list on 16 October found it running to Announcement No. 48 with nothing extending or replacing No. 70. The basis for downgrading this page’s own verdict on the 6 October countdown piece (checked 20 Oct 2026)](<https://rareearthdirectory.com/policy/china-export-controls-timeline/>)  
[China Rare Earth Industry Association — price index posted at 263.6 on 9 October 2026, against 267.0 on 4 August, with NdPr oxide quoted at ¥729.8–749.8/kg, bracketing Benchmark’s 25 September assessment of RMB729,500/t. Relayed through a paywalled trade outlet rather than read from the association directly, and carried on that basis; the small-volume terbium transactions above $12,000/kg reported in the same item are trade talk with no assessor behind them and are not carried as an assessment (9 Oct 2026)](<https://rareearthexchanges.com/news/china-rare-earth-price-index-ndpr-terbium/>)  
[Nidec Corporation — “Nidec Receives an Investigation Report from the Executive Responsibility Investigation Committee”: the Committee established 13 March 2026 to examine whether current and former directors, auditors and vice presidents “bear legal liability, including whether they breached their duties in the execution of their responsibilities” over the improper accounting; the Report to be disclosed only after redaction for trade secrets and privacy; Nidec to “deliberate legal actions such as lodging a claim for damages”; Board and Audit and Supervisory Committee meetings convened the same day. Carried on Business Wire at 12:23 AM EDT, which is 13:23 JST, inside the Tokyo afternoon session. The claim audit’s fifty-first row (16 Oct 2026)](<https://www.nidec.com/en/corporate/news/2026/1016-01/>)  
[Schaeffler AG — preliminary figures for the first nine months of 2026: revenue €17.5bn (prior year €17.7bn), EBIT margin before special items 4.5% (4.2%), nine-month free cash flow before M&A −€40m (+€47m), Q3 EBIT margin 4.2% and free cash flow +€260m, about €240m of one-off E-Mobility special charges, humanoid-robotics orders on hand of approximately €400m after two new Greater China customers, and confirmation of 2026 guidance and the 2028 mid-term targets; validated statement due 3 November (13 Oct 2026)](<https://www.schaeffler.com/en/media/press-releases/press-releases-detail.jsp?id=88222017>)  
[Japan Exchange Group — extension of Tokyo Stock Exchange trading hours with the arrowhead 4.0 launch: the afternoon session runs 12:30–15:30, continuous trading ending 15:25 ahead of a 15:30 itayose closing auction, effective 5 November 2024. The basis for treating a 15:00 Tokyo quote as a pre-auction print rather than a close (read 16 Oct 2026)](<https://www.jpx.co.jp/english/corporate/news/news-releases/1030/20230920-01.html>)  
[Investing.com — “At the close in Tokyo, the Nikkei 225 added 1.66%” against an actual 15:30 close of 65,324.29, +1.39%; the 1.66% is the 15:00 print of 65,493.63. The claim audit’s forty-ninth row (16 Oct 2026)](<https://in.investing.com/news/stock-market-news/japan-shares-higher-at-close-of-trade-nikkei-225-up-166-5636588>)  
[24/7 Wall St — USA Rare Earth “up 4%” at $10.77, MP “up 2%” at $40.82 and REMX “down 0.9%” at $55.26, stamped 1:53pm ET against closes of $11.09 (+7.46%), $41.61 (+4.16%) and $55.74 (−0.07%). The claim audit’s fiftieth row (15 Oct 2026)](<https://247wallst.com/investing/2026/10/15/usa-rare-earth-climbs-4-as-month-long-selloff-pauses-mp-materials-ticks-up-united-states-antimony-dips/>)  
[ad-hoc-news.de — a market report stamped 15 Oct 13:20 CEST quoting Schaeffler’s “Xetra close of EUR 5.94 on October 14, 2026” and labelling its own −2.86% figure an intraday decline: the first confirmation of a reconstructed Schaeffler close from outside the price vendor (15 Oct 2026)](<https://www.ad-hoc-news.de/boerse/news/nebenwerte/schaeffler-confirms-guidance-as-schaeffler-stock-falls-2-86-percent/70329414>)  
[Yaskawa Electric — consolidated results for the first half of the year ending 28 February 2027: revenue ¥285,598m (+9.8%), operating profit ¥22,058m (−5.5%), and the revised FY2/27 forecast of ¥600,000m revenue and ¥57,500m operating profit (9 Oct 2026)](<https://www.yaskawa-global.com/wp-content/uploads/2026/10/20261009_en.pdf>)  
[Nidec — internal-control improvement page: the 28 October 2025 Security on Special Alert designation in its timeline, and the prose dating it 28 October 2026; the claim audit’s forty-first row (read 9 Oct 2026)](<https://www.nidec.com/en/corporate/internal-control-improvement/overview/>)  
[Reuters — China Rare Earth Group in talks to buy Shenghe Resources, the holder of roughly 3% of MP Materials, and Shenghe’s Shanghai exchange filing denying any transfer of control (18 Sep 2026)](<https://www.reuters.com/markets/deals/china-rare-earth-group-talks-buy-mp-materials-shareholder-shenghe-resources-2026-09-18/>)  
[Milk Road AI — humanoid supply chain post (9 Aug 2026)](<https://x.com/MilkRoadAI/status/2086577616853815592>)  
[Jiji via nippon.com — Nidec’s FY3/26 results: ¥564.6bn net loss, ¥518.9bn operating loss, revenue ¥2,708.7bn, impairments above ¥632bn, and PwC Japan’s disclaimer of opinion (30 Sep 2026)](<https://www.nippon.com/en/news/yjj2026093000856/>)  
[Reuters via Investing.com — Nidec slumps after its auditor declines to sign off; the ¥1,888 intraday low, and the end-October internal-control filing or delisting (1 Oct 2026)](<https://www.investing.com/news/stock-market-news/nidec-corp-shares-slump-after-auditor-declines-to-sign-off-on-earnings-4926185>)  
[Investing.com — the “slumped 17.2% to ¥1,949” figure, against a ¥2,100 close; the claim audit’s thirty-third row (1 Oct 2026)](<https://www.investing.com/news/stock-market-news/why-is-nidec-stock-plunging-today-93CH-4926188>)  
[TDK and Taiyo Yuden — memorandum of understanding to begin discussions on a potential alliance in MLCCs and inductors, with a possible future capital alliance (29 Sep 2026)](<https://passive-components.eu/tdk-and-taiyo-yuden-explore-alliance-for-mlccs-and-inductors/>)  
[Benchmark Mineral Intelligence — Q3 2026 rare-earths price review: Chinese PrNd oxide RMB729,500/t (~$108.7/kg), ex-China PrNd $106,500/t CIF, dysprosium $3,500/kg CIF Europe +56%, terbium $6,745/kg +23% (25 Sep 2026)](<https://source.benchmarkminerals.com/article/ex-china-heavy-rare-earths-continue-to-surge-q3-2026-rare-earths-price-review>)  
[Perpetua Resources at Mining Forum Americas — EXIM loan closing guided to Q4 2026, FID after close, initial production late 2029, construction capital above $3bn (28 Sep 2026)](<https://www.investing.com/news/transcripts/perpetua-resources-at-mining-forum-americas-2026-shift-to-construction-93CH-4921335>)  
[24/7 Wall St — the “surges 6% to $48.29” figure for MP, against a $47.29 close and a $48.665 session high (30 Sep 2026)](<https://247wallst.com/investing/2026/09/30/mp-materials-surges-6-as-traders-weigh-critical-minerals-growth-story-usa-rare-earth-rises-3-critical-metals-stays-flat/>)  
[Ambarella — CV7 launch, CES 2026](<https://www.ambarella.com/news/ambarella-launches-powerful-edge-ai-8k-vision-soc-with-industry-leading-ai-and-multi-sensor-perception-performance/>)  
[Ambarella — FY2026 results](<https://ambarella.gcs-web.com/news-releases/news-release-details/ambarella-inc-announces-fourth-quarter-and-fiscal-year-2026>)  
[Schaeffler — Hexagon Robotics partnership](<https://www.schaeffler.com/en/media/press-releases/press-releases-detail.jsp?id=88184987>)  
[The Robot Report — Schaeffler robot deployment](<https://www.therobotreport.com/schaeffler-humanoid-partner-build-deploy-hundreds-robots/>)  
[Investing.com — Nidec results delay and probe](<https://www.investing.com/news/stock-market-news/nidec-delays-q3-results-amid-accounting-irregularity-probe-93CH-4469763>)  
[Nidec — internal control investigation overview](<https://www.nidec.com/en/corporate/internal-control-improvement/overview/>)  
[Nidec — receipt of the third-party committee’s final report on accounting (17 Apr 2026)](<https://www.nidec.com/files/user/www-nidec-com/corporate/news/2026/0417-01/260417-01en.pdf>)  
[Nidec — Q1 FY3/27 disclosure beyond 45 days (5 Aug 2026)](<https://www.stocktitan.net/news/NJDCY/nidec-announces-the-disclosure-of-the-financial-results-for-the-h4pqqx2pszoe.html>)  
[Nidec — shareholder request to file action (5 Aug 2026)](<https://finance.yahoo.com/markets/stocks/articles/nidec-announces-request-shareholder-file-015300997.html>)  
[Nidec — second, joint shareholder request to file action (21 Aug 2026)](<https://www.businesswire.com/news/home/20260821331686/en/Nidec-Announces-a-Request-by-Shareholders-to-File-an-Action>)  
[Nidec — annual securities report filing deadline extended to 30 September 2026](<https://www.marketscreener.com/news/nidec-announces-approval-of-an-extension-of-the-filing-deadline-of-annual-securities-report-for-the-ce7f5fdcd98bf62d>)  
[USA Rare Earth — completion of the Serra Verde combination (3 Sep 2026)](<https://www.globenewswire.com/news-release/2026/09/04/3356489/0/en/usa-rare-earth-completes-combination-with-serra-verde-group.html>)  
[Ambarella — Citi Global TMT Conference remarks (9 Sep 2026)](<https://www.investing.com/news/transcripts/ambarella-at-citis-2026-global-tmt-conference-edge-ai-push-93CH-4894482>)  
[Nikkei — Nidec limit-down, −¥500 (17.60%) to ¥2,340 on the impairment report (28 Sep 2026)](<https://www.nikkei.com/article/DGXZQOFL280T10Y6A920C2000000/>)  
[Bloomberg — Nidec’s statement: executive changes and large-scale impairment under consideration, nothing decided (28 Sep 2026)](<https://www.bloomberg.com/jp/news/articles/2026-09-28/TM1Z03T96OSI00>)  
[Kyodo — Nidec considering large-scale impairment and management changes (28 Sep 2026)](<https://news.yahoo.co.jp/articles/205bf38bad7f8a91f48281488696557ab1bed95d>)  
[Nikkei — the earlier “several hundred billion yen” impairment report (15 Sep 2026)](<https://www.nikkei.com/article/DGXZQOUF153BG0V10C26A9000000/>)  
[Ambarella — Developer Zone on Google Cloud, dated 15 Sep 2026, not 25 Sep](<https://www.globenewswire.com/news-release/2026/09/15/3362176/23306/en/ambarella-expands-its-developer-zone-with-remote-access-to-live-silicon-and-agentic-ai-development-on-google-cloud-platform.html>)  
[Niron Magnetics — first steel at Sartell and a Honda investment (10 Sep 2026)](<https://www.prnewswire.com/news-releases/niron-magnetics-marks-steel-raising-at-sartell-manufacturing-site-302875334.html>)  
[Niron Magnetics — Moog qualifying iron nitride magnets in actuators, AUSA 2026 (2 Oct 2026)](<https://nironmagnetics.com/moog-and-niron-magnetics-to-showcase-rare-earth-free-motor-technology-at-ausa-2026/>)  
[Nidec — receipt of notification of litigation on derivative action, Kyoto District Court case 1749 of 2026 (10 Sep 2026)](<https://www.nidec.com/en/corporate/news/2026/news0910-01/>)  
[24/7 Wall St — rare earths attributed to “U.S.–China thaw hopes” (10 Sep 2026)](<https://247wallst.com/investing/2026/09/10/rare-earth-stocks-tumble-on-u-s-china-thaw-hopes-usa-rare-earth-sinks-4-mp-materials-drops-5-critical-metals-slips/>)  
[CSIS — the 2026 Trump–Xi summits, including the 24 September Washington visit](<https://www.csis.org/programs/trump-xi-2026-summits>)  
[USA Rare Earth — groundbreaking at Blacksburg, South Carolina: 6,400 tpa NdFeB, ~$1.2bn (9 Sep 2026)](<https://www.globenewswire.com/news-release/2026/09/09/3358813/0/en/usa-rare-earth-breaks-ground-on-advanced-rare-earth-metal-and-magnet-manufacturing-facility-in-blacksburg-south-carolina.html>)  
[Reuters via Mining.com — Chinese rare-earth firms halt some US shipments after the RBA sanctions (4 Sep 2026)](<https://www.mining.com/web/china-rare-earth-firms-halt-some-us-shipments-over-geopolitical-worries-sources-say>)  
[Rigzone — Brent +6.3% to $107.63 on Hormuz shipping attacks (10 Sep 2026)](<https://www.rigzone.com/news/wire/oil_soars_as_hormuz_risks_rattle_markets-10-sep-2026-184590-article/>)  
[Nikkei Indexes — Nikkei 225 daily close](<https://indexes.nikkei.co.jp/en/nkave>)  
[Jefferies — Schaeffler target cut to €8.70 from €9.45, Buy retained, analyst Vanessa Jeffriess (23 Sep 2026)](<https://www.finanzen.net/analyse/schaeffler_buy-jefferies_&_company_inc._1106825>)  
[Jefferies — the earlier Schaeffler cut, €10.45 to €9.45 (5 Aug 2026)](<https://www.onvista.de/news/2026/08-05-analyse-flash-jefferies-senkt-ziel-fuer-schaeffler-auf-9-45-euro-buy-0-10-26540026>)  
[Schaeffler −2.24% on the session of the Jefferies cut (23 Sep 2026)](<https://www.ad-hoc-news.de/boerse/news/nebenwerte/schaeffler-stock-falls-2-24-percent-as-jefferies-cuts-target/70169026>)  
[NBC News — Bessent: the Busan Agreement extended from 10 November to 10 January (23 Sep 2026)](<https://www.nbcnews.com/business/economy/us-china-extend-trade-truce-trump-rcna599525>)  
[Reuters via Investing.com — Bessent and He Lifeng meet before the summit; truce extension, 11:40 ET (23 Sep 2026)](<https://www.investing.com/news/stock-market-news/us-treasurys-bessent-chinas-he-to-meet-on-unfinished-business-before-trumpxi-summit-4913529>)  
[Investing.com — Asian rare-earth names fall after the truce extension; Lynas −2% (24 Sep 2026)](<https://www.investing.com/news/stock-market-news/asia-rare-earth-stocks-fall-after-uschina-extend-trade-truce-to-january-4914443>)  
[Bloomberg — China’s magnet exports to the US fall 21% to 512 t in August, on customs data (21 Sep 2026)](<https://www.bloomberg.com/news/articles/2026-09-21/china-rare-earth-magnet-exports-to-us-drop-before-xi-trump-meet>)  
[SMM — China’s total rare-earth magnet exports 5,010 t in August, −6% MoM, −18% YoY](<https://news.metal.com/newscontent/104124359-chinas-rare-earth-magnet-exports-down-6-mom-18-yoy-in-august-2026>)  
[USAGOLD — hawkish Fed remarks, dollar index back above 100, metals liquidation (23 Sep 2026)](<https://www.usagold.com/daily-precious-metals-market-report-september-23-2026/>)  
[Nidec — receipt of the Investigation Committee’s report, publication set for 4 September (2 Sep 2026)](<https://www.businesswire.com/news/home/20260901419270/en/>)  
[Nidec — disclosure of the quality Investigation Report and the company’s response (4 Sep 2026)](<https://www.nidec.com/en/corporate/internal-control-improvement/overview/>)  
[Nidec — response to suspected quality-related misconduct](<https://www.nidec.com/en/corporate/disclosure/quality/>)  
[News On Japan — Nikkei −3.2% on AI selling and bond yields (19 Aug 2026)](<https://newsonjapan.com/article/150426.php>)  
[24/7 Wall St — robotics names in the 18 Aug AI selloff](<https://247wallst.com/investing/2026/08/18/robotics-stocks-punished-tuesday-in-ai-sell-off-ouster-down-10-aeva-technologies-down-12-symbotic-down-4/>)  
[Antimony spot price and year-to-date change](<https://strategicmetalsinvest.com/antimony-prices/>)  
[Shanghai Metals Market — Pr–Nd oxide benchmark](<https://www.metal.com/rare-earth-oxides/201102250162>)  
[EE Times — Nidec gear design for humanoids](<https://www.eetimes.com/how-nidec-is-rethinking-gear-design-for-humanoid-and-mobile-robots/>)  
[Investing.com — Harmonic Drive humanoid orders](<https://www.investing.com/news/stock-market-news/this-japanese-stock-is-surging-on-positive-outlook-for-humanoid-robot-orders-4367329>)  
[JPX/TDnet — Harmonic Drive earnings forecast revision](<https://finance.biggo.com/news/jpx_tdnet_140120260421507113>)  
[Harmonic Drive — FY3/27 forecast raised to ¥74.5bn / ¥8.5bn (7 Aug 2026)](<https://www.tipranks.com/news/company-announcements/harmonic-drive-systems-lifts-full-year-earnings-forecast-on-strong-automation-demand>)  
[MP Materials — DoD partnership (Jul 2025)](<https://www.mpmaterials.com/news/mp-materials-announces-transformational-public-private-partnership-with-the-department-of-defense-to-accelerate-u-s-rare-earth-magnet-independence/>)  
[MP Materials — Q2 2026 results](<https://investors.mpmaterials.com/investor-news/news-details/2026/MP-Materials-Reports-Second-Quarter-2026-Results/default.aspx>)  
[MP Materials — 10X site selection, ~10,000 t total capacity](<https://investors.mpmaterials.com/investor-news/news-details/2026/MP-Materials-Selects-Northlake-Texas-as-the-Site-of-10X-a-New-U-S--Rare-Earth-Magnet-Manufacturing-Campus/default.aspx>)  
[CNBC — Pentagon stake in MP Materials](<https://www.cnbc.com/2025/07/10/pentagon-to-become-largest-shareholder-in-rare-earth-magnet-maker-mp-materials.html>)  
[Nikkei Asia — 15% polysilicon tariff](<https://asia.nikkei.com/economy/trade-war/trump-tariffs/us-imposes-15-tariff-on-polysilicon-in-pushback-against-china>)  
[MINING.COM — China blacklists MP and USA Rare Earth](<https://www.mining.com/china-blacklists-mp-materials-usa-rare-earth-in-critical-minerals-war/>)  
[MINING.COM — USA Rare Earth pulls timeline forward](<https://www.mining.com/usa-rare-earth-brings-forward-commercial-production-plans-by-two-years/>)  
[USA Rare Earth — Stillwater Phase 1a commissioning](<https://www.theglobeandmail.com/investing/markets/stocks/USAR/pressreleases/36563055/usa-rare-earth-achieves-major-operational-and-strategic-milestone-with-commissioning-of-phase-1a-magnet-production-at-stillwater-facility/>)  
[USA Rare Earth — Q2 2026 results, cash position, CEO transition](<https://www.globenewswire.com/news-release/2026/08/10/3342269/0/en/usa-rare-earth-reports-second-quarter-2026-financial-results.html>)  
[USA Rare Earth — 8-K, TMRC acquisition closing (7 Aug 2026)](<https://www.sec.gov/Archives/edgar/data/0001970622/000121390026086814/ea0301206-8k_usarare.htm>)  
[Bloomberg — Oasis urges Nidec to explore privatization on delisting risk (7 Oct 2026)](<https://www.bloomberg.com/news/articles/2026-10-07/oasis-urges-nidec-to-explore-privatization-on-delisting-risk>)  
[Seeking Alpha — Ambarella gains on Betaville report of Qualcomm interest, unconfirmed (9 Oct 2026)](<https://seekingalpha.com/news/4651826-ambarella-gains-on-report-qualcomm-may-be-interested-in-takeover>)  
[USA Rare Earth — Serra Verde definitive agreement](<https://investors.usare.com/news-releases/news-release-details/usa-rare-earth-announces-definitive-agreement-acquire-serra>)  
[USA Rare Earth — 8-K, $1.55bn SPV capitalisation for Serra Verde offtake (24 Aug 2026)](<https://www.sec.gov/Archives/edgar/data/1970622/000121390026092810/ea030292401ex99-1.htm>)  
[The Motley Fool — MP Materials Q2 2026 earnings call transcript (gadolinium deal)](<https://www.fool.com/earnings/call-transcripts/2026/08/13/mp-materials-mp-q2-2026-earnings-call-transcript/>)  
[Lynas — US project updates](<https://lynasrareearths.com/us-project-updates/>)  
[Lynas — Q3 FY26 record revenue](<https://rare-earth-mining.com/lynas-rare-earths-results-q3-fy26/>)  
[Lynas — FY26 results and heavy rare-earth capex escalation (26 Aug 2026)](<https://rareearthexchanges.com/news/lynas-fy26-profit-hre-capex-escalation/>)  
[Rare Earth Exchanges — Lynas Texas project status](<https://rareearthexchanges.com/news/lynas-texas-project-in-limbo-amid-u-s-america-first-shift/>)  
[CNBC — Perpetua secures $2.9bn EXIM loan](<https://www.cnbc.com/2026/05/21/miner-perpetua-resources-secures-2point9-billion-us-loan-for-idaho-gold-antimony-project.html>)  
[MINING.COM — Perpetua breaks ground at Stibnite](<https://www.mining.com/perpetua-starts-building-1-3b-stibnite-gold-antimony-mine/>)  
[Novonix — $103m 48C tax credits certified](<https://www.stocktitan.net/news/NVX/novonix-us-103-million-48c-tax-credits-certified-for-synthetic-9kepfx8eb4ct.html>)  
[Novonix — divesting non-core business](<https://www.novonixgroup.com/news/novonix-to-divest-non-core-business-to-focus-on-synthetic-graphite/>)  
[Novonix — 6-K, one-for-ten ADS ratio change for Nasdaq compliance (12 Aug 2026)](<https://www.sec.gov/Archives/edgar/data/0001859795/000119312526345642/nvx-ex99_1.htm>)  
[Novonix — ASX half-year accounts and Appendix 4D, going-concern material uncertainty (31 Aug 2026)](<https://ir.novonixgroup.com/financial-information/asx-announcements>)  
[Niron Magnetics — $150m Department of War commitment](<https://www.prnewswire.com/news-releases/niron-magnetics-receives-conditional-150-million-department-of-war-commitment-302846367.html>)  
[Northern News Now — Niron Sartell plant](<https://www.northernnewsnow.com/2026/08/08/pentagon-invests-150m-minnesota-magnet-company-niron-magnetics/>)  
[Arnold Magnetic — USA Rare Earth distribution agreement](<https://www.stocktitan.net/news/CODI/arnold-magnetic-technologies-announces-new-usa-rare-earth-e3ux5o4jp3tv.html>)  
[Compass Diversified — Q2 2026 results, Sterno sale, Arnold segment, CEO transition](<https://www.stocktitan.net/news/CODI/compass-diversified-reports-second-quarter-2026-financial-0y129wl1vqw5.html>)  
[The Motley Fool — Compass Diversified Q2 2026 earnings call transcript](<https://www.fool.com/earnings/call-transcripts/2026/08/17/compass-diversified-codi-q2-2026-earnings-call-transcript/>)  
[Compass Diversified — completion of restatement](<https://ir.compassdiversified.com/news-releases/news-release-details/compass-diversified-completes-restatement-previously-issued>)  
[Investing.com — CODI restated financials after Lugano fraud](<https://www.investing.com/news/company-news/compass-diversified-files-restated-financials-after-lugano-fraud-93CH-4395663>)  
[Bain Capital — Hitachi Metals tender offer close](<https://www.baincapital.com/news/bain-capital-led-consortium-announces-successful-close-tender-offer-hitachi-metals-common>)  
[MarkLines — Hitachi Metals renamed Proterial](<https://www.marklines.com/en/news/277915>)  
[TDK — magnet product range](<https://product.tdk.com/en/products/magnet/index.html>)  
[TDK as an NdFeB manufacturer — profile](<https://rare-earth-mining.com/tdk-rare-earth-magnets/>)  
[SCMP — defence mineral supply chain order](<https://www.scmp.com/news/us/politics/article/3361267/trump-order-targets-china-linked-military-mineral-supply-chains>)  
[CSIS — critical minerals executive order analysis](<https://www.csis.org/analysis/new-executive-order-ties-us-critical-minerals-security-global-partnerships>)  
[Rare Earth Exchanges — China price index, Aug 2026](<https://rareearthexchanges.com/news/china-rare-earth-industry-association-reports-267-0-index-as-ndpr-prices-retreat/>)  
[S&P Global — rare earth supply bottlenecks 2026](<https://www.spglobal.com/energy/en/news-research/latest-news/metals/012726-rare-earth-supply-bottlenecks-set-to-persist-in-2026>)  
[Rare Earth Exchanges — humanoids and magnet control](<https://rareearthexchanges.com/news/humanoids-are-entering-the-hype-peak-and-china-controls-the-magnets-that-make-them-move/>)  
[Seeking Alpha — REMX composition analysis](<https://seekingalpha.com/article/4922005-remx-rare-earth-exposure-is-only-one-half-and-all-is-not-well-on-the-other-half>)  
[VanEck — REMX holdings & performance](<https://www.vaneck.com/us/en/investments/rare-earth-strategic-metals-etf-remx/>)  
[Yahoo Finance — Nikkei 225 chart data, checked against index closes each run](<https://query1.finance.yahoo.com/v8/finance/chart/%5EN225>)  
[Yahoo Finance — Nomura Nikkei 225 ETF (1321.T), used to cross-check the index print](<https://query1.finance.yahoo.com/v8/finance/chart/1321.T>)  
[Yahoo Finance — Nikko Nikkei 225 ETF (1330.T), the second cross-check on the index print](<https://query1.finance.yahoo.com/v8/finance/chart/1330.T>)  
[Reuters — Japan’s 10-year JGB yield reaches 3%, first since 1996 (1 Sep 2026)](<https://www.investing.com/news/economy-news/japans-benchmark-bond-yield-rises-to-3-for-first-time-in-30-years-4883532>)  
[Novonix — land purchase dropped and Riverside ramp slowed, per the half-year report](<https://newschannel9.com/news/local/novonix-backs-off-land-purchase-next-to-chattanooga-plant-slows-production-ramp-up-riverside-synthetic-graphite-panasonic-battery-materials-175-acres>)  
[Ambarella — Q2 FY2027 financial results (3 Sep 2026)](<https://www.globenewswire.com/news-release/2026/09/03/3356244/23306/en/ambarella-inc-announces-second-quarter-fiscal-year-2027-financial-results.html>)  
[News On Japan — Nikkei +2.12% to 66,399.84 on a chip-led rebound (7 Sep 2026)](<https://newsonjapan.com/article/150650.php>)  
[Sunday Guardian — the mid-session 8 September Nikkei figure corrected in the claim audit](<https://sundayguardianlive.com/business/japan-stock-market-today-september-8-nikkei-225-gains-04-while-topix-falls-06-check-why-boj-rate-hike-bets-and-stronger-yen-are-driving-markets-279813/>)  
[Sunday Guardian — the mid-session 14 September Nikkei figure corrected in the claim audit](<https://sundayguardianlive.com/business/nikkei-225-falls-today-september-14-japanese-benchmark-drops-17-amid-ai-safety-fears-and-oil-price-surge-check-top-falling-stocks-and-key-reasons-284322/>)  
[Business Recorder — Reuters Japan wrap, 14 Sep 2026: the Topix/Nikkei split, and a second morning index level](<https://www.brecorder.com/news/40439387/japans-nikkei-slumps-as-ai-leaders-raise-safety-concerns-topix-gains>)  
[FinanzNachrichten — Schaeffler Xetra order book, used for the 7 September close](<https://aktienkurs-orderbuch.finanznachrichten.de/sha0.htm>)  
[finanzen.net — Schaeffler daily price history, cross-check for the German feed gaps](<https://www.finanzen.net/historische-kurse/schaeffler>)  
[Yaskawa FY2026 earnings commentary](<https://note.com/kabuya66/n/nc4dd1543fdfc?hl=en>)  
[Nabtesco FY2025 results breakdown](<https://note.com/yama_shukatsu/n/nfd0ea20c5e8b?hl=en-US>)  
[Federal Reserve — FOMC statement, 16 Sep 2026: first hike since 2023, to 3.75–4.00%](<https://www.federalreserve.gov/newsevents/pressreleases/monetary20260916a.htm>)  
[Nikkei 225 at 65,018.95 after the Bank of Japan decision, and Tokyo shut for the holiday run](<https://www.bbntimes.com/financial/nikkei-225-today-tokyo-shut-for-respect-for-the-aged-day-leaving-index-at-65-018-95-after-boj-fuelled-rally>)  
[CNBC — US-listed Greenland names surge after the Denmark security deal (21 Sep 2026)](<https://www.cnbc.com/2026/09/21/greenland-trump-denmark-deal-stocks-mining.html>)  
[Al Jazeera — what the Greenland security agreement does and does not cover](<https://www.aljazeera.com/news/2026/9/19/trump-says-us-has-permanent-control-of-greenland-security-does-it>)  
[24/7 Wall St — the intraday USAR and MP figures corrected in the claim audit](<https://247wallst.com/investing/2026/09/21/critical-metals-surges-25-as-greenland-security-pact-puts-tanbreez-rare-earth-permit-in-focus-usa-rare-earth-climbs-6-mp-materials-rises-4/>)  
[The USAR-has-no-Greenland-assets point, made contemporaneously](<https://blockonomi.com/usa-rare-earth-usar-stock-surges-9-despite-zero-greenland-operations/>)  
[StockAnalysis — Harmonic Drive 6324 daily history, independent cross-check for the 8 Oct close of ¥6,160 (−7.09%)](<https://stockanalysis.com/quote/tyo/6324/history/>)  
[StockAnalysis — TDK 6762 daily history, cross-check for the 8 Oct close of ¥3,353 (−5.01%) and the ¥3,628 peak of 6 Oct](<https://stockanalysis.com/quote/tyo/6762/history/>)  
[StockAnalysis — Nabtesco 6268 daily history, cross-check for the 8 Oct close of ¥4,795 (−4.29%)](<https://stockanalysis.com/quote/tyo/6268/history/>)  
[News on Japan — 7 Oct 2026 Tokyo wrap: Nikkei 70,035.71, −648.27 points, with Tokyo Electron, Advantest, Fast Retailing, Kioxia and TDK the largest drags](<https://newsonjapan.com/article/150979.php>)  
[ad-hoc-news — the 11:00 JST intraday Harmonic Drive quote of ¥6,380 corrected in the claim audit, and the 29th Mechanical Components Technology Exhibition dates](<https://www.ad-hoc-news.de/boerse/news/nebenwerte/harmonic-drive-stock-falls-3-77-percent-during-osaka-expo/70260654>)  
[24/7 Wall St — the 7 Oct US minerals session, independent cross-check for MP $46.31 and USAR $13.23](<https://247wallst.com/investing/2026/10/07/usa-rare-earth-and-mp-materials-fall-5-as-rare-earth-stocks-slide-together-critical-metals-declines-3/>)  
[The 23 Sep 2026 Bessent extension of the US–China truce to 10 January 2027 — the date the 6 Oct countdown coverage had missed](<https://startupfortune.com/us-and-china-extend-their-rare-earth-truce-by-just-two-months-to-january/>)  
[The 6 Oct 2026 piece still counting down to 10 November, corrected in the claim audit](<https://startupfortune.com/chinas-rare-earth-export-truce-expires-november-10-with-no-renewal-in-sight>)  
[Schaeffler AG — “Humanoids making humanoids: Schaeffler opens lighthouse factory in Taicang”: roughly 7,500 m², six humanoid robots across five assembly stations building linear and rotary actuators, with a training and data centre and a digital twin of the line, described as the start of mass production of those components; the release restates humanoid order intake at around €400m after two new Greater China customers in Q3 2026, the same figures as the 13 October nine-month preliminaries (20 Oct 2026)](<https://www.schaeffler.com/en/media/press-releases/press-releases-detail.jsp?id=88222851>)  
[Nidec Corporation — “Nidec Discloses an English Version of the Digest of the Executive Responsibility Investigation Committee’s Investigation Report”, Business Wire, 01:04 EDT: the English translation of the digest attached to the 20 October release, carrying no new substance beyond it (23 Oct 2026)](<https://www.financialcontent.com/article/bizwire-2026-10-23-nidec-discloses-an-english-version-of-the-digest-of-the-executive-responsibility-investigation-committees-investigation-report>)  
[The 23 Oct 2026 Tokyo wrap putting the Nikkei at −0.25% and 62,752.00 against a close of 62,693.16, and carrying TDK correctly at −5.61% to ¥2,624.50 in the same document — corrected in the claim audit](<https://uk.investing.com/news/stock-market-news/japan-shares-lower-at-close-of-trade-nikkei-225-down-025-4927392>)  
[Nabtesco Corporation — “Notice Regarding the Status and Completion of Acquisition of Own Shares”: 3,306,600 shares for ¥14,999,799,800 between 3 August and 30 October 2026 against caps of 4 million shares and ¥15bn, October alone 1,445,500 shares for ¥6,422,216,300, open-market purchase on the Tokyo Stock Exchange, all of it to be cancelled 15 December 2026 (2 Nov 2026)](<https://www.nabtesco.com/cms/wp-content/uploads/News-release-20261102e.pdf>)  
[Nabtesco Corporation — “Notices of change the scheduled date for the announcement of its financial results for FY2026/Q3”: moved from 30 October to **13 November 2026**, because the September amendments to prior-year internal control reports require an independent auditor’s review report to be attached to the quarter (14 Oct 2026)](<https://www.nabtesco.com/cms/wp-content/uploads/News-release-20261014e.pdf>)  
[NOVONIX Limited — September 2026 quarterly activities report and Appendix 4C, filed on Form 6-K: cash US$33.652m at 30 September, net operating outflow US$9.765m, investing outflow US$9.169m, item 8.5 estimated quarters of funding 3.4, facilities US$58.1m fully drawn with nothing unused, customer receipts US$582,000, and the PowerCo agreement restated with St Thomas production deferred from 2027 to 2029 (30 Oct 2026)](<https://www.sec.gov/Archives/edgar/data/1859795/000119312526439824/nvx-ex99_1.htm>)  
[NOVONIX Limited — “NOVONIX and PowerCo amend supply agreement”: the November 2024 binding offtake for a minimum 32,000 tonnes over five years from 2027 replaced by an Amended and Restated Supply Framework Agreement under which volume and price are to be negotiated after qualification (19 Oct 2026)](<https://www.sec.gov/Archives/edgar/data/1859795/000119312526424624/nvx-ex99_1.htm>)  
[Nidec Corporation — “Notice Regarding the Status of the Development and Operation of the Internal Management System”: states that the Written Confirmation of Internal Management System was submitted to the Tokyo Stock Exchange that day, one year after the 28 October 2025 Security on Special Alert designation, and that the exchange will examine it and lift the designation if the system is judged properly established (30 Oct 2026)](<https://www.nidec.com/en/corporate/news/2026/news1030-01/>)  
[24/7 Wall St — USA Rare Earth “at $9.52, down 10%” and MP “at $39.21, down 6%”, stamped 2:08pm ET against closes of $9.62 (−8.56%) and $39.57 (−4.97%) — corrected in the claim audit (30 Oct 2026)](<https://247wallst.com/investing/2026/10/30/critical-metals-sinks-11-as-european-lithium-deal-nears-new-share-issuance-usa-rare-earth-tumbles-10-mp-materials-falls-6/>)  
[tech-insider.org — the Dow at “52,342.56, down just 0.25%” and the Nasdaq Composite at “26,184.72, down 1.38%” given as Friday 30 October closes, both being the 29 October session to the decimal against actual Friday closes of 51,770.31 (−1.09%) and 25,437.00 (−2.86%) — corrected in the claim audit (30 Oct 2026)](<https://tech-insider.org/dow-vs-nasdaq-october-2026-big-tech-selloff-gap/>)  
[MP Materials Corp — third-quarter 2026 results confirmed for after the close on Thursday **5 November 2026**, call at 5:00pm ET (announced 8 Oct 2026)](<https://investors.mpmaterials.com/investor-news/news-details/2026/MP-Materials-Announces-Date-for-Third-Quarter-2026-Financial-Results-and-Webcast/default.aspx>)
