Research console

Next actionon the Q3 2026 revenue update (~late October), read Miu Miu's quarter first. It is the topic management has now dodged two quarters running (dodge log). Then read Prada-brand growth against Q2's +6.3%. Before that date, the owner may want to ask whether the Milan labour investigation's document request (2025-12-04, names Prada and Versace) has produced anything. It is absent from every filing read.

9 of 10 closed · 29 edges · 8 documents · 1 dossier · worked 23 Sep 2026

Open profile →

Next actionon the Q3 2026 10-Q (late October), re-run one table — Note 5's "(favorable) and adverse net PYD by segment and tail length". The whole name turns on it. Q2 2026 was short-tailed $201m favourable against long-tailed $32m adverse; a year earlier it was $152m favourable against $1m favourable. Two more quarters on that trajectory and the long-tail charge starts to exceed what the property book can release, at which point the reported combined ratio moves toward the 82.5% underlying figure without a single catastrophe.

6 of 10 closed · 11 edges · 8 documents · 1 dossier · worked 22 Sep 2026

Open profile →

Next actionwatch Item 5.02 8-Ks between now and the Q3 call. Two of the company's most senior people left in fifteen days — the Executive Chair who built the restructuring and the CEO of General Insurance, which after the Corebridge exit is effectively the whole company. The filings say no disagreement and the stock did not move (−0.9% across both announcements). A third departure at that level, or a named external hire into General Insurance, is the fact that decides whether this is an orderly succession or the start of something. The Q3 call on ~2026-11-04 is the first time anyone gets to ask.

6 of 10 closed · 19 edges · 10 documents · 1 dossier · worked 23 Sep 2026

Open profile →

Next actionread the September 2026 monthly catastrophe release (~2026-10-15) before the quarter closes. July and August 2026 are already $1.43bn against $558m for the whole of Q3 2025. If September lands anywhere near the July/August run rate, Q3 2026 catastrophe losses are roughly three times the prior-year quarter, against a Q2 2026 Property-Liability underwriting income of $2,006m — and the quarter will be the first clean test of whether the reserve releases are used to fill the hole.

6 of 10 closed · 17 edges · 12 documents · 1 dossier · worked 23 Sep 2026

Open profile →

Next actiontrack warrant vesting. AMD has issued warrants over 320 million shares at $0.01 to OpenAI (2025-10) and Meta (2026-02). None had vested at 2026-06-27. Each 10-Q's warrant note is the single most valuable recurring read on this name, because vesting is where the customer relationships turn into ~20% dilution.

5 of 10 closed · 25 edges · 7 documents · 1 dossier · worked 21 Sep 2026

Open profile →

Next actionre-test "immaterial in the aggregate" at the FY2026 10-K. The SEC's Office of Manufacturing put exactly that question to Amphenol about its 2024 and 2025 acquisitions, the company answered that they were not material individually or together, and the staff closed the review on 2025-06-25 without requiring anything. Seven months later Amphenol closed CommScope for ~$10.5bn — by its own words "the largest acquisition in the Company's history" — which added ~$7.0bn of goodwill in one quarter and took goodwill above book equity. The answer that satisfied the staff was about a different roll-up than the one that now exists.

6 of 10 closed · 25 edges · 10 documents · 1 dossier · worked 21 Sep 2026

Open profile →

Next actiondecide whether Arm's FY26 growth is a business result or a parent's decision. Related-party revenue — SoftBank Group and Arm China — rose $676m, 82%, and supplied 74% of the $913m of total revenue growth, while revenue from external customers grew 7%. The auditor's sole critical audit matter is revenue recognition on long-term contracts "including related parties". A controlling shareholder at 86.4% can time a licence. Q1 FY27 partly answers it — external revenue reaccelerated to +24% — and the next two quarterly splits settle it.

6 of 10 closed · 13 edges · 5 documents · 1 dossier · worked 21 Sep 2026

Open profile →

Next actionwatch the second capacity step. ASML has planned +30% on 2027 low-NA EUV (off ~65 units) and DUV immersion (off ~130), and is "investigating" a further 30% for 2028. The 2026-09-20 cluster file named TSMC's capacity discipline the best single bubble indicator for this complex and flagged that TSMC is not on the watchlist. ASML is the same indicator one layer up, it publishes quarterly, and it is already on the list. Whether "investigating" becomes "planning" is the disclosure to wait for; it would be the earliest credible top signal available on this source ladder.

6 of 10 closed · 16 edges · 5 documents · 1 dossier · worked 21 Sep 2026

Open profile →

Next actionthe customer-concentration number has moved a long way and it is the single most important change any name in this slot produced. One distributor customer went from 32% of FY25 revenue to 50% of Q3 FY26 revenue; the top five end customers went from ~40% to ~55%. Update _clusters/ai-capex-2026-09-20.md, whose AVGO row still reads 32%.

5 of 10 closed · 26 edges · 7 documents · 1 dossier · worked 21 Sep 2026

Open profile →

Next actionon the fiscal-Q1 2027 10-Q (~late October) read the deposit table and answer one question the 8-K does not: did the ~$1.3bn of Capital One IRA balances that did not transfer arrive later, go somewhere else, or never exist by closing? Then read the purchased-loan average balance and yield in the MD&A average-balance table — that is the FY2027 earnings bridge, and it is the number most likely to be missing from any model built off the headline.

5 of 10 closed · 11 edges · 7 documents · 1 dossier · worked 22 Sep 2026

Open profile →

Next actionread the Q2 2026 prepared remarks. This run reached the transcript only from turn 2, where the CEO's opening is a single short paragraph and the call jumps to Q&A — so the only forward number in hand ("revenue up more than 3x next year") is a UBS analyst's restatement that the CEO did not dispute. Everything in S8 downstream of FY2027 rests on it. One more transcript call with offset=0 and a higher limit, or the Q2 earnings 8-K, closes that gap.

6 of 10 closed · 17 edges · 6 documents · 1 dossier · worked 21 Sep 2026

Open profile →

Next actionread the FY2025 10-K (2026-03-02) and the DEF 14A (2026-04-22). This pass built S2 and S3 almost entirely off the Q2 2026 10-Q and the September 8-K because the promotion made those the live documents; the annual report's risk factors, the auditor's report and the Class B voting arithmetic are all still unread. Second: the DDTL 5.5 facility, announced on the Q2 call as the instrument that finances shorter-duration customer contracts — it is the mechanism behind the margin uplift management would not decompose, and it should appear in the next 10-Q's debt table.

6 of 10 closed · 28 edges · 7 documents · 1 dossier · worked 23 Sep 2026

Open profile →

Next actionread the FY2026 10-K (Feb–Mar 2027) for Vantage's first loss-development table before anything else, then the equity-sleeve size against net reserves ($1.55bn at 2026-06-30). Before that, the Q3 10-Q (~2026-11-05/-10): Vantage capital contributions, the equity sleeve's share of invested assets, and whether the S&P action is disclosed.

9 of 10 closed · 34 edges · 15 documents · 1 dossier · worked 23 Sep 2026

Open profile →

Next actionread the DEF 14A filed 2026-03-23. This pass located it and did not read it, so insider ownership %, the compensation structure headline and what the bonus is paid on are all unfilled — and they matter more than usual here, because the U.S. government is a large holder that has agreed to vote as the board recommends, which makes the proxy the only place the governance arithmetic is written down. Second: the Q3 2026 call (guided for October) is the first read on whether the Escrowed Shares mark reverses.

6 of 10 closed · 26 edges · 10 documents · 1 dossier · worked 23 Sep 2026

Open profile →

Next actionon the Q3 2026 release (~2026-10-22) read two numbers before anything else: gross written premium ex-Commercial-Property, and the current accident year ex-catastrophe loss ratio. The first says whether a 32.7% decline in one division is a division problem or the front edge of a company problem — ex-CP growth was +3.7% in Q2 and +4.8% over the half, so it is still a division problem today. The second is the only underwriting series a reserve release cannot flatter, and at 57.0% (from 58.1%) it says the underwriting is still excellent. The bear case requires both to turn; either one alone is noise.

6 of 10 closed · 11 edges · 7 documents · 1 dossier · worked 22 Sep 2026

Open profile →

Next actionwatch the Q3 2026 annual assumption review, reported with Q3 results around 2026-10-29, against the leverage ratio in the same release. The assumption review has charged net income in two of the last three years and is the single biggest unmodelled item on this name. The leverage ratio was 25.1% against a 25% target at 2026-06-30 — before $437.3m of preferred equity came out of the stack on 2026-09-10 and $500m of subordinated notes stayed in it. Those two numbers, in one release, decide whether the capital rebuild of June-September was an arbitrage or a step down.

6 of 10 closed · 21 edges · 13 documents · 1 dossier · worked 23 Sep 2026

Open profile →

Next actionprice the NVIDIA preferred as a governance fact, not a funding one. NVIDIA paid $2.0bn on 2026-03-31 for stock convertible into ~21.8m Marvell shares at ~$91.84 while the stock trades at $244.25 — an ~$3.3bn unrealised gain for the counterparty, sitting alongside a partnership that routes Marvell's XPUs into NVIDIA's ecosystem and a 10-K that lists NVIDIA among the companies that "compete directly with our businesses". Whether that shareholding constrains Marvell's freedom to sell against NVIDIA is the question the filings do not answer and the next transcript read might.

6 of 10 closed · 14 edges · 6 documents · 1 dossier · worked 21 Sep 2026

Open profile →

Next actionon the Q3 2026 release (~2026-10-21) read book value per share first, then EAD against the $0.75 dividend. The stock fell 7.9% from $23.29 to $21.44 between 2026-08-20 and 2026-09-22 with no company filing to explain it beyond the Series I redemption notice; if the book fell with it, the falsifier ($18.45) is one more basis shock away. Then check how the $442.5m Series I redemption on 2026-10-01 was funded.

9 of 10 closed · 23 edges · 11 documents · 1 dossier · worked 23 Sep 2026

Open profile →

Next actionat the 2026-12-01 Q2 call, test the pull-forward question — product GM, inventory and purchase commitments against the 1H/2H "50/50" claim

7 of 10 closed · 66 edges · 14 documents · 1 dossier · worked 20 Sep 2026

Open profile →

Next actionre-test the cluster pass's central claim. _clusters/ai-capex-2026-09-20.md concluded the top layer "is funded from operating cash flow rather than debt." NVDA issued $25.0bn of notes in June 2026 and long-term debt went $7.5bn → $32.4bn in two quarters. The claim needs restating for the sellers, not just the buyers.

5 of 10 closed · 28 edges · 7 documents · 1 dossier · worked 21 Sep 2026

Open profile →

Next actionfind out what moved the stock on 2026-09-18. The Mexican exchange queried a breach of its traded-volume parameter; OMA answered that it knows of no cause, that no director or relevant officer traded, and that the buyback was not operating. The ADR traded 252,609 shares that day against a FINRA average of 75,452 — about 3.3x normal. A 29.9% shareholder group (VINCI, via SETA and CONCESSOC) sits behind a 12.9% control block; block movement is the obvious hypothesis and the public record does not confirm it. Watch for a 13D/G amendment or a follow-up 6-K.

5 of 10 closed · 13 edges · 6 documents · 1 dossier · worked 21 Sep 2026

Open profile →

Next actionread the DEF 14A (2025-09-26) for Ellison's beneficial ownership percentage and the compensation structure — this pass has the 50-million-share 10b5-1 plan but not the denominator it is struck against. Second: Oracle's October Investor Day, which the CFO twice named as where the gross- margin and free-cash-flow questions would be answered. That is the single highest-information event on this name's calendar and it falls before the next 10-Q.

6 of 10 closed · 29 edges · 8 documents · 1 dossier · worked 23 Sep 2026

Open profile →

Next actionon the Q3 2026 release (late October), reconcile the PROPWR contract book line by line: contracted megawatts at the start of the quarter, gross additions, capacity that came off contract, and the closing total. The 2026-09-22 release headlines a 230 MW addition and reports a 510 MW total against a 350 MW starting point — arithmetic that implies about 70 MW came off contract and is disclosed only as a subordinate clause with no number attached. Second: whether the 1.1 GW of equipment already ordered or delivered has closed any of the gap to the ~510 MW contracted.

4 of 10 closed · 21 edges · 11 documents · 1 dossier · worked 23 Sep 2026

Open profile →

Next actionread the FY2026 10-K (due ~2026-11-05) for how the Amazon warrant is accounted for. Three of the last six names in this rotation turn on the same question and each answered it differently: AMD's OpenAI/Meta warrants, Cerebras' penny warrant amortised as a reduction of revenue ($44.3m in one quarter), NVIDIA's guarantees. Qualcomm's warrant vests against up to $60bn of a customer's purchases, and PwC's sole critical audit matter is already QCT revenue recognition with customer incentives as variable consideration. If the 25m shares are charged against the very revenue line the $5bn FY27 / $15bn FY29 data-centre targets are measured on, those targets and the reported revenue that tests them are not the same number — and the ACIP pays management on Adjusted revenue and Adjusted operating income, with no capital-returns or dilution metric anywhere in the plan.

6 of 10 closed · 29 edges · 12 documents · 1 dossier · worked 22 Sep 2026

Open profile →

Next actionInvestor Day 2026-10-06 — read FY28–30 targets and non-seat ACV mix against the 17% FY27 guide

9 of 10 closed · 70 edges · 22 documents · 2 dossiers · worked 20 Sep 2026

Open profile →