← All companies

ACGLArch Capital Group Ltd.

Industry Insurance Updated 24 Sep 2026

The idea

8.6x 2026E earnings with book compounding and shrinking the share count into a soft market; underwriting remains excellent on a current-accident-year basis.

Events to watch

  • Q3 2026 10-Q (late Oct): Note 5 prior-year development by tail length — long-tail adverse development was $32m in Q2
  • buyback pace ($1.95bn in H1)
How we got the target

Scorecard: base ~$105 (~1.14x modelled 2028 BVPS of ~$92). Time frame: 2028 model line.

Target: from the research notes. Prices: market data get_latest_stock_price, close of 2026-09-22 (fetched 2026-09-23)

What would prove it wrong

7 warning signs
  1. Long-tailed prior-year development exceeds $75m adverse in a single quarter, or exceeds the short-tailed release in any quarter.

    Details

    Q2 2026 was $32m adverse against $201m released; H1 2026 was $48m against $437m. The ratio, not the net number, is the series to watch.

  2. Group ex-cat ex-PYD combined ratio crosses 85%.

    Details

    It was 80.9% (Q2 2025) → 82.5% (Q2 2026). Two more years at 1.6 points a year gets there.

  3. Net premiums written decline more than 10% year on year for three consecutive quarters while the buyback continues at the H1 2026 pace.

    Details

    Shrinking the book is the right answer to a soft market; shrinking it while adding $1.56bn of senior notes to buy stock at 1.41x book is a different bet, and it is the one being made.

  4. Senior notes exceed $5bn, or equity declines for two more consecutive quarters.

    Details

    Equity went $24,206m → $24,030m over H1 2026 on $2,104m of earnings.

  5. The Middle East / political-violence book is quantified and the number is larger than one quarter's insurance-segment cat load.

    Details

    — or a second year passes with the war live and no segment figure disclosed. Arch leads this market at Lloyd's, has booked losses from it in two consecutive quarters, says "there could be more", and has declined to size it to two analysts in one call. The Q3 2026 10-Q or release is where it settles.

  6. Buybacks continue above 1.6x book, or are funded by new debt for a third consecutive quarter.

    Details

    Management's own rule (call 2026-07-29 [51]) puts buybacks at ~1.45-1.6x book and special dividends at 2x. Q2 2026 was bought at $96.77 against $68.04 of book = 1.42x, with $1,977m of new borrowings raised in H1 and $2bn of senior notes issued in May. The rule is testable; breaking it is the signal, not the buyback itself.

  7. The FY2026 catastrophe load is reported against the 6-8% range rather than the 7-8% one guided in February.

    Details

    — i.e. the widening is made permanent in the 10-K. A planning range that moves toward the outcome after the outcome is known stops being a planning range.

How far along

7 of 10 steps done · 3 done in a group review

Next step

on 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.

Needs your decision

S6 product test / expert call, S7 conviction.

Failed checks

none. S0/S4/S7 are not closed by design (cluster stages, carried forward). S3 is closed with the transcript deviation named above — and that deviation is now partly closed: the 2026-09-24 backfill read the two most recent calls out of the archive, so the S3 transcript gap is narrowed to the earlier quarters. S8 is closed reliable with one withdrawn row (the cat load quietly widened from 7-8% to 6-8%), S2's capital-allocation verdict is compounds and S3's earnings-quality check carries 2 flags.

Who runs it

The Front Office

Step by step

  1. S0How it does as AI gets cheapDone in a group review

    carry-forward (cluster: Insurance — Stage 0 re-score, 2026-09-22)

    RE-SCORED 2026-09-22 under the SKILL.md §4.3 risk-carrying variant, which supersedes the.

    ⚠️ 2/4 this line used to carry from Insurance & reinsurance — cluster pass, 2026-09-22. lens 4/4: reserve honesty 1 (favourable development 5 of 5 years, −355/−769/−538/−507/ −600 $m), edge survives the cycle 1 (group combined ratio plus development added back: 89.6/89.6/83.6/85.9/86.3, below 95 in 5 of 5), capital theirs 1 (NPW/equity 0.68x, ceded share flat at 27-29% while gross grew 79%), no dilution 1 (reserves $17.8bn → $33.5bn on diluted shares 400.3m → 375.9m). Glyph records the lens ran, not its verdict (owner decision 2026-09-22); the score is in this detail.

    ⚠️ CARRY THE COMPOSITION, NOT JUST THE TOTAL: the favourable development is mostly MORTGAGE — 64%/56%/39% of the group total in 2023/2024/2025 — and the mortgage loss ratio has already gone through zero (−1.2% → 6.5%). Long-tailed lines were adverse in all three disclosed years (+93/+60/+82 $m). Owner checks dropped 2026-09-21 (README §Stage 0 override). Liquidity: ADV $349.7m/day (21 sessions to 2026-09-21, market data NASDAQ:ACGL) — no constraint at any plausible size.

  2. S1Where the idea came fromDone

    watchlist (owner, Apple Stocks, 2026-09-20). No Tier 0 promotion; first member of the first insurance name slot.

  3. S2Quick deal-breaker checkDone

    FY2025 revenue $19,929m (+14.3%), net income $4,399m, net premiums earned $17,065m, shareholders' equity $24,206m, reserves $33,547m (XBRL). Five-year share count 406.7m (2020) → 359.0m (2025) → 340.9m (2026-06-30): −16% and accelerating.

    Full notes

    No cash-runway question; interest paid $65m against $2.1bn of half-year net income. Survives comfortably.

    ⚠️ but the direction has turned on three lines at once: Q2 2026 gross premiums written −1.1%, net written −6.9%, net earned −8.1%, underwriting income −19.7%, and the combined ratio EXCLUDING cats and prior-year development 82.5% against 80.9%. Equity FELL $176m over the half on $2,104m of net income because $1,949m went to buybacks. · capital allocation (2026-09-24, dossier-2026-09-24.md): compounds. FY2021-25 operating cash flow $25,835m; $3,733m of buybacks and a single $1,866m special dividend returned = 21.7% of it, the rest retained into a book that doubled (NPE $8,082m to $17,065m). ROE 16.7/11.6/29.7/22.8/20.1%, five-year average 20.2%; BVPS $30.31 (2020-12-31) to $65.11 (2025-12-31) = +16.5%/yr, +18.3%/yr with the $5.00 special added back. Diluted shares 400.3m to 375.9m (-6.1%), 340.9m at 2026-06-30; SBC $548m over five years is 2.1% of OCF. One acquisition (MCE, a net cash INFLOW of $852m in FY2024), goodwill impairment $2m in five years. Buyback price: ~$87 implied for FY2025, $90.07 average in Q4 2025 (10-K Item 5), $96.77 for Q2 2026's 12.4m shares = 1.34x-1.42x book. Management's stated rule is a book multiple: buybacks at ~1.45-1.6x, special dividends at 2x (call 2026-07-29 [51]).

    ⚠️ XBRL CommonStockDividendsPerShareDeclared reports $5.00 for CY2025 as well as CY2024; the FY2025 10-K cash-flow statement shows common dividends paid of $7m in 2025 against $1,866m in 2024. Use the cash-flow line.

  4. S3Reading the company's reportsDone

    (gate-minimum)

    FY2025 10-K risk-factor summary, Item 1A ordering and PwC's report.

    Full notes

    Q2 2026 10-Q Note 5 (reserve roll-forward and the PYD-by- tail-length table), Note 3 (buyback), Note 9 (senior notes) and the cash flow statement; the 2026-07-28 results 8-K; the 2026-03-24 DEF 14A ownership table; and a filing-index red-flag sweep (no NT 10-K, no UPLOAD/CORRESP, no 10-K/A or 10-Q/A anywhere in the index. PwC, PCAOB 238; sole CAM is reserve valuation). Three disclosed risks named with locators, all from management's own ordered summary.

    ⚠️ Deviation: no earnings-call transcript was read — the Equibles connector failed to connect for this session and fool.com has no Q2 2026 ACGL transcript (404 on both plausible URLs). Management commentary comes from the 8-K release, which is primary and locatable, plus a summary-kind read of the Q1 2026 fool.com transcript that is marked as such and is not quoted anywhere. EQ: 2 flags (2026-09-24, dossier-2026-09-24.md) — "transaction costs and other" added back for a third year at a steady $26m/quarter ($6m FY2023, $81m FY2024, $75m FY2025), which is mild; and 12.1% of FY2025 pre-tax income coming from prior-year releases ($600m of $4,979m; 11.3% FY2024, 15.9% FY2023) that are increasingly short-tail and mortgage-sourced while long-tail lines run adverse. Clean: SBC 0.74% of revenue and 2.4% of OCF; OCF/net income 1.29-1.55x with accruals negative in all three years; premiums-receivable days 136 to 122; DAC DOWN 1.0% on earned premium up 13%. The non-GAAP bridge is conservative in the rare direction — its largest adjustment REMOVES equity-method income ($504m FY2025, 340bp of ROE).

  5. S4Rivals and suppliersDone in a group review
  6. S5Who owns the stockDone

    13F by manager at 2026-06-30 (BlackRock 30.2m sh / $2,933m, State Street 16.3m, BAMCO/Baron 15.9m), 13D/G census, proxy group ownership (17 officers and directors = 3.3%, every individual under 1%), the $3.0bn authorisation added 2026-04-19 with $2.2bn remaining.

    Full notes

    Who is on the other side: index money plus Baron, no strategic block. Shelf: none read this run (no S-3 in the recent index).

    ⚠️ the Q2-26 13F quarter is incompletely ingested (609 filers vs 898) — both Vanguard entities, Artisan, WCM, Geode, JPMorgan, Northern Trust, Morgan Stanley and Dimensional have no filing in it, so no "exit" may be read from an absence.

    ⚠️

    ⚠️ the Sixth Street row is NOT a holding — see the log.

  7. S6What workers and customers sayDone

    carry-forward (cluster: Insurance & reinsurance — cluster pass, 2026-09-22)

    named in the 2026-09-07 Real Eisman Playbook reinsurance-consolidation passage, plus a Jefferies downgrade to Hold, $100 PT. Product test / expert call open OPEN (user)

  8. S7The case for it, and pricesDone in a group review

    carry-forward (cluster: Insurance & reinsurance — cluster pass, 2026-09-22)

    conviction open OPEN (user) · valuation (bear/base/bull; moved from S8 2026-09-23): 8.6x 2026E and 9.1x 2028E model EPS at $95.78 (2026-09-21); 1.41x book ($68.04 BVPS at 2026-06-30, 1.04x the modelled 2028 BVPS of ~$92)

    Full notes

    Base ~$105, bear ~$77, bull ~$130.

  9. S8Do the bosses keep their word?Done

    7 rows (2026-09-24, dossier-2026-09-24.md): 2 beat, 4 met, 1 withdrawn

    reliable. All 7 belong to the current team (Papadopoulo CEO from 2024-10-13, 8-K 0000947484-24-000123. Morin EVP/CFO throughout); no departed rows.

    Full notes

    Arch guides almost nothing — no revenue, EPS, margin or ROE guidance in eight calls, and capital return is refused in terms ("We don't set targets and how much we're going to buy back", call 2026-04-29 [32]). What it does guide are expense lines and loss loads, and it hits them: amortisation $99m against ~$100m and $193m against ~$195m; the wildfire net loss inside $450-550m; the FY2025 tax rate 14.9% against a guided 16-18%. MCE first-year premium >$1.5bn as promised.

    ⚠️ The one withdrawn row is the informative one: the FY2026 group cat load was guided at 7-8% on the call of 2026-02-10 [4] and described on 2026-07-29 [33] as "the 6-8% ... that has not changed" — a range widened downward while being called unchanged. Consequence for S7: no guidance haircut (there is no guidance to haircut), but model the cat load at the TOP of whatever range is current, not the midpoint.

  10. S9What to watch nextDone

    trigger written; next earnings expected late October 2026 (DERIVED from filing cadence: Q3 releases went 2025-10-27 and 2024-10-28. market data's calendar needs a paid plan and Equibles was unreachable)

    Full notes

    Tier 0 EDGAR sweep covers the feed (cik 947484)

Sources and notes

Sources 8 documents
Short nameDocumentWhere it came fromRead on
10K-2025Arch Capital 10-K for FY2025, filed 2026-02-26https://www.sec.gov/Archives/edgar/data/947484/000094748426000017/acgl-20251231.htm2026-09-22
10Q-Q2-26Arch Capital 10-Q for the quarter ended 2026-06-30, filed 2026-08-04https://www.sec.gov/Archives/edgar/data/947484/000094748426000124/acgl-20260630.htm2026-09-22
8K-Q2-26Arch Capital 8-K filed 2026-07-28, Exhibit 99.1 — 2026 second quarter results releasehttps://www.sec.gov/Archives/edgar/data/947484/000094748426000118/ex-991release63026.htm2026-09-22
DEF14A-26Arch Capital DEF 14A filed 2026-03-24 (ownership table as of 2026-03-09)https://www.sec.gov/Archives/edgar/data/947484/000094748426000038/acgl-20260324.htm2026-09-22
13F-Q2-26Institutional holders by manager, 13F report date 2026-06-30 (ingestion incomplete — see the edge note)python ledger/research/tools/threespread.py holders ACGL2026-09-22
SIXTHST-13FSixth Street Partners Management Company 13F-HR for 2026-06-30, accession 0001752724-26-000051 — filing indexhttps://www.sec.gov/Archives/edgar/data/1812095/000175272426000051/index.json2026-09-22
SIXTHST-13FASixth Street Partners Management Company 13F-HR/A (RESTATEMENT) for 2026-06-30, accession 0001752724-26-000055https://www.sec.gov/Archives/edgar/data/1812095/000175272426000055/primary_doc.xml2026-09-22
CLUSTER-INSInsurance & reinsurance cluster pass, 2026-09-22, slot cluster-insuranceledger/research/_clusters/cluster-insurance-2026-09-22.md2026-09-22
Links to other companies 11 links
context 3issuer of 2officer of 1auditor of 1holds 1litigation watch 1governance structure 1competitor 1
FromLinkToDateWhere it says so
ACGLcontextACGL2026-06-3010Q-Q2-26 — Note 5, Reserve for Losses and Loss Adjustment Expenses — table 'summarizes (favorable) and adverse net PYD by segment and tail length'
ACGLissuer ofArch Capital senior notes (carrying value $4.3bn at 2026-06-30)2026-06-3010Q-Q2-26 — Note 9, Fair Value — senior notes carrying value paragraph
ACGLissuer ofArch Capital share repurchase program2026-06-3010Q-Q2-26 — Note 3, Shareholders' Equity — share repurchase program paragraph
Nicolas Papadopouloofficer ofACGL2026-07-288K-Q2-26 — Exhibit 99.1, CEO quote following the results bullets
ACGLcontextACGL2026-02-2610K-2025 — Item 1A, RISK FACTORS SUMMARY, second bullet under 'Risks Relating to Our Industry, Business and Operations'
PricewaterhouseCoopers LLPauditor ofACGL2025-12-3110K-2025 — Item 8, Report of Independent Registered Public Accounting Firm — Critical Audit Matters heading
BlackRock, Inc.holdsACGL2026-06-3013F-Q2-26 — holders table, row 1; WARNING line and 'Prior holders with NO 13F ingested' line beneath it
Sixth Street Partners Management Company, L.P.litigation watchACGL2026-08-14SIXTHST-13F — Archives directory listing for accession 0001752724-26-000051 — the information-table filename
Sixth Street Partners Management Company, L.P.contextCharles Schwab Investment Management, Inc.2026-08-14SIXTHST-13FA — primary_doc.xml, summaryPage — tableEntryTotal and tableValueTotal
ACGLgovernance structureFrançois Morin2026-03-09DEF14A-26 — Security Ownership table, group row
ACGLcompetitorKinsale Capital Group, Inc.2026-06-30CLUSTER-INS — §1 — The cycle, from a filing rather than from commentary; §5 cluster thesis
History 4 entries
  • 2026-09-22 — first pass, slot names-insurance-1, model-drafted unattended by the cloud research routine. S1, S2, S3 (gate-minimum, no transcript — Equibles down, fool.com 404), S5 and S9 closed; S0/S4/S6/S7 carried forward from Insurance & reinsurance — cluster pass, 2026-09-22; S8 modelled and left open. Run finding: the prior-year-development split by tail length confirms the cluster hypothesis — the release is short-tail property, the long tail is already adverse and worsening ($1m favourable → $32m adverse in the quarter, $5m → $48m adverse over the half). Second finding: Arch raised $1,977m of borrowings in H1 2026 (zero in H1 2025), took senior notes from $2.7bn to $4.3bn, and spent $1,949m on buybacks at ~$92 against $68.04 of book. Tooling finding affecting every S5 in this repo: the "Sixth Street Partners Management Company" row that 3spread reports for ACGL, KNSL, AX and QCOM is Charles Schwab Investment Management's position counted twice. Sixth Street's 13F-HR accession 0001752724-26-000051 has Schwab's information table (13F_0000884546_20260630.xml) attached to it; Sixth Street restated the same day (0001752724-26-000055, 9 holdings, $579,182,557, none of them these names) and 3spread has not ingested the restatement. Discard the row. Evidence and access limits in dossier-2026-09-22.md. Nothing here is human-verified.
  • 2026-09-22 — S0 re-scored, slot cluster-insurance-rescore, model-drafted unattended. The S0 line above now points at Insurance — Stage 0 re-score, 2026-09-22 instead of Insurance & reinsurance — cluster pass, 2026-09-22, which scored this name ⚠️ 2/4 with the AI-surplus four and a reworded fourth test. Under the instrument built for risk carriers (SKILL.md §4.3, owner decision 2026-09-22) the same evidence reads 4/4. Nothing else in this scorecard changed — S4, S6 and S7 still carry forward from the 2026-09-22 cluster file, whose industry stages stand. New fact the re-score produced, and the reason the ⚠️ above is in the S0 detail: the FY2025 10-K's prior-year-development table by segment and tail length shows the mortgage segment supplying $344m of $538m (2023), $282m of $507m (2024) and $235m of $600m (2025) of the group's favourable development, while long-tailed lines were adverse in every one of those years.
  • 2026-09-23 — stage redefinition, owner decision (WATCHLIST_PLAN.md §25): S8 is now Management credibility and is not yet run for this name; the old S8 valuation figures moved to the valuation: line under S7, and any proposed buy price or position size was removed (the pipeline no longer drafts either). No research was redone. S2 capital allocation and S3 earnings quality are also pending, in the same backfill slot.
  • 2026-09-24 — backfill, slot backfill-2026-09-23-5, model-drafted unattended by the cloud research routine. The three checks added on 2026-09-23 are now written: S8 management credibility (7 rows, 2 beat / 4 met / 1 withdrawn — reliable), S2 capital allocation (compounds) and S3 earnings quality (2 flags). Evidence: EDGAR only (FY2025 10-K, four quarterly 8-K Ex. 99.1 releases, five Item 5.02 8-Ks, XBRL companyfacts) plus the eight archived market data transcripts. 0 Equibles calls, 0 market data calls, 0 3spread calls. No other stage was redone. Run finding 1: Arch issues no revenue, EPS, margin or ROE guidance at all, and refuses capital- return targets in terms — so S8 had to be scored on expense lines, loss loads and one integration commitment. It hits them; the record is clean and it is a record of arithmetic, not of performance. Run finding 2: the FY2026 catastrophe load was guided at 7-8% of net earned premium on 2026-02-10 and described five months later as "the 6-8% ... that has not changed". That is the only withdrawn row and it is the one that touches reported earnings. Run finding 3 (tooling, affects any screen reading XBRL): ACGL tags CommonStockDividendsPerShareDeclared at $5.00 for CY2025 as well as CY2024, while the FY2025 10-K cash-flow statement shows $7m of common dividends paid in 2025 against $1,866m in 2024. The per-share tag will make Arch look like a dividend payer in a year it paid none. Run finding 4: the non-GAAP bridge runs backwards from the usual direction — its largest adjustment removes $504m of equity-method income from the headline. Nothing here is human-verified.
Full notes, 24 Sep 2026 19 min read

ACGL — dossier, 2026-09-24 (backfill: S8, S2 capital allocation, S3 earnings quality)

Slot backfill-2026-09-23-5. Model-drafted unattended by the cloud research routine; nothing here is human-verified. This file holds only the three checks added on 2026-09-23 (WATCHLIST_PLAN.md §25). Every other stage stands as written in dossier-2026-09-22.md. Anything from background knowledge is marked [background].

Sources this pass. EDGAR, User-Agent ddboard-research gorkem@celebico.com (CIK 947484): the FY2025 10-K (0000947484-26-000017, filed 2026-02-26) for the consolidated statements, MD&A and Item 5; the quarterly earnings releases 8-K Ex. 99.1 of 2025-04-29 (Q1 2025), 2026-02-10 (Q4/FY 2025), 2026-04-28 (Q1 2026) and 2026-07-28 (Q2 2026); the Item 5.02 8-Ks of 2024-10-15, 2024-11-08, 2025-02-06, 2026-01-22 and 2026-06-03; XBRL companyfacts for CIK 947484. Transcripts entirely from the archive ledger/research/_transcripts/ACGL/ (market data, eight quarters FY2024-Q3 through FY2026-Q2, call dates from _transcripts/index.json). Equibles: 0 calls. market data: 0 calls. 3spread: 0 calls. Nothing was fetched from the open web.

Quote discipline: every transcript quote was located by exact substring in the archived JSON bytes and copied from it; filing quotes were matched against a whitespace-collapsed text extraction of the EDGAR HTML. Turn numbers [n] are newline positions in the market data content string.

Scoring convention used below. For a cost or loss guide (cat load, amortisation, corporate expense, tax rate), landing below the guided figure is scored beat, because that is the favourable direction for earnings; landing above is missed. It is stated here because Arch guides almost nothing else, so the convention decides most of the table.

S8 — management credibility

How Arch guides: it mostly does not. There is no revenue, EPS, margin or ROE guidance in any of the eight calls or any earnings release in the window. What management does put a number on is a small set of planning assumptions and expense lines — the group catastrophe load as a percentage of net earned premium, intangible amortisation, corporate expenses, the effective tax rate, interest expense — plus discrete event estimates (the California wildfire loss) and one acquisition-integration commitment (MCE premium volume). Capital return is explicitly refused as a commitment, twice and in the same words:

We don't set targets and how much we're going to buy back.

— François Morin, call of 2026-04-29 [32], answering Cave Montazeri.

We do not certainly do not have targets or plans to buy back a certain number or dollars of shares

— François Morin, call of 2026-07-29 [8], answering Elyse Greenspan.

A management that commits to nothing checkable is itself the finding (SKILL.md §S8); Arch is a partial case — it commits to little, and what it does commit to it hits.

Who promised, and who is still here. market data labels turns by name only; roles are read off EDGAR as of each call date.

  • Nicolas Papadopoulo, Chief Executive Officer — appointed effective 2024-10-13, 8-K 0000947484-24-000123 Item 5.02: "the Board approved the appointment of Nicolas Papadopoulo as Chief Executive Officer of the Company and as a member of the Board, effective October 13, 2024." Still in the role (named as CEO in the 8-K of 2026-06-03). Made rows 3 and 5 below.
  • François Morin, Executive Vice President, Chief Financial Officer and Treasurer — across the entire window; his title is read from his signature block on the 8-K of 2026-01-22 (0000947484-26-000004). He made or co-made every row below.
  • Marc Grandisson, CEO until 2024-10-15 (same 8-K: "Mr. Grandisson will cease to be Chief Executive Officer of the Company and a member of the Board effective October 15, 2024"). He made none of the scored rows — the earliest is the call of 2024-10-31, after he had gone.
  • David Gansberg, President — stepped down 2026-06-03 (8-K 0000947484-26-000094 Item 5.02), with Maamoun Rajeh taking the insurance segment "as the Company moves forward under a single President model." He made none of the scored rows.

So every row belongs to the current team, and there are no departed rows. That is unusual and it cuts both ways: there is no old-team record to discount, and no long record either.

Say vs do (guides made 2024-10-31 → 2026-07-29)

#Promise (period · metric · figure)Made where · by whomResult · locatorScore
1Q4 2024 · group amortisation of intangibles ≈ $100mcall 2024-10-31 [2] · François Morin (EVP & CFO): "We expect our overall amortization expense across the group to be approximately $100 million in the fourth quarter of this year and $195 million in 2025, spread evenly throughout the four quarters."$99m — release 2026-02-10 (Ex. 99.1), Q4 2024 comparative column of the segment reconciliation, "Amortization of intangible assets (99)"met
2FY2025 · group amortisation of intangibles ≈ $195msame sentence, same call · Morin$193m — FY2025 10-K, Consolidated Statements of Income, 2025 column: "Amortization of intangible assets 193"met
3Q1 2025 · California wildfires · net loss $450–550mcall 2025-02-11 [1] · Nicolas Papadopoulo (CEO): "As noted in yesterday's press release, we expect the wildfires to result in a net loss between $450 million and $550 million based on an industry loss estimate of $35 billion to $45 billion."release 2025-04-29 (Ex. 99.1): "Pre-tax current accident year catastrophic losses for the Company's insurance and reinsurance segments, net of reinsurance and reinstatement premiums, of $547 million, primarily related to the California wildfires." The wildfire share is therefore at or below $547m — inside the range, at its topmet
4FY2025 · group catastrophe load ≈ 7–8% of full-year group-wide net earned premiumcall 2025-02-11 [2] · Morin: "we expect our cat load to represent approximately 7% to 8% of our full-year group-wide net earned premium."≈6.0%. FY2025 10-K MD&A: insurance segment "included 4.4 points of current year catastrophic event activity"; reinsurance segment "8.5 points for current year catastrophic event activity". Applied to segment net premiums earned of $7,771m and $8,122m (10-K segment note) that is ~$342m + ~$690m = ~$1,032m on group net premiums earned of $17,065m = 6.05%beat (below the load — but see the caveat under the tally)
5MCE (MidCorp & Entertainment) · > $1.5bn of annualised premium in the first year of ownershipcall 2025-07-30 [2] · Morin: "We are now on track to write slightly more than $1.5 billion of annualized premium for the first year of owning this business, which is slightly higher than the forecast at the time of the acquisition."confirmed on the next call, 2025-10-28 [59] · Morin, sizing a program run-off against "$1.5 billion to $1.6 billion book is -- which was the overall MCE premium volume". The first year of ownership ended 2025-07-31 (acquisition closed 2024-08-01, release 2026-02-10)met
6FY2025 · effective tax rate on pre-tax operating income 16–18%the standing guide, restated and scored by management itself on the call of 2026-02-10 [4] · Morin: "For the 2025 year, our effective tax rate on pretax operating income was 14.9% reflecting the mix of income by tax jurisdiction. It was slightly below the 16% to 18% previously guided range mostly due to a 1.4% benefit from discrete items."14.9%, i.e. outside the guided range on the favourable side — and management said so unpromptedbeat (outside the range, favourable)
7FY2026 · group catastrophe load 7–8%call 2026-02-10 [4] · Morin: "For 2026, our current estimate of the full year catastrophe losses stands within a range of 7% to 8% of ov[erall net earned premium]"redefined mid-year without acknowledgement. Call 2026-07-29 [33] · Morin, to David Motemaden: "Cat load going forward, I mean, we quoted the 6-8% kind of for the on an annual basis for you know, the group, that has not changed." The range quoted in February was 7–8%; the range described in July as unchanged is 6–8%withdrawn

Not scored (open).

  • FY2026 corporate expenses "approximately between $80 million and $90 million" (call 2026-02-10 [4], Morin). H1 2026 actual is $43m — $31m in Q1 (release 2026-04-28: "Corporate expenses for the 2026 first quarter were $31 million") plus $12m in Q2 (release 2026-07-28: "Corporate expenses for the 2026 second quarter were $12 million"). On the guided first-quarter skew that is on track and inside the range; it comes due with the FY2026 10-K.
  • FY2026 reinsurance-segment operating expense ratio "between 3.9-4.5%" (same call). Not checked — the quarterly releases give the combined underwriting expense ratio (22.9% in Q2 2026), not the operating component; it needs the 10-Q segment table.
  • FY2026 effective tax rate back in the 16–18% range (same call). Comes due with the FY2026 10-K.
  • Interest expense "approximately $60 million to $63 million for each of the next two quarters" (call 2026-07-29 [2], Morin) — Q3 and Q4 2026.
  • Insurance-segment operating expense ratio to "revert to historical levels during the second half of the year" (call 2026-07-29 [2], Morin) — H2 2026.
  • Uses of the May 2026 $2bn senior-note raise (same turn): redeem "the $500 million of 10-year senior notes maturing later this year" (not yet due) and "purchase $418 million of our 2043 and 2046 senior notes" through a tender offer described as "recently completed" (done).

Tally: 7 rows — 2 beat, 4 met, 1 withdrawn. All 7 belong to the current team; 0 departed rows.

Two caveats sit on top of that tally and they matter more than the count:

  1. Row 4 is weather, not execution. A catastrophe load is a planning assumption about events nobody controls. Coming in 100–200bp under it in a year whose only major event was the California wildfires says the year was quiet, not that management delivered. The row is scored because it is a number they put on the record; it is worth less than the others.
  2. Nothing in this table is an earnings promise. Arch guides expense lines and loss loads and refuses to guide anything the market scores it on. The record is clean, and it is a record of arithmetic rather than of performance.

Dodge-log roll-up (§3.1). scorecard.md carried no dodge log — the 2026-09-22 pass closed S3 without a transcript (Equibles was down that session). The log below is opened from the two most recent calls, read in this pass.

  1. Quantifying the Iran / Middle East losses — dodged in two consecutive quarters, and twice inside the second call. Q1 2026 (2026-04-29): Michael Zaremski asked "Maybe you can kind of just elaborate on the Iran conflict, how you guys are thinking about that? Is it all IBNR, are there real losses or..." [21]; Morin answered "there's nothing paid, but it's certainly -- there are some real losses, specialty book out of London, like terror, political violence... we took a first stab at it this quarter" [22] — composition, no figure. Q2 2026 (2026-07-29): David Motemaden asked management to "quantify the Iran losses this quarter that impacted the insurance segment... I am interested also in any sort of IBNR versus actual loss detail you could share" [32]; Morin answered with the majority-share statement, the cat load, and reserving philosophy ("case reserves have been set up. it is not a hypothetical IBNR") — no number, no split [33]. Mike Zaremski then came back to it in the same call — "I am not sure if you did quantify the exact cat loss to David's question" [70] — and got the case-by-case answer [71] plus Papadopoulo's industry figure, not Arch's: "I think the industry in general is still around $3 billion for the for the Middle East war losses" [72]. Two analysts, one call, and the segment number was never given. Promoted from a note to a finding.
  2. Notes, not dodges — stated refusals. The buyback-pace question (Greenspan 2026-07-29 [7], Montazeri 2026-04-29 [31] and 2026-07-29 [21]) is answered every time with an explicit "we don't set targets", which is a policy, not an evasion. Same for the mid-year property rate forecast (Greenspan 2026-04-29 [4] → "we don't have a crystal ball... So we don't really have a forecast there" [5]).

Verdict — reliable. Six of seven scored rows landed at or better than the guide, and the company's own CFO volunteered the one figure that fell outside its range. Against that: the single withdrawn row is the informative one, because the cat load was widened from 7–8% to 6–8% inside six months while being described as unchanged, and the cat load is the one number in this table that touches reported earnings directly. That is not enough for over-promises (1 of 7), and it is enough to withhold conservative: a company that guides this little should be scored on whether the few numbers it gives stay put, and one of them did not. Consequence for S7: the base case does not need a guidance haircut — Arch supplies no guidance to haircut — but the catastrophe load used in any forward model should be taken at the upper end of whatever range is current (8%), not at the midpoint, because the range has moved down once already without notice.

S2 — capital allocation

All figures $m, calendar fiscal years. Sources: XBRL companyfacts (CIK 947484) for the five-year series — OCF NetCashProvidedByUsedInOperatingActivities, buybacks PaymentsForRepurchaseOfCommonStock, acquisitions PaymentsToAcquireBusinessesNetOfCashAcquired, SBC AllocatedShareBasedCompensationExpense, preferred dividends PaymentsOfDividendsPreferredStockAndPreferenceStock, goodwill GoodwillImpairmentLoss — and the FY2025 10-K's Consolidated Statements of Cash Flows for the common-dividend line (2023–2025 columns).

FY2021FY2022FY2023FY2024FY20255y total
Operating cash flow3,4253,8165,7496,6736,17225,835
Buybacks1,2345860241,8893,733
Common dividends paid——01,86671,873
Preferred dividends4740404040207
Acquisitions (cash, net of cash acquired)n/t00(852)0(852)
Stock-based compensation878892133148548
SBC as % of revenue0.9%0.9%0.7%0.8%0.7%
Goodwill impairment002002
Diluted weighted-average shares (m)400.3377.6378.8381.8375.9−6.1%

(n/t = not tagged for that year. The FY2025 10-K's cash-flow statement covers 2023–2025 only; Arch paid no common dividend before the 2024 special — the 10-K describes it as the one event: "In 2024, the Board of Directors of Arch Capital (the "Board") has declared and paid a special cash dividend of $1.9 billion to common shareholders, representing $5.00 per outstanding common sh[are]".)

A tagging discrepancy worth carrying. XBRL CommonStockDividendsPerShareDeclared reports $5.00 for CY2025 as well as CY2024 (accessions 0000947484-26-000017 and 0000947484-25-000017). The FY2025 10-K's own cash-flow statement shows common dividends paid of $7m in 2025 against $1,866m in 2024, and its liquidity discussion says "Activity in 2025 consisted of $1.9 billion of share repurchases under our share repurchase program, while activity in 2024 included a $1.9 billion special dividend paid to common shareholders." Use the cash-flow line, not the per-share tag; a screen that reads the tag will show Arch paying a $5.00 dividend in a year it paid none.

Where the cash went. $25.8bn of operating cash flow over five years, against $3.7bn of buybacks and $1.9bn of a one-off special dividend — 21.7% of operating cash flow returned. The rest stayed in the business, which for an underwriter is not hoarding: net premiums earned went from $8,082m (2021) to $17,065m (2025), reserves from $17.8bn to $33.5bn (scorecard.md S0), and equity from $13.5bn to $24.2bn. The retention funded the growth. The pattern changed in 2025 and again in H1 2026: buybacks went $24m (2024) → $1,889m (2025) → $1.95bn in the first half of 2026 alone (call 2026-07-29 [1], Papadopoulo: "Bringing the total for the first half of the year to $1.95 billion"), i.e. "approximately 94% of our net income" (same call [2], Morin). That is the soft market arriving: the capital cannot be deployed into the book, so it is being handed back.

Buybacks at what price. FY2025: $1,889m against a 21.6m increase in treasury shares (219.2m → 240.8m, FY2025 10-K balance sheet) — ≈$87 a share implied, including a small number withheld for taxes. The 10-K's Item 5 table prices the fourth quarter exactly: 8,870,269 shares at an average of $90.07. H1 2026: Q2 alone was "12.4 million shares at an aggregate cost of $1.2 billion" (call 2026-07-29 [2], Morin) = $96.77. Against book value per common share of $65.11 at 2025-12-31 (release 2026-02-10) and $68.04 at 2026-06-30 (scorecard.md S7), Arch is buying at 1.34x–1.42x book, and doing it with $1,977m of new borrowings raised in H1 2026 (scorecard.md log, 2026-09-22) and a $2bn senior-note issue in May. Management's own framing of the buyback-versus-dividend switch is a book-multiple rule: "back in 2024, we are at 2x book, so it was very much a you know, to us was very clear that buybacks did not make sense... Right now, we are trading in the kinda 1.5x, 1.6x range, 1.45x, whatever. So I think it is more, still makes sense to do buybacks" (call 2026-07-29 [51], Morin). That is an explicit, testable discipline, and it is the line to hold them to: the rule says stop somewhere between 1.5x and 2.0x.

What the acquisitions were worth. One acquisition in the window — the U.S. MidCorp and Entertainment businesses from Allianz, closed 2024-08-01. It shows as a net cash inflow of $852m in FY2024 (PaymentsToAcquireBusinessesNetOfCashAcquired = −852), which is normal for a book transfer where the reserves arrive with assets against them, and its transaction costs ran through BusinessCombinationAcquisitionRelatedCosts at $81m (2024) and $75m (2025). Goodwill impairment across the five years is $2m, in 2023 — nothing has been written off. The integration commitment (row 5 above) was met and management calls the cutover finished: "we just finished the cutover... the first goal was to move the business over to Arch. So we did this a year ago" (call 2026-04-29 [39] / [54] of 2026-07-29, Papadopoulo).

Returns (insurer form — ROE and book value per share compounded with dividends, per SKILL.md). Common equity = StockholdersEquity less the $830m of preferred ($780m before 2021); net income available to common = NetIncomeLossAvailableToCommonStockholdersBasic.

FY2021FY2022FY2023FY2024FY2025
Net income to common ($m)2,0941,4364,4034,2724,359
Average common equity ($m)12,52112,39814,80218,75721,683
ROE16.7%11.6%29.7%22.8%20.1%

Five-year average ROE ≈20.2%, and only one year (2022, the Ukraine/Ian year) below 15%. Book value per common share: $30.31 at 2020-12-31 ($12,326m over 406.7m shares) → $65.11 at 2025-12-31 (stated in the release of 2026-02-10). That is +16.5% a year compounded, and +18.3% a year adding back the $5.00 special dividend. Arch's own framing of 2025: "resulted in book value growth for 2025 of 22.6%" (FY2025 10-K).

Dilution. Diluted weighted-average shares 400.3m (2021) → 375.9m (2025), −6.1%, and 340.9m at 2026-06-30 (scorecard.md S2) — −15% on five and a half years. SBC of $548m over the window is 2.1% of operating cash flow and was more than offset every year the company bought back at all. There is no dilution to explain.

Verdict: compounds. 20% average ROE, book value per share up 16.5% a year (18.3% with the dividend), a falling share count, one acquisition that has not been written down, and a stated book-multiple rule governing the buyback. The one thing to watch is in scorecard.md already and is a direction, not a verdict: equity fell $176m over H1 2026 because $1,949m went to buybacks funded in part by new senior notes, while net premiums written shrank. Buying 1.4x book with borrowed money in a softening market is a bet that the softening is shallow; it is the right answer if it is, and it converts a compounder into a levered one if it is not.

S3 addendum — earnings quality

Sources: the non-GAAP reconciliation in the earnings releases of 2026-02-10 (Q4/FY2025) and 2025-04-29; the FY2025 10-K consolidated statements, MD&A and the prior-year-development schedule; XBRL companyfacts.

  1. The non-GAAP bridge. Arch's headline non-GAAP measure is after-tax operating income available to Arch common shareholders. The Q4 2025 reconciliation (release 2026-02-10) runs, per diluted share: net income $3.35 → net realized (gains) losses $(0.06) → equity in net (income) of investments accounted for using the equity method $(0.42) → net foreign exchange (gains) losses $0.02 → transaction costs and other $0.07 → income tax $0.02 → operating income $2.98. This bridge runs the unusual way: it mostly removes income. The largest single adjustment strips out equity-method investment income — $504m in FY2025, $580m in 2024, $278m in 2023 (10-K income statement) — which management itself says "contributed 340 basis points to our annualized net income return on average common equity in the quarter" (call 2026-07-29 [2], Morin). An adjustment that lowers the headline is the opposite of the drift this test hunts for. Clean, and conservative. Flag 1 (mild) — "transaction costs and other" is now in its third year. $6m (2023) → $81m (2024) → $75m (2025) of BusinessCombinationAcquisitionRelatedCosts, running at $26m a quarter in both Q4 2024 and Q4 2025 (release 2026-02-10). Two full years after the MCE close, a line added back as transaction costs at a steady quarterly rate is closer to an operating expense than to a one-off. It is small (0.4% of revenue) and it is going the right way, so it is logged as mild.
  2. Stock-based compensation. Clean. $148m in FY2025 = 0.74% of revenue and 2.4% of operating cash flow. Against $1,889m of buybacks in the same year it is not a tenth of the capital returned, and the diluted share count fell. Nothing here.
  3. Cash conversion. Clean. Operating cash flow over net income: 1.29x (2023), 1.55x (2024), 1.40x (2025). Accruals ((net income − OCF) / average total assets): −2.5%, −3.6%, −2.4% — negative in all three years, i.e. cash runs ahead of reported profit, which is what a growing underwriter's float should do.
  4. Working capital. The insurer reading is premiums receivable against earned premium. PremiumsReceivableAtCarryingValue $4,644m (2023) → $5,634m (2024) → $5,723m (2025), against net premiums earned of $12,440m → $15,100m → $17,065m. Days: 136 → 136 → 122, i.e. receivables grew 1.6% while earned premium grew 13%. Clean, improving.
  5. Costs moved onto the balance sheet. Clean. Deferred policy acquisition costs $1,531m (2023) → $1,734m (2024) → $1,717m (2025): down 1.0% on earned premium up 13%. DAC is not being used to hold costs off the income statement. No capitalised-software line is separately tagged.
  6. Sector equivalent — prior-year reserve development as a share of pre-tax income. Flag 2, and it is the one that matters. Favourable development (XBRL SupplementalInformationForPropertyCasualtyInsuranceUnderwritersPriorYearClaimsAndClaimsAdjustmentExpense, negative = favourable) against income before income taxes (FY2025 10-K income statement: $4,979m / $4,474m / $3,385m for 2025/2024/2023; earlier years from XBRL):
FY2021FY2022FY2023FY2024FY2025
Favourable PYD ($m)355769538507600
Pre-tax income ($m)2,3681,5623,3854,4744,979
PYD as % of pre-tax income15.0%49.2%15.9%11.3%12.1%

Favourable in five of five years and the share is falling, which reads well — but the composition does not, and it is already the scorecard's ⚠️: the mortgage segment supplied $344m of $538m (2023), $282m of $507m (2024) and $235m of $600m (2025), the mortgage loss ratio has gone through zero (−1.2% → 6.5% in Q2 2026, release 2026-07-28), and long-tailed lines were adverse in each of those three years. Q2 2026 says it out loud from the other side: 5.3 points of favourable development in the reinsurance segment alone (release 2026-07-28) in a quarter whose group ex-cat ex-PYD combined ratio was 82.5% against 80.9% a year before. Management's own description of where it comes from: "most of the favorable is in the short-tail lines in the last 2 to 3 accident slash underwriting years" and "Casualty at a high level is kind of neutral" (call 2026-07-29 [19], Morin). The reported combined ratio is being held down by releases on business written two or three years ago in a hard market; the underlying number is 1.6 points worse year on year. That is not an accounting flag — the releases are real — it is a durability flag, and it is exactly the series scorecard.md's ➡️ next action already points at.

EQ summary: 2 flags — transaction costs added back for a third year at a steady quarterly rate (mild), and reported earnings carrying 12% of pre-tax income from prior-year releases that are increasingly short-tail and mortgage-sourced while long-tail lines run adverse. Everything the test normally catches — SBC, accruals, receivables, DAC, the non-GAAP bridge — is clean, and the bridge is conservative in the rare direction of removing income rather than adding it back.

Full notes, 22 Sep 2026 14 min read

ACGL — Arch Capital Group Ltd. — dossier, 2026-09-22

Slot names-insurance-1. Model-drafted unattended by the cloud research routine (Opus 5), following .claude/skills/stock-sourcing-pipeline/SKILL.md. Nothing here is human-verified. Sourced facts carry the document they were read from; anything from background knowledge is marked [background]; anything from a search-result summary is marked [search-summary]. Private: this directory never reaches dist/.

Access this run

RungState
1. SEC EDGARReachable. Carried essentially the whole pass — submissions index, XBRL company concepts, and four documents read in full.
4. EquiblesUNREACHABLE. The connector failed to connect for this session (equible (502) … CLIENT_HTTP_NOT_IMPLEMENTED). No transcript, no short interest, no valuation multiples from this rung. 0 of the 25-call budget spent — because none could be.
4b. 3spreadReachable (key present, HTTP 200). Used for 13F holders and 13D/G. See the defect below.
5. market dataReachable for prices; the earnings calendar still needs a paid plan.
6. WebSearchReachable.
7. www.fool.comReachable, but has no Q2 2026 ACGL transcript — 404 on /2026/07/28/ and /2026/07/29/. Only the Q1 2026 transcript exists, and WebFetch returns a model summary of it rather than a locatable span, so it is used as [search-summary] and quoted nowhere.

§1 — The finding: the release is short-tail, the long tail has already turned

The 2026-09-22 cluster pass ended with this instruction: "Its S3 gate-minimum read should start with Arch's Q2 10-Q reserve roll-forward, because the ex-cat ex-development gap identified in §1 is the thing a name pass can either confirm or kill." It is confirmed, and the 10-Q gives a split the cluster pass did not have — prior-year development by tail length.

Note 5 of the Q2 2026 10-Q, table headed "summarizes (favorable) and adverse net PYD by segment and tail length", $ millions, negative = favourable:

Q2 2026 shortQ2 2026 longQ2 2026 totalQ2 2025 shortQ2 2025 longQ2 2025 total
Insurance(33)6(27)(13)5(8)
Reinsurance(123)26(97)(75)(6)(81)
Mortgage(45)—(45)(64)—(64)
Total(201)32(169)(152)(1)(153)

Six months: 2026 short (437) / long 48 / total (389); 2025 short (355) / long 5 / total (350).

Read it in the order that matters:

  1. The headline improved. Net favourable PYD went $153m → $169m in the quarter, $350m → $389m over the half. On that number alone, reserving looks fine and slightly better than a year ago.
  2. The whole improvement, and more, came from short-tail lines. Short-tail releases grew from $152m to $201m (+32%) in the quarter and $355m to $437m (+23%) over the half.
  3. The long tail flipped. It went from a $1m credit in Q2 2025 to a $32m charge in Q2 2026, and from $5m to $48m adverse over the half. In Arch's own words, at the same locator:
"Long-tailed lines included $ 26 million of adverse development in casualty, primarily from the 2022 and 2023 underwriting years." — 10-Q Q2 2026, Note 5, paragraph headed 2026 Second Quarter, reinsurance segment

and in insurance: "Long-tailed lines primarily included adverse development in programs business, primarily from the 2021 to 2023 accident years." (same note, same paragraph block).

That is the cycle showing up in the accounts before it shows up in the combined ratio. The releases are coming out of 2024 and 2025 property and property-catastrophe underwriting years — a short-tail book being released a year or two after it was written, which is precisely the source that a softening property market stops replenishing. The charges are coming out of 2021-2023 casualty, which is the industry's known problem and which has years left to develop.

The mortgage segment is the third leg and it is fading on its own schedule: $64m released in Q2 2025, $45m in Q2 2026. The cluster pass already flagged the mortgage loss ratio going −1.2% to +6.5%.

Why this matters more than the headline. Arch's own release puts the combined ratio excluding catastrophes and prior-year development at 82.5% against 80.9% a year earlier. That 1.6-point deterioration is the underlying business. The reported result is held up by the $169m release. If the short-tail source dries up while the long-tail charge keeps growing, the reported number moves toward the underlying one with no catastrophe required.

§2 — The second finding: a debt-funded buyback into a softening market

From the Q2 2026 10-Q, three facts that belong in one sentence:

  • Senior notes $2,729m at 2025-12-31 → $4,286m at 2026-06-30 (Note 9). The cash flow statement shows $1,977m of proceeds from borrowings in H1 2026 against zero in H1 2025, less $398m of repayments.
  • Buybacks of $1,949m of cash in H1 2026 (Note 3: "Arch Capital repurchased 20.7 million common shares … with an aggregate purchase price of approximately $ 1.9 billion"), against $359.7m in H1 2025 — a 5.3x step-up. Average price paid ≈ $92 against a book value per common share of $68.04, i.e. ~1.35x book.
  • Equity fell. $24,206m (2025-12-31) → $24,030m (2026-06-30), despite $2,104m of net income.

The Board added $3.0bn of authorisation on 2026-04-19; $2.2bn remained at 2026-06-30. Shares outstanding 359.0m → 340.9m, −5.0% in six months.

None of this is imprudent on its own — Arch is over-capitalised by any conventional measure and the cluster pass correctly reads the capital return as a cycle call ("the disciplined actor"). But the financing of it is new. A company that genuinely believes premium rates are unattractive returns capital out of earnings; a company that borrows $1.98bn to return $1.95bn is expressing a view on its own share price as well, at 1.35-1.41x book, while its underlying combined ratio deteriorates 1.6 points and its long-tail reserves turn adverse. Those two views can both be right. They are not the same view, and the cluster file recorded only the first.

§3 — S2 kill test

FY2021FY2022FY2023FY2024FY2025
Revenues $m9,2489,61313,63417,44019,929
Net income $m2,1571,4764,4434,3124,399
Net premiums earned $m8,0829,67912,44015,10017,065
Shareholders' equity $m13,54612,91018,35320,82024,206
Loss reserves $m17,75720,03222,75229,36933,547
Diluted weighted shares m400.3377.6378.8381.8375.9

(XBRL companyconcept, CIK 947484 — Revenues, NetIncomeLoss, PremiumsEarnedNet, StockholdersEquity, LiabilityForClaimsAndClaimsAdjustmentExpense, WeightedAverageNumberOfDilutedSharesOutstanding.)

Share count over five years, the stage's loudest trap check: 406.7m (2020-12-31) → 359.0m (2025-12-31) → 340.9m (2026-06-30). No dilution anywhere; the 2023 equity jump came from retained earnings, not issuance.

2026 to date, and the turn. Q2 2026 against Q2 2025, from the results 8-K:

Q2 2026Q2 2025Δ
Gross premiums written $m6,1266,196−1.1%
Net premiums written $m4,0494,348−6.9%
Net premiums earned $m3,9854,337−8.1%
Underwriting income $m657818−19.7%
Loss ratio55.1%53.1%+2.0 pt
CR ex-cat ex-PYD82.5%80.9%+1.6 pt
Net income to common $m1,047 (=$3.00/sh)1,182 (=$3.23/sh)−11%
Operating income to common $m893 (=$2.56/sh)979 (=$2.58/sh)−9%
BVPS$68.04—+2.8% vs 2026-03-31

Survives or dies: survives, easily and with room. $24.0bn of equity, $1.1bn of cash, $65m of interest paid in a half against $2.1bn of net income, no maturity wall disclosed, a $2.2bn unexercised buyback authorisation and a reserve position the auditor treats as the single critical estimate rather than a solvency question. Nothing in this pass puts the company's existence in question. What it does put in question is the level of earnings: gross premium flat, net premium falling faster than gross because Arch is buying more reinsurance or ceding more, earned premium falling faster still, the underlying combined ratio deteriorating, and the released reserves that close the gap coming out of a short-tail book that is itself shrinking.

§4 — S3 gate: the three biggest risks management itself discloses

All three are taken from Arch's own ordered risk-factor summary (10-K FY2025, Item 1A, "RISK FACTORS SUMMARY", bullets 1-4 under Risks Relating to Our Industry, Business and Operations) — i.e. management's own ranking, not this pass's.

  1. Competition. "We operate in a highly competitive environment, and we may not be able to compete successfully in our industry." (Item 1A summary, bullet 1)
  2. Cyclicality — placed second, above catastrophes. "The insurance and reinsurance industry is highly cyclical, and we may at times experience periods characterized by excess underwriting capacity and unfavorable premium rates." (Item 1A summary, bullet 2) This is the cluster thesis stated by the company, in the company's own priority order, and it is the reason the S0 score of 2/4 is read as "cyclical business" rather than "bad company".
  3. Catastrophe volatility. "Claims for natural catastrophic events could cause large losses and substantial volatility in our results of operations and could have a material adverse effect on our financial position and results of operations." (Item 1A summary, bullet 4)

Reserving appears further down the same summary ("Underwriting risks and reserving for losses are based on probabilities and related modeling, which are subject to inherent uncertainties.") and is the auditor's sole critical audit matter: "Valuation of Reserve for Losses and Loss Adjustment Expenses", on a $33.5bn balance. PwC, PCAOB ID 238.

Red-flag sweep. Run against the EDGAR filing index for CIK 947484 back to 2010 rather than via efts.sec.gov full-text: no NT 10-K or NT 10-Q, no UPLOAD or CORRESP (no SEC comment-letter correspondence at all), no 10-K/A and no 10-Q/A. Nine 8-K/A filings, which is ordinary. This is the cleanest filing index of the three names in this slot.

§5 — S5 ownership, and a defect that invalidates a row in every 13F table in this repo

13F holders at 2026-06-30 (threespread.py holders ACGL), top five: BlackRock 30,220,703 sh / $2,933.2m (−1,133,501 QoQ); State Street 16,251,212 / $1,577.3m (+490); BAMCO Inc (Baron) 15,872,450 / $1,540.6m (−837,309); Invesco 4,947,084 (+887,704); Royal Bank of Canada 3,495,371 (−36,673).

⚠️ The quarter is incompletely ingested — 609 filers against 898 for 2026-03-31. Prior holders with no filing in the quarter include both Vanguard entities (45.2m and 14.7m shares), Artisan (22.7m), WCM (12.6m), Geode (10.6m), JPMorgan, Northern Trust, Morgan Stanley and Dimensional. No exit may be read from any of those absences. This is the fourth consecutive slot in which the warning has fired.

13D/G: all 13G, no 13D; no activist and no strategic block.

Proxy (DEF 14A, 2026-03-24, as of 2026-03-09): "All directors and executive officers (17 persons) (22)" hold 11,618,680 common shares = 3.3% of 356,272,841 outstanding; every individual officer is below 1%. Director Brian S. Posner holds 3,000 Series F and 5,000 Series G preferred shares — the only preferred held by any insider.

⚠️⚠️ The Sixth Street row is not a holding. It is Schwab's book, counted twice.

3spread reports Sixth Street Partners Management Company, L.P. as a NEW holder of exactly 2,177,713 ACGL shares ($211.4m) — byte-identical to Charles Schwab Investment Management's line in the same table. The same pair appears in this slot at KNSL (199,715 shares each, $65.9m) and AX (751,934 each, $73.2m), and appeared at QCOM in slot names-ai-capex-2b (22,310,326 each, $4,122.7m), where the run recorded it as an unexplained duplication and did not carry it as a fact. That instinct was right. Here is the cause, traced to the source:

  • Sixth Street's 13F-HR for 2026-06-30, accession 0001752724-26-000051, contains an information-table file named 13F_0000884546_20260630.xml — 2,049,306 bytes. 0000884546 is Charles Schwab Investment Management's CIK. The filer agent attached Schwab's holdings table to Sixth Street's submission.
  • Sixth Street's own cover page in that same accession declares <tableEntryTotal>9</tableEntryTotal> and <tableValueTotal>579182557</tableValueTotal> — 9 holdings, $579m — against a table holding thousands of rows.
  • Sixth Street filed a 13F-HR/A, amendmentType RESTATEMENT, the same day (2026-08-14, accession 0001752724-26-000055) with the correct 4,662-byte table 13F_0001812095_20260630.xml. Its nine lines are: Atlanta Braves Holdings, Caris Life Sciences (×2), Essent Group, MasterBrand, PG&E, Sixth Street Specialty Lending and Sprinklr (×2). No ACGL, no KNSL, no AX, no QCOM.
  • 3spread has ingested the original and not the restatement.

Consequences for this repository, stated plainly so the next run does not re-derive them: (a) every "Sixth Street Partners Management Company" row in a threespread.py holders output is Schwab's position duplicated and must be dropped; (b) the shares reported total at the top of each table is overstated by that amount; (c) the QCOM scorecard's S5 note, which called it "read as duplication, not carried as a fact", is correct and can now name the cause; (d) this is not a 3spread bug — 3spread faithfully ingested a defective filing — so the fix is to prefer the latest 13F-HR/A for a given manager-quarter, not to distrust the vendor. threespread.py gained a mechanical duplicate-row warning in this commit so the next run is told rather than having to notice.

§6 — S8 valuation and sizing (proposals only — ⏳ OPEN (user))

Price $95.78 at 2026-09-21 (market data NASDAQ:ACGL). Shares 340.9m → market cap $32.6bn. Common equity $23,200m ($24,030m less $830m of preferred) → BVPS $68.04, P/B 1.41. ADV $349.7m over the 21 sessions to 2026-09-21. LTM net income $4,692m (FY2025 $4,399m less H1 2025 $1,811m plus H1 2026 $2,104m) → LTM P/E ≈ 7.0x.

Out-year model. The driver is not premium volume — it is the three-part combined ratio, so the model is built on it. Assumptions named so they can be disagreed with: net earned premium −7% in 2026 and −5% a year after; ex-cat ex-PYD combined ratio worsening 1.5 points a year from 82.5%; catastrophe load 6 points; favourable PYD fading from ~4 points to ~1.5 points by 2028 as the short-tail source shrinks; buybacks continuing at ~$2.7bn a year; tax ~14%.

2025A2026E2027E2028E
Net earned premium $m17,06515,90015,10014,350
CR ex-cat ex-PYD~80.5%82.5%84.0%85.5%
Cat + PYD, net pts~+1+2+3.5+4.5
Reported combined ratio~81.5%84.5%87.5%90.0%
Net income to common $m4,359~3,910~3,640~3,380
Average diluted shares m375.9~348~319~291
EPS$11.60~$11.20~$11.40~$11.60
Year-end BVPS$64.5~$72.4~$82.0~$92.3

Note what the model says: EPS is flat for three years while the combined ratio deteriorates 8.5 points, because the buyback shrinks the share count 23% over the period. Per-share earnings hide the cycle; book value per share does not, and it compounds ~11% a year even after the buyback drag.

BasisValue
Bear2028 BVPS $85 (faster reserve strengthening, one bad cat year) at 0.9x~$77
Base2028 BVPS $92.3 at 1.25x, cross-checked at 9x $11.60 EPS = $104~$105
Bull2028 BVPS $95 at 1.45x, cross-checked at 12x EPS~$130

Against $95.78, the base case is +10% over roughly 2.3 years, before a dividend of about 0.1%. That is the sixth name in the last nine in this rotation whose correct operating forecast produces little or no return at the traded price (MRVL, ARM, ASML, CBRS, QCOM, now ACGL), with APH the exception.

Proposed buy price and size: ⏳ OPEN (user). For what it is worth to that decision: ADV of $349.7m/day imposes no liquidity constraint at any size the owner is likely to run, and the shares trade at 1.04x the modelled 2028 book value, which is the least demanding entry arithmetic of the three names in this slot.

§7 — S9 monitoring

  • Next earnings: late October 2026, DERIVED from filing cadence, not scheduled. Q3 results 8-Ks were filed 2025-10-27 and 2024-10-28; the 10-Q followed on 2025-11-06. market data's calendar needs a paid plan and Equibles — which gave QCOM a real scheduled date last slot — did not connect this session.
  • Filing feed: the Tier 0 EDGAR sweep already covers CIK 947484 (watch_edgar.py), so a 10-Q, 8-K, 13D/G or Form 4 cluster promotes the name automatically.
  • Leading indicator between quarters: January renewal commentary from the European reinsurers and the retrocession market. Arch's own gross-flat/net-down-10% signature is a pricing statement; if 1/1 2027 property-cat rates fall again, the short-tail release source for 2027 shrinks with them.
  • The specific trigger that converts this to a buy: the long-tailed PYD column returns to zero or better for two consecutive quarters while the ex-cat ex-PYD combined ratio stops rising — i.e. the casualty years stop deteriorating before the property release runs out — with the shares still inside 1.15x book. That combination would mean the cycle risk is priced and the reserve risk has peaked. Neither condition holds today.

§8 — Evidence that is thin, named

  • No earnings-call transcript at all. Equibles did not connect; fool.com has no Q2 2026 ACGL transcript. The Q1 2026 fool.com transcript exists but WebFetch returns a model summary of it, not a locatable span, so it is [search-summary] and is quoted nowhere in this dossier or in relationships.yaml. Management commentary here is entirely from the 8-K release and the filings.
  • No short interest. FINRA is not reachable from this environment and Equibles, the usual route, was down. Four previous name slots had this figure; this one does not.
  • No analyst-coverage count. Standing gap; the cluster pass proposed folding it into the S6 search, which was not re-run for this name.
  • No shelf read. No S-3 appears in the recent filing index, but absence in an index read back to 2025-06 is not proof that no effective shelf exists; ACGL is a WKSI and may have one outside the window read.
  • Proxy compensation structure not extracted — ownership table only. This is the fourth name slot to leave it, and it is now the longest-standing unclosed item in the S3 gate-minimum table.
  • Segment-level 2026 premium detail not read beyond the group table in the release; the cluster file's segment figures (reinsurance NPW −10.4%, insurance −5.1%) are carried from there.

Process support, not investment advice.