0Q99 — ageas SA/NV

ageas SA/NV (0Q99) — deep-value analysis

Date: September 4, 2026 · Symbol from the task: 0Q99 (LSIN) · Real trading line: AGS.BR, Euronext Brussels · ISIN BE0974264930 Price: EUR … (close 09/03/2026) · Shares outstanding: 209.2 mn · Market cap: EUR 15,962 mn 52-week range: 55.85 – … — the stock sits at the top of the range Reporting currency = quotation currency = EUR (no FX adjustment) Prior GBL score (tracker, evaluated 09/02/2026): 72.0% — BUY (Graham 8.5 · Buffett 6.5 · Lynch 7.0 · total 22.0/30 · F-Score 6.0 · denominator 59.0 · discount to Graham Number …)

A note on the symbol — and the report’s first finding. 0Q99 is the London Stock Exchange international line for ageas SA/NV’s ordinary share. It is not an ADR and has no conversion ratio: on 07/17/2026 both lines closed at EUR 72.35, an identical figure — a 0Q99 share is an AGS.BR share. Ageas is not SEC-registered: it has no CIK, no 10-K and no 20-F, so the EDGAR step was intentionally skipped (deep_data_pack.py already contains the alias). The primary sources for this report are the audited 2025 Annual Report (12.4 MB PDF, downloaded from downloads.ctfassets.net and extracted in full locally with pdftotext -layout — 20,387 lines), the H1 2026 results release from 08/27/2026, the FY2025 release from 02/25/2026 and the FY2023 release from 02/28/2024 (for the 2022–2023 comparatives). Figures marked “(AR25 p. N)” come from the audited consolidated financial statements; those marked “(H1’26)” or “(FY25)” from the “Key Figures” appendices of the releases.

A note on method regarding the vault. vault_semantic_search is not exposed in this session — no vault MCP tool is available. This is stated explicitly, per the rule, before switching to disk search. On disk, two prior deep analyses of the same company were found, filed under the AGS ticker (2026-09-01-deep-AGS.md, 2026-09-02-deep-AGS.md), plus the data pack data-pack-AGS-20260902.md. These were used as a page map, not as a source of figures: every number in this report was re-extracted from the primary filings downloaded today. Where my verification confirms a prior finding, I say so; where it corrects or refines it, I flag it as such. This report additionally adds the entire H1 2026 half-year (published 08/27/2026) and the full 2022–2026 series, which don’t appear in the prior artifacts.

Executive summary

The thesis, in one sentence: Ageas is a well-run, competently managed European insurer, but at EUR … there is no margin of safety left — because two-thirds of reported profit is profit recognized from stakes Ageas doesn’t control, of which it actually collected only 189 million out of 1,123 recognized in 2025, and the gap between those two figures explains why, in the company’s most profitable year ever, free operating capital generation halved.

Finding zero: the tracker’s 72% “BUY” score rests on a July 2022 price

This is not a valuation opinion, it’s a data defect that has been fully diagnosed and must be fixed before any discussion of the company.

The 0Q99 row in the tracker has price=… and eps=4.406. Both come from the info object of the Yahoo symbol 0Q99.L, whose regularMarketTime field has the value 1658475299 — i.e. July 22, 2022, 07:34 UTC. The quote has been frozen for four years and six weeks. The … price fits perfectly with Ageas’s 2022 range (34.61 – …), and the 4.406 EPS is a 2022-vintage EPS. At the same time, the bookValue = … field is current — it matches exactly the 10,223 mn EUR of equity at 06/30/2026 divided by 209.2 mn shares (= 48.87). Yahoo refreshes the company’s fundamentals block, but not the quotation block for the London line.

The effect on the score is mechanical and large:

With the correct price, the “41% discount to the Graham Number” — one of the inputs that raises the Graham score to 8.5/10 — becomes a 9.6% premium. The sign flips. Additional check: the 0Q99.L line does actually trade (220 trading sessions in the last year, zero days with zero volume, the last close in the series was … on 07/17/2026); it is not a dead line, only the real-time quote field is broken. Operational recommendation: the 0Q99 row must be re-fed from AGS.BR, or removed as a duplicate of it. The “BUY 72.0%” verdict is not a reasoning error, but a data pipeline one.

The six things found in the audited statements

1. Of EUR 1,123 mn of profit recognized from investees, Ageas collected EUR 189 mn. Note 4, “Equity-accounted investments” (AR25 p. 232), publishes both columns, side by side. Net income attributed to Ageas from associates and joint ventures was EUR 1,123 mn in 2025; dividends actually received from the same entities, EUR 189 mn. Cash conversion: 16.8%. For China Taiping Life alone: EUR 880 mn recognized, EUR 104 mn collected — 11.8%. In 2024 the ratio was 564 recognized / 148 collected (26.2%), so conversion has worsened. Since net income attributable to shareholders was EUR 1,712 mn, unconsolidated investees produced 65.6% of the group’s profit and 11% of its cash. This is the central finding of the report and is the O’Glove method applied literally: profit is an opinion, cash is closer to fact.

2. The 2025 improvement in the combined ratio doesn’t come from underwriting, but from buying less reinsurance — verified segment by segment. The operating segment note (AR25 p. 311–312) breaks the combined ratio into three components. At the group level: loss ratio 59.1% → 60.0% (worse by 0.9 pp), expense ratio 28.2% → 28.6% (worse by 0.4 pp), reinsurance ratio 6.3% → 3.9% (lower by 2.4 pp). Both technical components worsened; the entire 1.1-point improvement in the combined ratio comes from retaining more risk on the balance sheet. My refinement versus prior analyses: the segment breakdown shows where it comes from, and the source is Europe — there the loss ratio rose from 61.0% to 63.5% (+2.5 pp) while the reinsurance ratio collapsed from 8.2% to 4.3% (−3.9 pp). Europe is the segment esure entered. What happens when you retain more risk and a hail-storm summer comes was seen in H1 2026: the combined ratio jumped to 95.2%, from 92.1%, with roughly 5 points of weather impact (H1’26).

3. Operational earnings per share DECLINED in H1 2026, even though operational earnings grew 6%. From the company’s own table (H1’26, Appendix 1): net operating result EUR 776 mn versus 734 (…); average share count 197 mn versus 185 (…); net operating result per share EUR 3.94 versus 3.96 — minus 0.5%. The entire growth was absorbed by the share issuance. The Elevate27 target is average earnings-per-share growth of 6… over the cycle.

4. Book equity grew 21.8%; tangible equity DECLINED 17.6%, and available solvency capital fell 3%. Shareholders’ equity: 7,752 → 9,441 mn EUR (AR25 p. 149). Goodwill and intangibles: 1,626 → 4,390 mn EUR (AR25, consolidated balance sheet). The difference — tangible equity — fell from 6,126 to 5,051 mn EUR. Available solvency capital, the framework that actually determines what can be distributed and where intangibles are worth zero: 20,077 → 19,486 mn EUR (FY25). And the solvency ratio has declined for three half-years running: 240% (H1’25) → 211% (FY25) → 195% (H1’26) (H1’26).

5. The 2021 long-term incentive plan vested ZERO, because Ageas’s relative TSR was below the 25th percentile of the peer group — and the remuneration committee instead proposed a one-off exceptional bonus. Verbatim, AR25 p. 34: “Ageas’s relative TSR was below the 25th percentile of the peer group. As such, there was no vesting of the LTI-plan 2021.” And on p. 35: “The Remuneration Committee, with the endorsement of the Board, proposes the allocation of an exceptional, one off bonus to the members of the Executive Committee… 35% for the Chief Executive Officer; 25% for the Chief Financial Officer…”. The narrative “the stock has tripled since 2020” is true and irrelevant: the European insurance sector re-rated faster than Ageas, and the company’s own relative-performance instrument certifies it.

6. Management holds practically no shares — and doesn’t meet its own threshold. The ownership-requirement check table (100% of base salary, AR25 p. 34, valued at EUR 59.80/share): CEO Hans De Cuyper 9,161 shares = EUR 547,828 versus 800,000 base salary → 68%, below threshold; CFO Wim Guilliams 3,500 shares → 41%; CRO Christophe Vandeweghe 177 shares → 2%; Ben Coumans 195 shares → 2%; Karolien Gielen and Heidi Delobelle — empty box, zero shares. Only 2 of 8 executive committee members meet their threshold. Total held by all board and executive committee members combined: 59,586 shares = 0.028% of the company. The “30.5% insiders” shown by yfinance are strategic shareholders (BNP Paribas Cardif and affiliates), not management.

The FCF bridge, briefly

CFO − capex gives … − 322 = EUR … mn, i.e. a 16.1% yield on market cap — an impossible figure, because an insurer’s IFRS operating cash flow contains policyholders’ money. The proof is right in the cash flow statement (AR25 p. 151): inside that CFO sits a gross movement of −… mn EUR at “Net changes in all other operational assets and liabilities,” offset by +… mn EUR on the next line. The correct figure is Holding Free Cash Flow, which the company publishes: EUR 774 mn in 2025 (FY25), independently confirmed by an operating free capital generation of 793 mn. Normalized through five independent paths (below), EUR 950 mn — FCFE yield 5.95%. Divergence from “simple FCF”: ….

Estimated value and verdict

Five triangulated models give a range from EUR 40.12 (pessimistic DCF) to EUR 103.29 (optimistic DCF), with the margin-of-safety median at …. Monte Carlo over 20,000 scenarios: median intrinsic value EUR …, probability of undervaluation …%, probability of a margin above 30% only 9.9%. External check: the average target of the 11 analysts covering the stock is EUR 72.76 (range 63.40 – 90.00), aggregate recommendation “hold” — the EUR … price is above the market’s average target and at the top of the 52-week range.

Verdict: HOLD / DO NOT BUY at EUR …. A good company at a full price. The re-entry threshold for a 25% margin of safety is below ~…; for 15%, below ~…. For those already holding below …: the thesis has played out, the ~5% dividend yield and 0.53 beta justify keeping a reduced position, not adding.