SPNT — SiriusPoint Ltd.

Deep-value report — SiriusPoint Ltd. (SPNT) · REFRESH

Analysis date: 2026-09-04 · Price: … · Market cap: …B · Shares: 116,065,965 common issued (30.06.2026) / 119.0M average diluted Q2 2026 · Exchange: NYSE · Sector: Financial — Specialty Insurance/Reinsurance (Bermuda)

Regime: REFRESH over the reference thesis from 2026-07-24-deep-SPNT.md (41 days). Mechanical trigger: 2 new filings (10-Q and 8-K, both 29.07.2026) and one falsifier that remained unverifiable (YoY revenue growth, missing its second quarter).

What I inherit without re-derivation (nothing in the new filing touches it): segment structure and the moat verdict; the CM Bermuda transaction history (2024-2025) and the Armada/Arcadian sales; the four 2025-2026 rating upgrades; the long-term debt structure; the catastrophe risk profile (1-in-250 PML). What I fully re-derive: normalized earning power, balance-sheet quality (Q2 2026 moves receivables significantly), underlying underwriting result, the valuation, and all five models.

New sources versus the July report: 10-Q Q2 2026 (filed 29.07.2026, accession 0001576018-26-000085) and 10-Q Q1 2026 (07.05.2026, 0001576018-26-000065) — downloaded today, missing from the file set because the downloader only covers fiscal-2025 quarters; 8-K/Exhibit 99.1 Q2 2026 earnings release (29.07.2026, 0001576018-26-000084); 10-K FY2025 (24.02.2026) for debt, tax, and receivables notes; data-pack-SPNT-20260904.md; research-SPNT-20260904.md; yfinance for current multiples and peers, 04.09.2026.

Executive summary

SiriusPoint reported on July 29, 2026 a quarter that looks good at the headline and less so underneath. Net profit available to common shareholders was …M in Q2 2026 (… versus …M in Q2 2025), diluted EPS … operating EPS … versus … last year — practically flat [8-K 29.07.2026, Exhibit 99.1]. For the half-year, net profit available to common shareholders rose 44% to …M and operating EPS 17% to … and the Core combined ratio improved 2.3 points, to 90.1%. Book value per diluted share ex-AOCI grew 8% from 31.12.2025, to …

Below the headline, however, the half-year improvement is entirely attributable to the absence of catastrophes. Catastrophe losses were …M (0.5 points of combined ratio) in H1 2026 versus …M (5.3 points) in H1 2025, when the California wildfires hit. The Core attritional combined ratio — the one excluding both catastrophes and prior-year reserve releases — deteriorated from 90.9% to 93.4%, by 2.5 points (computed from published components: attritional loss ratio 58.6% + acquisition cost ratio 27.5% + other underwriting expense ratio 7.3%, versus 59.1% … … last year) [8-K 29.07.2026, segment tables]. On a standalone Q2 basis, the deterioration is 2.1 points (93.8% vs 91.7%), and the reported Core combined ratio ROSE, from 89.5% to 91.4%. The attritional loss ratio improved 0.5 points; the acquisition cost rose 2.1 points. The company explains the rise through “profit commission accruals related to favorable loss experience” — a partly self-limiting explanation, but the net arithmetic remains negative by 1.6 points.

Revised thesis: earning power is higher than I thought in July — TTM operating net profit (H2 2025 + H1 2026) is …M, i.e. an operating ROE of 14.7% on common equity ex-AOCI of …M, at the top of the 12… “across the cycle” band the company itself targets. But the quality of this earning power is weaker: 3.8 points of the Core combined ratio come from prior-year reserve releases (…M Core in H1 2026), technical receivables grew 13.8% year over year while net earned premium stagnated (…), and cash conversion stayed weak (CFO/net profit 0.53 TTM, 0.22 in FY2025). Egan’s turnaround is real; what’s changed over the last two quarters is that the underlying underwriting margin has stopped improving and started eroding, and profit is being held up by weather luck and the reserve cushion.

Estimated value (triangulated, five models): … – …/share, model median … versus the current price of … The normalized owner-earnings yield (10.6% on …M) is consistent with the data pack’s FCF yield (9.4%) and with yfinance’s forward-EPS consensus yield (11.1% on …) — my figure isn’t an outlier versus the market, it’s slightly more conservative.

Verdict: MONITOR / SPECULATIVE — unchanged from July, but for a different reason. In July the argument was that the price already reflected the turnaround. Today the argument is that the price reflects the turnaround and a hurricane-free year and a reserve cushion being consumed. The 1.27x P/B is the highest in SiriusPoint’s history (2021-2026); the calendar-year median over the period is 0.84x. It’s not a sell thesis — capital is well allocated, ratings are rising, the capital structure is clean after redeeming the preferreds — but there’s no margin of safety, and the negative catalyst (a normal hurricane season + normalization of reserve releases) is more likely than the positive one.