AII — American Integrity Insurance Group, Inc. · 2026-09-11 · Verdict: WATCH · Conviction 3
Price $25.41 (screen row, screen dated 2026-09-11; no live quote) · Mkt cap $498M · EV ~$498M (long-term debt $0.4M) · P/E 5.7x LTM GAAP · Price / normalised after-tax profit 7.1x · P/TBV 1.35x · ROTE 24.9% LTM / 20.0% normalised · Net debt: none · ADV $5.5M Sources read: 10-K 2026-02-27 (Items 1, 1A, 7), 10-Q 2026-08-13 (Item 2), DEF 14A 2026-04-17, 8-Ks 2026-05-12 (Q1 results), 2026-06-01 (reinsurance placement), 2026-08-11 (Q2 results), 2026-09-08 (2Q26 investor deck), Form 4s (12m, 22 rows), Q2 2026 earnings call transcript.
Desk stats - Revenue trend. FY2025 total revenues $276.5M, up 35.3% (10-K, Item 7). Q2 2026 revenues $115.2M, up 54.6% (10-Q). Almost none of that 54.6% is business growth: gross premiums earned rose only 8.3% and gross premiums written 13.8%; the rest is the January 1, 2026 cut in the non-catastrophe quota share cession from 40% to 25%, which converted ceded premium into net premium (10-Q, Item 2). H1 2026 gross premiums written grew 9.5%. - Normalised after-tax profit. LTM (H2 2025 plus H1 2026) pre-tax income $121.7M, GAAP net income $88.1M. Adjustments: add $0.9M executive transition cost (10-Q non-GAAP table) and $2.3M post-IPO transition expenses (10-K non-GAAP table); deduct $0.1M realised investment gains, $0.7M favourable prior-year reserve development, and a $30.0M normal catastrophe load, because the company recorded exactly zero catastrophe losses in FY2025 and in H1 2026 while FY2024 carried $32.2M (10-K and 10-Q underlying-loss tables). Normalised pre-tax $94.1M, taxed at 25% = $70.6M normalised after-tax, against $88.1M GAAP. - EV / normalised after-tax profit. Market cap $497.9M (19,593,427 shares at $25.41); long-term debt $0.4M and no netting of insurer cash, since the $698M of cash and investments backs $555M of unearned premium and $254M of loss reserves (10-Q balance sheet). 7.1x, versus the screen's 5.1x. - Financial-lane equivalents. P/E 5.7x LTM GAAP; P/TBV 1.35x ($369.5M tangible common equity, no goodwill or intangibles on the balance sheet); ROTE 24.9% LTM and 20.0% normalised; tangible book per share $15.42 to $18.86, up 22.3%, or 28.9% adding back the $1.02 special dividend paid March 30, 2026. Balance-sheet leverage is nil (debt/equity 0.1%, equity/assets 22.9%). The real leverage is underwriting: $35M first-event catastrophe retention is 9.5% of equity and the $75M four-event aggregate retention is 20.3%. - Is the growth sustainable? Organic and cash-backed (H1 2026 operating cash flow $129.8M on $54.1M of net income; voluntary policies in force up 22%), but the reported growth rate is flattered by the quota-share change, and the Citizens take-out engine that produced 2025 is over: 665 policies assumed in H1 2026 against 33,867 in 2025 and 68,844 in 2024. No guidance is given. - What the screen got wrong. Four things. The 5.1x P/E uses FY2025 net income struck at a 13.4% effective tax rate because of a one-time $9.7M deferred tax benefit on the LLC-to-corporation conversion; at a normal rate FY2025 earnings were about $86M, not $99.6M. The screen charges no catastrophe load in a year with zero catastrophes. Its +150.7% net income growth and +54.6% quarterly revenue growth conflate the Citizens windfall and the quota-share change with organic growth. And it mixes periods, taking earnings from CY2025 and book value from June 30, 2026.
1. What the business actually does
American Integrity writes admitted personal residential property insurance, mostly Florida homeowners (HO-3 is 67% of policies), plus dwelling, manufactured home, vacant-property, watercraft and, since October 2025, small commercial residential. Florida was 92.0% of policies in force at June 30, 2026, with Georgia, South Carolina and North Carolina the balance (10-Q). Founded 2006, IPO May 9, 2025 at $16.00. At June 30, 2026 it had 461,714 policies and $1,029M of in-force premium, the second largest Florida residential book by policy count (2Q26 deck, FLOIR data). The carrier AIIC holds the statutory capital while separate subsidiaries do underwriting, claims and reinsurance analytics. It cedes heavily, 60.5% of gross premiums earned in H1 2026, and buys a $2.27B first-event catastrophe tower costing a guided $430M to $440M for the 2026 treaty year (8-K 2026-06-01).
2. Why it is mispriced — the edge case
There is none, and that caps this at WATCH. The stock sits 4.6% below its 52-week high after rising 41.1% in six months on $5.5M of daily volume. Nothing is forcing anyone to sell; the flow runs the other way. Founder and CEO Robert Ritchie sold 550,000 shares for $13.4M between August 3 and August 31, 2026 at $20.68 to $26.63, cutting his holding from 2,433,997 to 1,883,997 and dropping out of 10% owner status on August 26 (Form 4s). No insider bought a share in the open market in twelve months across 22 transaction rows. Sowell Investments Holding Co. still owns 22.8% and holds demand registration rights (DEF 14A). The screen's $25.41 sits in the middle of the CEO's own selling range. What the screen found is a real accounting artifact, the untaxed and catastrophe-free 2025, not a dislocation.
3. Underwriting quality, combined ratio and reserve development
The combined ratio was 80.9% in 2024, 63.7% in 2025, 75.0% in Q1 2026 and 63.4% in Q2 2026 (10-K; 8-K 2026-05-12; 10-Q). The 2025 number is not a run rate. It benefited from two windfalls the company names itself: policies assumed from Citizens carry "immaterial upfront acquisition costs" and are covered free by the in-force treaty, so "periods of heavy take-out activity result in lower expense ratios and loss ratios" (10-Q, Item 2); and there were no catastrophe losses at all.
The cleanest read on underwriting is the gross underlying loss and LAE ratio, which strips out reinsurance, catastrophes and reserve development. It was 17.0% in both 2024 and 2025. It has since risen to 18.6% in H1 2026 from 15.8%, and to 18.1% in Q2 2026 from 15.5% (10-Q tables). That 2.6 to 2.8 point deterioration coincides exactly with the mix shift. Tri-County (Miami-Dade, Broward, Palm Beach) went from 185 to 7,636 voluntary new policies year over year and from about 1% to 33% of voluntary new business premium; middle-aged homes went from 437 to 9,062 policies and from 4% to 24% (2Q26 deck, transcript). Together that is 57% of new voluntary premium in segments the company deliberately shrank during the litigation crisis. Asked about margin, the CEO said on the Q2 call: "we model and expect the underlying gross loss ratio to be a few points higher than the overall portfolio, but the premium we are collecting accommodates for that." That is the right answer and it is also unproven; one clean season of loss data is not evidence.
Reserving offers no cushion now. Favourable prior-year development was $3.2M in 2024, $1.8M in 2025 and zero in both quarters of 2026 (10-K, 10-Q). Total reserves of $254.2M at June 30, 2026 are 95.2% IBNR and only $12.1M case, so almost the entire liability is actuarial estimate rather than claim-specific. Average gross written premium per policy was $2,156 in Q2 2026 against $2,178 a year earlier, and Citizens' own book has fallen to 278,000 policies, its lowest in 20 years (2Q26 deck): the depopulation opportunity is exhausted and everyone now competes in the voluntary market.
4. Balance sheet and capital allocation
Essentially unlevered: $412,000 of long-term debt against $369.5M of equity, 22.9% equity to assets, $698M of cash and investments at AA average credit and 1.3 to 2.0 years duration with no equity exposure (2Q26 deck). About $200M of cash was deployed into fixed income just after quarter end at new-money yields in the high fours against a mid-four book yield (transcript, CFO). The June 1, 2026 renewal took first-event retention flat at $35M despite 19% peak-season exposure growth and cut the four-event aggregate retention from $95M to $75M. A $20M special dividend was paid in Q1 2026; there is no buyback authorisation. Share count is flat at 19.6M with 2.12M shares available under the LTIP.
Compensation alignment is the weak spot. The 2025 bonus plan paid on adjusted ROE with a 4% threshold, 8% target and 12% maximum, and a combined ratio of 98%/94%/90%, against actual results of 42.1% and 63.7%. Every named officer was paid 200% of target, and on February 27, 2026 the Compensation Committee approved the same thresholds for 2026 (DEF 14A). For a company whose own worst ordinary year printed 26.8% ROE, a 12% maximum is not a hurdle. Separately, CFO Ben Lurie resigned on April 6, 2026 after nine years to return as CFO of Sowell & Co., taking an $800,000 IPO success bonus, $300,000 of consulting fees and accelerated RSUs; his replacement Brian Foley, 35, is a former KBW banker with no prior public-company CFO experience. The company also changed auditors three times between December 2023 and November 2024, including discharging Deloitte before any report was issued, and had material weaknesses in internal control that it says were remediated as of December 31, 2025 (DEF 14A; 10-K Item 1A).
5. Management: said versus did
There is no guidance history to test: the company has been public sixteen months and issues none. What can be checked holds up. The June 1 announcement of a $430M to $440M treaty cost was reaffirmed verbatim on the Q2 call; the quota-share reduction was executed at no additional cost; the Tri-County and middle-aged-home re-entry has been delivered in volume. The CEO's framing is uniformly promotional ("the strongest quarter in our company's history," "our best years are still ahead") and the Q2 call contained no discussion of what a bad wind season does to the numbers.
6. Valuation
Base (50%): no major Florida landfall through 2027, normalised earnings of roughly $75M to $80M as the quota-share retention annualises and reinsurance costs fall again, tangible book about $20.90 at year-end 2026 and $24.50 at year-end 2027, market pays about 1.6x forward book, giving $33, or 9.5x normalised earnings. Bear (30%): a 2004-style four-event season costs the full $75M aggregate retention, about $56M after tax or 15% of book, and repricing the $435M treaty up 20% adds roughly $87M of annual pre-tax cost against $94M of normalised pre-tax income; book falls to about $17.40 and the multiple compresses to 1.0x, giving $17. Bull (20%): two more clean seasons, quota share to zero, 15% policy growth, gross underlying loss ratio holding near 18%, earnings near $115M and book near $29 by 2028 at 1.8x, discounted back to $45. Probability-weighted $30.60, about 20% above $25.41, with a bear case a third below.
Reverse DCF: at $497.9M of market value against $70.6M of normalised after-tax earnings and an 11% cost of equity, the price implies normalised earnings declining about 3% a year in perpetuity. That is not an obviously wrong assumption for a single-state, single-peril insurer at the top of a soft-loss-cost cycle, which is precisely why the multiple is 7x.
7. Catalysts and timeline
End of the Atlantic season on November 30, 2026 resolves the year's binary. Q3 2026 results in early November are the first quarter carrying wind-season exposure at the reduced quota share with the new Tri-County and middle-aged mix. The June 1, 2027 renewal decides whether the reinsurance cost tailwind repeats. Against those, a registered secondary by Sowell (22.8%, demand rights) is the live negative catalyst.
8. Pre-registered kill criteria
- Gross underlying loss and LAE ratio above 20.0% for two consecutive quarters (18.1% in Q2 2026, 17.0% in FY2025): the mix shift into Tri-County and middle-aged homes is worse than priced.
- Any quarter of unfavourable prior-year reserve development, after $3.2M, $1.8M and $0 of favourable development in 2024, 2025 and H1 2026.
- Sowell Investments registering or selling more than 1.5M of its 4,473,824 shares, or Robert Ritchie's holding falling below 1.4M shares from 1,883,997 at August 31, 2026.
- Average gross written premium per policy down more than 5% year over year (Q2 2026: $2,156 versus $2,178), the sign that Florida rate is eroding faster than reinsurance cost.
9. Verdict and one-paragraph summary
WATCH, conviction 3. American Integrity is a genuinely good business at a defensible price, but there is no edge case to underwrite and the screen's cheapness is half real. Correcting for the one-time deferred tax benefit that held FY2025's effective tax rate to 13.4%, and charging a $30M annual catastrophe load against a last-twelve-months period in which a Florida homeowners insurer recorded literally zero catastrophe losses, the multiple is about 7.1x normalised after-tax earnings and 1.35x tangible book, not 5.1x, for a business compounding tangible book at 22.3% on a 20% normalised return, with no debt, a $35M first-event retention that is only 9.5% of equity, and reinsurance costs falling 15% to 20% faster than primary rates. What stops it being an idea is that nothing is forcing anyone to sell: the stock is 4.6% off its high after a 41% six-month run, and the founder-CEO sold 550,000 shares for $13.4M into that strength in August at $20.68 to $26.63 with no insider buying anywhere in twelve months, while a 22.8% holder sits on demand registration rights. Underneath, the two engines of the 2025 print are both switching off, with Citizens take-outs down from 68,844 policies in 2024 to 665 in H1 2026 and Citizens' own book at a 20-year low, and the growth replacing them comes from Tri-County and middle-aged homes, exactly the segments the company shrank during the litigation crisis, where the gross underlying loss ratio has already risen from 15.5% to 18.1% and management concedes it models "a few points higher." Probability-weighted value of $30.60 is 20% above price against a bear case 33% below, and the decisive data is dated: get through November 30 and read Q3's gross underlying loss ratio.
Research for discussion, not investment advice. Positions and sizing are the reader's decision.