American Coastal Insurance Corporation ACIC

ACIC — American Coastal Insurance Corporation · deep-value re-analysis (falsifier triggered: dso)

Date: September 16, 2026 · Reference price: $9.14 · Market cap: $432.0 mil. (47,269,517 shares) · NASDAQ · Sector: property insurance, commercial-residential, Florida

Why this report exists. On 09.15.2026, the mechanical triage ran the 6 falsifiers of the thesis written on 08.16.2026 and one triggered: dso, with values of 261.8 and 241.6 days against a threshold of 90. This session's job is to answer a single question — does the hypothesis that supported the verdict still hold? — and only then redo the full analysis.

Primary sources: 10-Q Q2 2026 (filed 08.06.2026), 10-Q Q1 2026 (05.08.2026), 10-K FY2025 (09.03.2026), 10-K FY2024, SEC XBRL companyfacts (CIK 0001401521, direct extracts for the receivables, premium and reserve series), data-pack-ACIC-20260916.md, research-ACIC-20260916.md (web brief — treated as a deposition, two figures corrected below), Q2 2026 earnings call transcript (08.05.2026), GlobeNewswire press release 09.14.2026, teza-ACIC-20260816.json, simulation mc-ACIC-20260916.json, GBL tracker (score dated 09.06.2026).

Two corrections to the research brief, before anything else:

1. The brief writes "guidance for fiscal year 2026: profit $85-100 mil.," implying net profit. The company's original wording is "earnings before income tax of $85 million to $100 million" — profit BEFORE tax. The difference is $22 million a year and changes the valuation base by 32%. The correct figure, after the H1 2026 effective tax rate of 24.0% ($13,006 thousand / $54,156 thousand), is $64.6-76.0 mil. net profit, midpoint $70.3 mil.
2. The brief attributes to a "compiled SEC 8-K source" an FHCF contract effective 06.01.2026 with 90% coverage, mandatory layer ~$571.5 mil. There's no mention of FHCF anywhere in the Q2 2026 10-Q, the FY2025 10-K, or the reinsurance program description (Note 6 and the Reinsurance section). The filing's description of the program is explicitly private occurrence-based, with exhaustion at $1.68 bn on the first event. I don't use the FHCF figure anywhere in this report; if it exists, it isn't in the public file I read.

---

Executive summary (1 page: thesis, estimated value, verdict)

The answer to the question that commissioned this report: the falsifier triggered on a metric the thesis wasn't measuring. The mechanical checker computes dso as yfinance receivables / quarterly revenue × 91. For ACIC, yfinance's "Accounts Receivable" line is $237.3 mil. as of 06.30.2026 — that's premiums receivable $119,915 thousand PLUS reinsurance recoverable $117,435 thousand, a sum that reconciles to the last dollar with the 10-Q balance sheet. The reinsurance recoverable is a claims-recovery asset, not a revenue collection; including it inflates the ratio 2-3x. Consequence: the metric was already above 90 in every quarter yfinance retains — 176.0 (09.30.2025), 209.5 (12.31.2025), 241.6 (03.31.2026), 261.8 (06.30.2026). The falsifier was triggering from the day it was written. Between 08.16.2026 and today no new filing has appeared (Q2 was filed 08.06.2026; Q3 comes 11.04.2026), so zero new information caused the trigger.

On the metric the thesis actually measured, the verdict is more nuanced — and more interesting. I reconstructed the prior author's definition from his own figures: premiums receivable / TTM gross earned premiums × 365. It reproduces the cited series exactly (27.7 for FY2023, 26.6 for FY2024, 39.7 for FY2025, 72.9 for TTM as of 06.30.2026 — all four match to a decimal). At 72.9 days, the 90 threshold isn't reached: the falsifier is at 81% of the threshold. But the claim it was protecting wasn't "DSO stays below 90," it was "the receivables growth remains isolated — hasn't worsened into a structural pattern." That claim is falsified by the evidence, even if not by the number: 26.6 → 39.7 → 44.6 → 72.9 are four consecutive rising observations, and receivables relative to unearned premiums went from 13.9-19.2% (2024 – mid-2025) to 28.2% / 29.8% / 35.7% (Q4 2025, Q1 2026, Q2 2026). It's not an isolated quarter. It's a level shift over three quarters, plus a $43.2 mil. jump concentrated in Q2 2026.

Structural deterioration or noise? My verdict: a real symptom of a soft market, with near-zero credit risk, amplified by the worst possible measurement date. The reasons, each with its own figure: (a) the credit-loss allowance on premiums receivable is zero as of 06.30.2026 ($0; $26 thousand a year earlier), and the Note 17 roll-forward shows no write-offs; (b) the policy stated in the 10-K is a policy-level test whereby the receivable can't exceed the associated unearned premium — at $119.9 mil. receivables against $335.5 mil. unearned premiums, coverage is 2.8x, and a non-payer is canceled with the unearned premium released, so loss-given-default is near zero; (c) the half-year's operating cash flow is POSITIVE $52.8 mil., so the company generated cash after absorbing the $49.5 mil. receivables build; (d) June 30 is the balance-sheet date right after the June 1 renewal, when the entire portfolio is billed and uncollected — historically, the balance collapses in Q3 ($18.6 mil. at 09.30.2024, $30.3 mil. at 09.30.2025). Against: there's no explanation in any filing or on the analyst call, and the context — net price down 24%, new and renewal policies +11.3%, policies in force +1.2% — describes exactly a firm defending volume with wider commercial terms. The deciding test is the 09.30.2026 balance, published November 4.

What actually changed versus the prior analysis. One material thing, and it's positive: on 09.14.2026 the company announced a restructuring of the exclusive agency agreement with AmRisc, effective 07.01.2026. The old contract had a fixed expiration of 01.01.2029 — i.e. the firm's largest single-point-of-failure risk (97% of the commercial-lines business, 100% of condominium revenue, plus premium collection) had a contractual cliff 27 months away. The new agreement has no expiration date, auto-renews annually and requires 48 months' notice plus a minimum four years from notification. The risk hasn't disappeared, but its horizon has moved from 2029 to "at least 2030, with four years' notice." Otherwise: Q2 2026 is the same quarter the prior report read, the price fell from $9.86 to $9.14 (−7.3%), and KBRA raised AmCoastal's financial-strength rating to A from A− and the issuer rating to BBB from BBB− (07.21.2026, Note 18, subsequent events).

Estimated value. The stated owner-earnings base is $70.0 mil. — not CFO − capex, which over the trailing 12 months gives minus $31.0 mil. from post-Milton claims settlement. The external check the method requires: the 3.5-year average of operating cash flow minus capex (2023: −$136.0 mil.; 2024: +$243.5 mil.; 2025: +$71.0 mil.; H1 2026: +$52.8 mil.) is +$66.1 mil./year — within 6% of the stated base, so $70.0 mil. is confirmed by cycle-level cash, not invented. The implied owner-earnings yield, 16.2%, sits between the trailing yield (21.6%, P/E 4.64) and the published forward yield (11.9%, forward P/E 8.39) — consistent with what the market sees, unlike the "−7.2% FCF yield" in the data pack, which is pure calendar artifact. The five-model triangulation gives a margin-of-safety range of −21.6% to +129.2%, median +31.6%; Monte Carlo across 20,000 scenarios gives median intrinsic value $12.96/share, median MOS +41.8%, undervaluation probability 87.0%.

Verdict: INTERESTING — small accumulation maintained (1-2% of portfolio), explicitly conditional on the November 4 report. The thesis doesn't die. The falsifier that triggered was measuring something else; and on the business fundamentals, the only material change in the last month is a structural improvement (AmRisc). But the claim "receivables remain an isolated signal" was wrong and I withdraw it: the level shift is real, has lasted three quarters, and has no published explanation. Sizing stays as before — small, assuming a hurricane year, not a quiet one — and I add a verifiable condition: if premiums receivable as of 09.30.2026 doesn't fall below ~$50 mil., the seasonal explanation dies and the thesis must be rewritten without it.

---

The business and the moat (how it makes money, competitive advantage, durability)

Revenue mechanics. ACIC is a holding company operating through a single regulated insurer, American Coastal Insurance Company (AmCoastal), domiciled in Florida, plus a reinsurance captive (Shoreline Re), a Cayman vehicle reinsuring excess & surplus portfolios (BlueLine Cayman Holdings), an in-house MGA (Skyway Underwriters) and an E&S entity awaiting authorization (ACES Specialty Insurance Company, Arizona, which as of 06.30.2026 had not begun operations — 10-Q Q2 2026, Note 1). The product is multi-peril property insurance for condominium associations, apartments and assisted-living facilities in Florida. As of 06.30.2026 the company had 4,453 policies in force, versus 4,402 a year earlier (+1.2%), on an estimated total insured value of $78 billion as of 09.30.2026 (10-Q Q2 2026, Note 6).

The flow is: gross written premiums $216.3 mil. in Q2 → minus the change in unearned premiums → gross earned premiums $138.7 mil. → minus ceded earned premiums $69.0 mil. → net earned premiums $69.7 mil. The gross cession rate is nearly 50%: half the premium collected goes out to reinsurers. On top come net investment income ($5.4 mil. in Q2), realized gains ($3.3 mil.) and unrealized equity gains ($4.2 mil.), for total revenue of $82.6 mil.

The moat, the real part. It exists and is narrow but sharp: ACIC underwrites Florida hurricane risk on commercial-residential property at an underlying loss ratio of 10.8% of gross earned premiums in Q2 2026 (10.2% in Q2 2025). Ten percent. In a business where Florida's admitted-market competitors went through a string of insolvencies between 2021 and 2023, that's not one quarter's luck — the FY2025 10-K shows underwriting profit every year since inception, and the CEO repeated the claim in the 09.14.2026 press release. Risk selection comes from AmRisc: a general agency managing a network of independent agencies focused strictly on commercial-residential, with over two decades of data on the same buildings. The second moat element is the cost structure: general and administrative expenses were $23.6 mil. in FY2025 on gross earned premiums of $648.3 mil. — 3.6%. There's no in-house sales force, no branch network; distribution is a variable cost paid as a monthly management fee to AmRisc.

The moat, the false part. The advantage isn't a brand and isn't a barrier to entry. It's a contractual relationship with a single counterparty. The FY2025 10-K states it plainly in the risk factors: "a concentration of revenue because we source 97% of our commercial lines business from AmRisc, LLC" (line 1270), and in Item 1: "AmRisc is an unaffiliated third party and represents 100% of our condominium revenue based on our exclusive agreement in Florida" (line 155). Further, "AmRisc handles the underwriting, claims processing and premium collection" (line 211) — underwriting, claims, AND premium collection. ACIC is, functionally, the balance sheet carrying risk selected by someone else. An important clarification the prior report didn't state as sharply: the 10-K explicitly qualifies AmRisc as an unaffiliated third party, so this isn't a related-party transaction in the accounting sense, whatever the Executive Chairman's personal history founding AmRisc.

What changed on September 14 and why it matters more than it looks. Until now, the MGA contract had a fixed expiration of 01.01.2029. An investor buying ACIC was buying an earnings stream with a known expiration date less than three years out. The new agreement removes the fixed date, auto-renews annually and requires 48 months' notice for a convenience termination, with a minimum four years of operation from the notice date. Practically: no one can stop the flow before fall 2030, and only if they notify today. The CEO's phrasing — "puts that partnership on a permanent footing" — is marketing, but the notice arithmetic isn't. For a DCF with 10 years of explicit flows, the difference between "contract expires in year 3" and "contract guaranteed at minimum through year 4, with auto-renewal" is the difference between a model not allowed to use a Gordon terminal term and one that is.

Durability, tested by the cycle. The market is softening for the first time in 14 years. Net price fell 24% year over year (10-Q Q2 2026, Revenue section, both periods). Citizens Property Insurance approved an average statewide reduction of 8.7-8.8% for 2026 and has fallen to ~336,000 policies, −76% from the October 2023 peak, meaning the insurer of last resort is releasing risk back into the private market exactly as 17 new insurers have entered (research brief, sources flvoicenews/mynews13). ACIC responds correctly: it writes more policies (1,705 new and renewal in Q2 2026 versus 1,532, +11.3%) at much lower prices, keeping policies in force nearly flat. It isn't ceding market share, it's ceding margin. Management explicitly says the pressure on rates, deductibles, and acquisition costs continues into 2027.

---

Management and capital allocation (track record, buybacks/dividends/acquisitions, skin in the game)

The team. B. Bradford Martz — CEO and President since February 2025, previously group CFO and president. R. Daniel Peed — Executive Chairman, after transitioning from the CEO role. Troy Crawford — Chief Underwriting Officer since 02.17.2026, with responsibility expanded group-wide (previously ran only AmCoastal underwriting). On the board, Sherrill W. Hudson stepped down after the 2026 annual meeting, and Deirdre Brown (CPA) was nominated for risk oversight (research brief; I didn't find confirmation in the 10-Q, so I flag it as unsupported by the filing).

Skin in the game — enormous. Insiders hold 50.7% of shares (yfinance, heldPercentInsiders = 0.50685), concentrated with R. Daniel Peed and his affiliates. The 10-K lists this as a risk factor in itself: "the ability of R. Daniel Peed and his affiliates to exert significant control over us due to substantial ownership of our common stock, subject to certain restrictive covenants" (line 119). It's simultaneously the best and worst thing in the file. Good: the principal owner bears exactly the same loss as the minority shareholder, in a business where the alternative — management with no skin in the game underwriting tail risk — produced a graveyard in Florida. Bad: at 50.7%, the minority holder has no leverage, and the restrictive covenants mentioned in the 10-K limit what the company can pursue (AmCo and AmCoastal have non-compete and non-solicitation clauses barring activities "customarily performed by managing general agents" — line 450 of the 10-K; i.e. ACIC is contractually barred from becoming its own MGA on the core segment).

Capital allocation over the last 12 months, with figures. 2026 is the first year since the 2023 restructuring in which capital returns to the shareholder:

• Special dividend $0.75/share, $36,574 thousand, paid in Q1 2026 (cash flow statement, H1 2026). In H1 2025: zero.
• Buybacks $19,441 thousand in H1 2026, of which $14,441 thousand in Q2 (1,365,980 shares at ~$10.57 average). Cumulative year-to-date, over 1.8 mil. shares. In H1 2025: zero. Treasury shares rose from 212,083 to 2,016,809.
• Buyback authorization increased on 07.21.2026 by another $25 mil., total $50 mil., of which $30.6 mil. remaining as of 08.06.2026 (Note 18).
• Debt-reduction plan from $150 mil. to $75 mil. over the next 6-12 months, targeting debt/capital ≤20% (Q2 2026 call). This is new information versus the file from a month ago and is directly relevant to the covenant (see the balance-sheet chapter).
• Shares outstanding: 48,764,802 (12.31.2025) → 48,342,811 (03.31.2026) → 47,271,828 (06.30.2026), i.e. −3.1% in six months and −3.0% year over year. After years of dilution, the direction has reversed.

Where the money comes from — and why it's a choice, not a consequence. In H1 2026 the holding company received a substantial dividend from AmCoastal, and the effect shows directly in the statutory balance sheet: AmCoastal's surplus as regards policyholders fell from $316,594 thousand (12.31.2025) to $257,384 thousand (06.30.2026), −18.7% (10-Q Q2 2026, Note 8). Risk capital from the regulated company, converted into a special dividend and buybacks, right before the peak of hurricane season, while total insured value grew toward $78 bn. The company partly offset this through reinsurance — first-event retention fell from $49 mil. to $23.5 mil. as of 08.01.2026, for $8.4 mil. — but the order of operations remains: it pulled the capital out first, then bought the protection.

Contrarian or mechanical buying? Honestly: in between. The Q2 buybacks were made at ~$10.57, i.e. 1.47x book value of $7.21 — below the company's own 1.90x median over 13 quarters (2023 Q1 – 2026 Q1), but above today's price, $9.14 (1.27x book). The company bought cheap relative to its own history and expensive relative to today. Where discipline is clearly contrarian is in the reinsurance purchase, not shares: the 2026/27 CAT program has a $1.918 bn aggregate limit, +14.4% (+$241.5 mil.) versus the prior year, at a provisional total cost of ~$179.5 mil. versus $201.85 mil., −11.1% — more coverage for less money, precisely because the reinsurance market softened. The CEO's formula, "buy more protection when pricing is cost-effective and retain more risk on our balance sheet when it's not," is applied to the decision that actually matters at a catastrophe insurer.

Acquisitions and divestitures. The Interboro (IIC) divestiture closed 04.01.2025 for $25,679 thousand proceeds and a loss of $247 thousand net of tax — portfolio cleanup, not value creation. Expansion is organic and cautiously paced: ACES Specialty is to be capitalized and licensed in Q4 2026 to obtain an AM Best rating, with evaluation of a fronting relationship with Fortegra to give Skyway access to rated paper in the interim. Estimated 2026 E&S premiums were lowered to ~$50 mil. from an initial $50-80 mil. range — guidance cut openly, stated openly.

---

What changed over the last 4 quarters (balance sheet line by line from the data pack, margins, cash conversion — explain EVERY large variance)

The consolidated balance sheet, 06.30.2026 versus 12.31.2025 (10-Q Q2 2026, thousand USD), with an explanation for each large variance:

Item	06.30.2026	12.31.2025	Δ	Explanation
Total investments	332,799	354,890	−22,091	Net sales of $28.0 mil. to fund the special dividend and buybacks; the portfolio was drawn down, not marked down
Cash and equivalents	218,943	198,762	+20,181	June 1 renewal: ceded premium owed, not yet paid
Restricted cash	98,280	94,092	+4,188	$97,914 trust funds (reinsurance collateral, Shoreline Re) + $366 regulatory deposits. Not free cash
Premiums receivable, net	119,915	70,447	+49,468	The anomaly that commissioned this report. See the O'Glove chapter
Reinsurance recoverable	117,435	128,205	−10,770	Continued catastrophe-claims settlement; credit allowance $39 thousand on $117.4 mil.
Ceded unearned premiums	224,760	109,697	+115,063	Purely seasonal: the CAT program starts June 1, so as of June 30 nearly all of it is paid/owed and unearned
Deferred acquisition costs (DAC)	54,693	37,815	+16,878	Same season: commissions on Q2-written policies
Goodwill	59,476	59,476	0	Not tested for impairment interim
Total liabilities	903,006	755,167	+147,839
Unpaid claims and LAE	118,920	165,701	−46,781	Milton/Ian runoff. Composition: only $13,229 case reserves, $105,691 IBNR (88.9%)
Unearned premiums	335,452	249,616	+85,836	Seasonal, the June 1 renewal
Reinsurance premiums payable	203,882	66,841	+137,041	Symmetric with ceded unearned premiums; unwinds over the following months
Notes payable, net	149,519	149,353	+166	$150 mil. senior notes, 6.25%, maturing 12.15.2027
Equity	340,788	317,565	+23,223	Profit $41,150 − dividend $36,574 − treasury $19,441 + SBC $2,842 + OCI
Treasury stock, at cost	(19,872)	(431)	−19,441	The first material buybacks in years

Over four quarters (data pack, XBRL series): total assets $1.35 bn (06.30.2025) → $1.24 bn (06.30.2026), −7.7%; cash $315.5 mil. → $218.9 mil., −30.6%; total liabilities $1.05 bn → $903.0 mil., −14.4%; equity $292.3 mil. → $340.8 mil., +16.6%; accumulated deficit narrowed from −$137.0 mil. to −$73.4 mil. The balance sheet is deflating on both sides — that's what happens when an insurer settles a catastrophe event and collects the recoverable, while writing cheaper premium in parallel. Book value per share rose from $6.00 to $7.21, +20.2%, and +10.7% versus year-end 2025 alone.

Margins, decomposed. GAAP combined ratio Q2: 60.6% → 74.3%, +13.7 points. The decomposition, exactly as in the filing:

• net loss ratio 19.8% → 27.0% (+7.2pp), of which +4.5pp current catastrophes ($3,118 thousand, versus zero in Q2 2025) and +2.7pp reserve-development swing (from −$1,275 thousand favorable to +$767 thousand UNfavorable);
• the underlying loss ratio 21.4% → 21.5% on net premiums, 10.2% → 10.8% on gross premiums. That is, +0.1pp from underwriting;
• the expense ratio 40.8% → 47.3% (+6.5pp), entirely a denominator effect: total operating expenses rose +2.9% ($32,035 → $32,969 thousand) while net earned premiums fell −11.1%.

The prior report's conclusion is confirmed: ACIC isn't underwriting worse, it's selling cheaper. But it's worth adding a nuance operating leverage hides — on GROSS earned premiums, acquisition costs rose from 14.7% to 16.4% (+1.7pp) and total operating expenses from 19.4% to 23.8% (+4.4pp). The cost of acquiring a dollar of premium is rising while the dollar shrinks. That's the definition of a soft market and confirms the CEO's warning that pressure on acquisition costs holds through 2027.

Cash conversion. H1 CFO: +$52,760 thousand in H1 2026, versus +$154,392 thousand in H1 2025. The company explains the difference as "a reduction in premiums collected greater than the reduction in net claims paid," with detail: the change in recoverable fell $83,061 thousand and in ceded unearned premiums $21,255 thousand, partly offset by the change in unpaid claims (−$56,172 thousand) and unearned premiums (−$16,137 thousand). Translated: H1 2025 was an exceptional half-year because massive post-Milton recoverable was being collected; H1 2026 is the normal half-year after. Over 12 months, CFO comes out at −$30.6 mil., producing a mechanical "FCF yield" of −7.2% in the data pack. The figure is arithmetically real and economically meaningless: the TTM window catches claims outflows without the recoverable inflows that fell in Q2 2025. Over 3.5 years, the average is +$66.1 mil./year.

---

Balance-sheet analysis — Quality of Earnings (Thornton O'Glove method)

1. Receivables vs. revenue — the subject that commissioned this report

The full premiums-receivable series, extracted directly from XBRL companyfacts (us-gaap:PremiumsReceivableAtCarryingValue), placed next to the same quarter's gross earned premiums and unearned premiums:

Date	Premiums receivable	Gross earned premiums (Q)	Unearned premiums	Receivable / unearned prem.	YoY Δ receivable
12.31.2024	46.56	—	285.35	16.3%	—
03.31.2025	61.75	162.10	321.11	19.2%	+14.4%
06.30.2025	53.50	165.46	383.99	13.9%	+1.2%
09.30.2025	30.34	162.76	293.05	10.4%	+62.9%
12.31.2025	70.45	157.90	249.62	28.2%	+51.3%
03.31.2026	76.72	141.13	257.88	29.8%	+24.2%
06.30.2026	119.92	138.73	335.45	35.7%	+124.1%

(mil. USD; gross earned premiums = net + ceded, both from XBRL; Q4 2025 derived from annual minus Q1-Q3)

Days sales outstanding, on the prior author's definition (receivable / TTM gross earned premiums × 365), reproduced exactly to be comparable:

Period	TTM gross earned premiums	Receivable	DSO
FY2022	474.2	39.30	30.3
FY2023	604.7	45.92	27.7
FY2024	638.6	46.56	26.6
FY2025	648.3	70.45	39.7
TTM 03.31.2026	627.3	76.72	44.6
TTM 06.30.2026	600.5	119.92	72.9

Four consecutive rising observations, from 26.6 to 72.9 days. The denominator explains little (TTM gross earned premiums fell 7.4%), the numerator explains almost everything (receivables rose 70% versus year-end 2025). Receivables' year-over-year growth, +124.1%, against a 16.2% decline in the quarter's gross earned premiums, is a 140-percentage-point divergence — exactly the pattern O'Glove calls "revenue without collection" and ranks first in his signal hierarchy.

What the cash flow statement says, which can't be adjusted by estimates: the change in receivables consumed $49,468 thousand in H1 2026, versus $8,482 thousand in H1 2025. Broken down by quarter (from the Q1 and Q2 10-Qs): Q1 2026 −$6,272 thousand, Q2 2026 −$43,196 thousand; Q1 2025 −$16,746 thousand, Q2 2025 +$8,264 thousand. Annually: FY2023 +$16,048 thousand (a cash source), FY2024 −$920 thousand, FY2025 −$25,425 thousand, H1 2026 −$49,468 thousand. Escalation over four consecutive periods. This isn't an accounting reclassification; it's cash that didn't come in.

The arguments that it's still NOT an earnings-quality problem, each verified in the filing:

• Credit-loss allowance on premiums receivable: $0 as of 06.30.2026. The Note 17 roll-forward shows opening balance $0, additions $1 thousand, recoveries $(1) thousand, closing balance $0. A year earlier: $26 thousand opening, $(23) thousand, $(3) thousand, closing $0. There's no write-off in any year.
• The policy stated in the FY2025 10-K (line 1191): "We perform a policy-level evaluation to determine the extent to which the balance of premium receivable exceeds the balance of unearned premium." The receivable is contractually capped by its associated unearned premium. At $119.9 mil. against $335.5 mil., coverage is 2.8x. A non-paying insured is canceled, the unearned premium is released, and the economic loss is practically zero.
• AmRisc, not ACIC, does the collecting: FY2025 10-K line 211 — "AmRisc handles the underwriting, claims processing and premium collection." The receivable is, in large part, exposure to a single agency's remittance cycle, not to 4,453 homeowner associations. And the MGA contract was renegotiated exactly during this period, effective July 1, 2026. A change in the settlement calendar between AmCoastal and AmRisc could produce exactly this pattern at the June 30 balance-sheet date. It's a hypothesis, not a fact — the company hasn't commented — but it's the only one that simultaneously explains the quarterly jump and the absence of any credit loss.
• The half-year's operating cash flow remains positive ($52.8 mil.) after absorbing the receivables build.

Verdict on this point: not fraud and not aggressive revenue recognition. Most likely, a combination of wider commercial terms in a soft market and a changed remittance calendar, measured at the worst possible date. But it remains the most serious signal in the file, because it has no published explanation and has now lasted three quarters.

2. "Inventory" — the insurer equivalent: DAC

An insurer has no inventory; the closest analog is deferred acquisition costs (DAC) — capitalized expense that must be recovered from the associated unearned premium. DAC: $58.01 mil. (06.30.2025) → $54.69 mil. (06.30.2026), −5.7%, while the quarter's gross earned premiums fell 16.2%. Against the unearned premiums it must be amortized from: 15.1% → 16.3%. The direction is unfavorable and consistent with what shows in the income statement: acquisition cost per dollar of premium is rising. Not material in amount (a difference of ~$4 mil. if normalized), but it independently confirms the cycle thesis.

3. Debt — structure, maturity schedule, cost, covenants

• Balance: $150,000 thousand principal, $149,519 thousand net of issuance costs. Debt/equity = 0.4476; debt/total capital = 30.6%.
• Cost: 6.25% annually. History: 6.25% at issuance (2017), raised to 7.25% after the December 2022 KBRA downgrade, brought back to 6.25% from 12.16.2025 after the upgrade to BBB−, and on 07.21.2026 KBRA raised it another notch, to BBB. Interest paid H1 2026: $4,688 thousand, versus $5,438 thousand in H1 2025.
• Maturity schedule: a single maturity, 12.15.2027 — 15 months from today. No intermediate amortization.
• Covenant, the critical point: the senior notes prohibit incurring new debt if the leverage ratio exceeds 0.3:1. The filing states it in black and white: "at December 31, 2025, while the Company's leverage ratio was greater than the allowed ratio above, the Company did not incur any additional indebtedness during the period and as a result, the Company was in compliance" (10-Q Q2 2026, Note 9). In other words: at the current leverage level, the company CANNOT refinance the 2027 maturity with new debt. It must either repay it in cash, or bring leverage down first. The plan announced on the call — a reduction to $75 mil. over 6-12 months, targeting ≤20% debt/capital — is exactly the answer to this constraint, and is new information versus the file from a month ago.
• Debt used for dividends/buybacks? Formally no: no new debt was incurred (and none could be). Functionally, the capital distributed in H1 ($36.6 mil. special dividend + $19.4 mil. buybacks = $56.0 mil.) came from an increased dividend from AmCoastal that reduced statutory surplus 18.7%, with the debt intact and due in 15 months. It's a legitimate choice, but worth naming: if $75 mil. of debt is paid down over the next 12 months, that money no longer buys shares.

4. Discretionary expenses and accounting estimates

• G&A: $10,266 thousand in Q2 2026 versus $7,778 thousand, +32.0%. The gross figure is misleading and the filing explains it: Q2 2025 contained a non-recurring employee-retention tax credit of $2,939 thousand ($4,469 thousand for all of 2025). Adjusted, the comparable Q2 2025 base is $10,717 thousand, so comparable G&A FELL 4.2%. This is the classic O'Glove correction: the non-recurring item is in the base period, not the current one, and it flips the sign of the conclusion. The data pack reports "SG&A TTM −1.6% vs revenue TTM −8.6%" without this adjustment.
• Policy acquisition costs: −6.4% ($24,257 → $22,703 thousand), from a $5,105 thousand decline in outside management fees (mechanical, on lower premiums), partly offset by a $2,277 thousand loss of ceding commission (the external quota-share rate fell from 20% to 15% from 06.01.2025) and $1,536 thousand of additional commissions.
• Amortization: down $944 thousand, matching the symmetric decline in "other income" from the IP transaction, which has wound down. Intangibles amortization remains $610 thousand/quarter, constant.
• Accounting policy or estimate changes: none. 10-Q, Critical Accounting Policies section: "During the six months ended June 30, 2026, we reassessed our critical accounting policies and estimates... We have made no changes or additions with regard to those policies and estimates."
• The estimate that matters, reserves: favorable prior-year development compressed from $3,469 thousand (H1 2025) to $899 thousand (H1 2026), and in Q2 2026 was outright UNfavorable ($767 thousand, versus $1,275 thousand favorable in Q2 2025). Reserve composition as of 06.30.2026: $13,229 thousand case reserves and $105,691 thousand IBNR — 88.9% of the reserve is estimate, not reported claims. Not abnormal for a catastrophe insurer mid-runoff, but it means the cushion of favorable releases that flattered past years' results has thinned to near zero.

5. Synthesis — net profit vs. CFO divergence (accruals) over 3 years

Year	Net profit	CFO	CFO − capex	CFO/NP
2023	309.9 (of which 224.7 from discontinued operations)	−136.0	−136.2	n/a
2024	75.7	+243.5	+243.5	3.22x
2025	106.8	+71.0	+70.9	0.66x
H1 2026	41.2	+52.8	+52.4	1.27x
Cumulative, excl. the 2023 gain	308.9	231.3	230.6	0.75x

The Sloan ratio over the TTM window, computed by the data pack: +10.5% of assets ((NP 100.2 − CFO −30.6) / 1,240 mil.), i.e. in the zone the literature flags as poor earnings quality. The correct interpretation, however, differs from the mechanical one: the TTM window (Q3 2025 – Q2 2026) contains post-Milton claims outflows without the recoverable inflows, which fell in Q2 2025 ($127.9 mil. CFO in a single quarter). Over 3.5 years, conversion is 75 cents of operating cash per dollar of GAAP profit — below 100%, but explainable by catastrophe runoff and, in the last year, by the receivables build.

One last divergence, which no consolidated metric shows: AmCoastal's statutory net profit — the regulated entity that actually carries the risk — was $4,065 thousand in Q2 2026, versus $18,219 thousand in Q2 2025 (−77.7%), while consolidated GAAP profit was $21,896 thousand (Note 8). For the half: statutory $24,389 thousand versus $36,511 thousand (−33.2%), GAAP $41,150 thousand. The GAAP-statutory gap doubled in a year (from $8.2 mil. to $17.8 mil. in one quarter). Part of this is structural and legitimate — the Shoreline Re captive (Cayman) and the holding company don't enter AmCoastal's statutory figures, and SAP doesn't capitalize DAC and treats unrealized gains differently. But the economic consequence is real: AmCoastal's dividend capacity to the holding, the only channel by which money reaches the shareholder, is calculated from statutory profit, and that's running at ~$48 mil. annualized and declining, while the holding plans $75 mil. of debt repayment plus buybacks.

Quality of Earnings verdict: MEDIUM. Reported profit is real — confirmed by $231.3 mil. of cumulative operating cash over 3.5 years and by an underlying loss ratio stable at 10.8% of gross premiums. It's not HIGH for three reasons: receivables +124% with no published explanation, the favorable-reserve-release cushion exhausted (development already unfavorable in Q2), and a GAAP-statutory gap widening exactly where it matters for distributions.

Signal to watch at the November 4, 2026 report: the premiums-receivable balance as of 09.30.2026. The seasonal norm is a collapse — $18.6 mil. (2024), $30.3 mil. (2025). If the balance stays above ~$50 mil., the calendar explanation dies and only one remains: permanently wider commercial terms, i.e. a soft-market concession that isn't reversing.

---

CEO profile — Outsider traits (William Thorndike method)

B. Bradford Martz, CEO and President since February 2025 — 19 months in the role. Too new for a full-cycle verdict on capital allocation, and I state this explicitly rather than extrapolate: everything below covers 6 quarters of decisions, of which the first buybacks in seven years happened in the last two. Martz isn't an outside hire, though — he was group CFO and president through the entire 2022-2024 restructuring, so he carries the bad history too, not just the good result.

Capital allocation as priority #1 — partially proven (3/5). In six months: $36.6 mil. special dividend, $19.4 mil. buybacks, authorization raised to $50 mil., a plan to halve debt, first-event retention bought down $8.4 mil. Five capital decisions in a half-year, all announced with figures. What's missing for a higher score: no evidence of a price rule below which they buy and above which they don't. The Q2 buybacks were made at ~$10.57, at 1.47x book; today the stock is at $9.14, at 1.27x book, and we don't know whether the pace has accelerated.

Contrarian vs. mechanical buyback — ambiguous (2.5/5). The first material buyback in seven years, triggered after the price fell from $13.06 (52-week high) — so on a decline, not at highs, which favors it. But there's no track record of stopping when it's expensive, because there's no track record at all. What's clearly contrarian is the reinsurance purchase: +14.4% limit at −11.1% cost, with first-event retention cut from $49 mil. to $23.5 mil. exactly when the reinsurance market softened, articulated as policy: "buy more protection when pricing is cost-effective and retain more risk on our balance sheet when it's not." At a catastrophe insurer, the retention decision IS the capital-allocation decision; the behavior here is textbook.

Decentralization and frugality — strongly proven (4.5/5). Distribution is fully outsourced to AmRisc, as a variable fee. No in-house sales force. G&A $23.6 mil. on $648.3 mil. gross earned premiums in FY2025 = 3.6%. The company went from two external asset managers to one plus an internal manager (August 2025 / April 2026) — a small saving, but the kind Thorndike tracks. Run-off subsidiaries (UIM, Skyway Claims, Skyway Reinsurance Services) were shut down, not maintained.

Independence from Wall Street — medium (3/5). The company gives annual guidance (pretax profit and revenue) — not Outsider purity. But the framework it's stated in is unusually good: the stated goals aren't EPS, but "profitable every quarter even with one full CAT retention lost, and profitable annually even with three full retentions" — i.e. guidance expressed in capital at risk, not the number analysts want. Coverage is thin (1-3 analysts depending on the source; yfinance reports 1), which reduces the pressure anyway. Communication is honest on one verifiable point: at Q2 they cut revenue guidance from $335-365 mil. to $300-320 mil. and lowered the E&S premium estimate from $50-80 mil. to ~$50 mil., instead of letting them quietly lapse.

Skin in the game — maximum, with a flip side (5/5 on alignment, risk separate). Insiders 50.7%, concentrated with R. Daniel Peed, Executive Chairman. Equity incentives vest in equal tranches over three years, and SBC is modest — $2,842 thousand in H1 2026, i.e. 1.8% of the half-year's revenue.

"How much of an Outsider" score: 3.5 / 5. The cost structure, outsourced distribution, and reinsurance-buying discipline are authentically Outsider. Capital allocation to the shareholder is new and promising, but has six quarters of history, not six years.

Key-person risk. Not CEO succession — concentration at Peed. 50.7% of shares means the minority holder has no governance leverage, and the 10-K explicitly lists both the control and "restrictive covenants" between the company and Peed that limit what opportunities the group can pursue (including a bar on AmCo/AmCoastal performing typical MGA activities on the core segment — line 450). The second key-person risk is structural, not individual: 97% of the business comes through a single agency that handles underwriting, claims, AND collection. The 09.14.2026 restructuring pushed that risk further out in time; it didn't eliminate it.

---

Accounting red flags (accruals, dilution, one-offs, accounting-policy changes)

1. Premiums receivable +124.1% year over year with no published explanation. $53.50 mil. → $119.92 mil., while the quarter's gross earned premiums fell 16.2%. DSO (TTM gross earned premiums) 26.6 → 39.7 → 44.6 → 72.9 days across four observations. Cash consumption: $49,468 thousand in H1 2026 versus $8,482 thousand. Severity: medium-high, mitigated by a $0 credit allowance and 2.8x coverage from unearned premiums, aggravated by complete silence in filings and on the analyst call.

2. The favorable-reserve-release cushion, exhausted. Prior-year development: $3,469 thousand favorable (H1 2025) → $899 thousand (H1 2026), and in Q2 2026 unfavorable by $767 thousand. IBNR is 88.9% of the claims reserve ($105,691 of $118,920 thousand). Past years' profit was aided by releases; the aid has stopped. Severity: medium.

3. GAAP vs. statutory divergence, widening. Consolidated GAAP profit Q2 2026 $21,896 thousand; AmCoastal statutory profit, same period, $4,065 thousand (−77.7% year over year). The gap grew from $8.2 mil. to $17.8 mil. in a year. The difference forms legitimately in the Shoreline Re captive and at the holding, but has an economic consequence: Shoreline Re participates in a 45% ground-up quota share with $33,041 thousand capital at risk (Note 6), so "reinsurance" is partly bought from the company's own group — the risk doesn't leave, it moves between pockets, and the ~50% gross cession rate overstates the actual risk transfer outside the group. Severity: medium — not an accounting error, an interpretation trap, plus a real constraint on dividend capacity.

4. The $150 mil. maturity of 12.15.2027 under a covenant that blocks refinancing. The leverage ratio exceeded 0.3:1 as of 12.31.2025; the company stayed compliant only by not incurring new debt. Repayment must therefore be funded from cash or preceded by a leverage reduction — hence the plan to reduce to $75 mil. Severity: medium — $218.9 mil. of unrestricted cash makes repayment feasible, but it consumes capital that would otherwise buy shares.

5. A non-recurring effect in the base period that flips the sign of a reported rate. The employee-retention credit of $2,939 thousand (Q2 2025) / $4,469 thousand (2025) makes G&A look +32.0% year over year when comparable is −4.2%. Not a company misstep — the filing discloses it explicitly — but any automated screen (including this session's data pack) will misread it. Severity: low, flagged for series accuracy.

What I did NOT find, explicitly checked: accounting-policy or estimate changes (stated "none" in the 10-Q); new expense capitalizations; dilution — on the contrary, shares fell 3.0% year over year and treasury shares rose to 2,016,809; excessive SBC ($2,842 thousand in H1, 1.8% of revenue); reinsurance-recoverable collection issues (credit allowance $39 thousand on a $117.4 mil. balance); auditor change; going concern. The UPC litigation (the DFS claim on the $40 mil. D&O policy) improved materially: on 06.25.2026, Florida OIR determined that former UPC directors weren't a substantial contributing cause of the insolvency, and the company has accrued only the $1,500 thousand retention.

DEEPRF;ACIC;creante_vs_venit;0.70;0;Creantele din prime +124,1% an/an ($53,50 -> $119,92 mil.) in timp ce primele brute castigate scadeau 16,2%; DSO 26,6 -> 72,9 zile pe patru observatii consecutive, $49,5 mil. numerar consumat in H1 2026, nicio explicatie in depuneri sau pe apel DEEPRF;ACIC;rezerve_subdimensionate;0.35;0;Dezvoltarea favorabila pe ani anteriori s-a comprimat de la $3.469 mii (H1 2025) la $899 mii (H1 2026) si a devenit NEfavorabila cu $767 mii in T2 2026; IBNR = 88,9% din rezerva de daune ($105.691 din $118.920 mii) DEEPRF;ACIC;parti_afiliate;0.45;0;Captivul propriu Shoreline Re preia un quota share ground-up de 45% cu $33.041 mii capital la risc, deci rata de cesiune bruta de ~50% supraestimeaza transferul real de risc in afara grupului; profitul statutar AmCoastal T2 2026 $4.065 mii vs GAAP consolidat $21.896 mii DEEPRF;ACIC;refinantare_2027;0.40;0;$150 mil. senior notes scadente 15.12.2027 sub un covenant de levier 0,3:1 pe care compania il depasea la 31.12.2025 — nu poate emite datorie noua pentru refinantare, deci rambursarea consuma numerarul care altfel cumpara actiuni DEEPRF;ACIC;accruals;0.25;0;Raport Sloan +10,5% din active pe fereastra TTM (PN $100,2 mil. vs CFO -$30,6 mil.), dar conversia pe 3,5 ani e 0,75x ($231,3 mil. CFO cumulat vs $308,9 mil. profit excl. castigul din 2023) — artefact de calendar al decontarii post-Milton, nu de recunoastere

---

Triangulated valuation (conservative DCF with explicit assumptions + earnings power value + 5-year historical multiples + Monte Carlo from step 5; range, not point)

The FCF bridge — the chosen definition and why

The bridge in the data pack, for fiscal 2025:

  CFO (`NetCashProvidedByUsedInOperatingActivities`)          71,025,000
  − Capex PP&E (`PaymentsToAcquirePropertyPlantAndEquipment`)     155,000
  ───────────────────────────────────────────────────────────────────────
  = Simple FCF                                                70,870,000

Interest paid, $10,875,000, is reported in OPERATING (InterestPaidNet), so it's already deducted from CFO — it isn't deducted a second time. No interest line in investing/financing, no material finance lease ($95 thousand principal in 2024, absent in 2025). The divergence between simple FCF and FCFE is under 1%: at ACIC, interest placement changes nothing.

The real problem is different and specific to insurers: CFO − capex measures float timing, not earning power. The series speaks for itself — 2023: −$136.0 mil. · 2024: +$243.5 mil. · 2025: +$71.0 mil. · H1 2026: +$52.8 mil. · TTM as of 06.30.2026: −$31.0 mil. A $380 mil. swing between two consecutive years, at a company with $330 mil. of revenue, driven by hurricane claims settlement and reinsurance recoverable collections. It's not a quantity you can build a DCF on.

Definition chosen: FCFE = core income normalized for catastrophe load and favorable reserve releases, anchored to the company's own guidance. The anchor: earnings before income tax of $85-100 million, midpoint $92.5 mil.; effective tax rate 24.0% (H1 2026: $13,006 / $54,156 thousand) → $70.3 mil., rounded to $70.0 mil.

The mandatory external check, in two directions:

1. Against cash realized over the cycle: the average CFO − capex over 2023 – H1 2026 (3.5 years) = ($−136.2 + $243.5 + $70.9 + $52.4) / 3.5 = +$66.1 mil./year. The stated base of $70.0 mil. is 6% above the realized cash average over a window that includes a major hurricane year. Confirmed, not invented.

DEEPFCF;ACIC;FCFE = core income normalizat pentru sarcina CAT si eliberarile de rezerve, ancorat pe ghidajul pre-tax al companiei;70.0;CFO-capex oscileaza intre -136,2 si +243,5 mil./an din decontarea daunelor si a recuperabilelor (TTM -31,0 mil.), dar media 2023-H1 2026 e +66,1 mil./an, la 6% de baza declarata — deci baza e confirmata de numerar pe ciclu, nu de o marja mecanica

Net debt and access to cash (λ)

The balance sheet gives net cash of $159.3 mil. (cash, equivalents, restricted, and investments $650.0 mil. minus debt $150.0 mil. and adjustments), i.e. 36.9% of market cap — well above the 15% threshold at which the method requires the explicit question. The script issued the warning; here's the answer.

I ran with nd = 0, not nd = −159.3. I didn't apply λ, because nd = 0 is already the equivalent of λ = 0, i.e. the most conservative assumption possible. The reasons, in order of importance:

1. Double counting. The $332.8 mil. portfolio produces net investment income of $10.5 mil. per half-year, ~$21 mil./year, which is already included in the profit from which the $70.0 mil. owner earnings is derived. Capitalizing the flow AND adding the asset that produces it means counting the same money twice.
2. It's not shareholder cash. Of the $650.0 mil., $98.3 mil. is explicitly restricted ($97,914 thousand trust funds as reinsurance collateral + $366 thousand regulatory deposits, Note 4), and the rest is regulatory capital covering $335.5 mil. of unearned premiums and $118.9 mil. of unpaid claims — $454.4 mil. of insurance liabilities. The arithmetic check the method requires: net cash of $159.3 mil. does NOT exceed equity of $340.8 mil., so we're not in the AFL/SDR case where extraction would be counting policy reserves; but the economic test is stricter than the arithmetic one, and fails it.
3. The real access channel is statutory and narrow. The only way money reaches the minority holder is the AmCoastal dividend to the holding, capped by statutory surplus and RBC. Statutory profit is running at ~$48 mil. annualized and declining, and surplus has already fallen 18.7% in six months.
4. No double-taxing governance. I didn't add extra governance points to r for Peed's 50.7% control; the access penalty is taken once, here, through nd = 0.

The five models

All use the same two-phase engine as the tracker (intrinsic_value: 5 years at g1, 5 years at g1/2, then Gordon), with 48.5 mil. shares — derived, not guessed: 47,271,828 shares outstanding as of 06.30.2026 plus the difference between diluted and basic Q2 average (49,303,289 − 48,094,148 = 1,209,141 shares of overhang from unvested equity instruments).

#	Model	OE	g1	r	gt	Value/share	MOS
1	DCF FCFE bear	48.0	−2.0%	13.0%	1.5%	$7.17	−21.6%
2	DCF FCFE base	70.0	0.0%	12.0%	2.0%	$12.89	+41.1%
3	DCF FCFE bull	85.0	+4.0%	11.5%	2.5%	$20.95	+129.2%
4	EPV Greenwald	70.0	—	12.0%	—	$12.03	+31.6%
5	Own multiples, 13 quarters	—	—	—	—	$11.18	+22.3%

Model 1 (bear). The soft market doesn't stop in 2027 but continues into 2028; net price falls another 15%, operating expenses stay sticky, and owner earnings falls to $48 mil. and then shrinks 2%/year in phase one. r raised to 13% for the extra uncertainty. Terminal weight: 31%. Result: the stock is 21.6% expensive at $9.14. This is the scenario that matters for sizing.

Model 2 (base). Identical to the center of the Monte Carlo simulation. Versus the prior report I lowered g1 from +1.0% to 0.0%: the CEO explicitly said on the 08.05.2026 call that rates, deductibles, and acquisition costs remain under pressure into 2027, revenue guidance was cut ~10% ($335-365 → $300-320 mil.), and the 2026 E&S contribution was lowered to ~$50 mil. Zero growth in phase one means: 2027 down, 2028-2030 recovering, net flat. r stays at 12% — a deliberate choice, not inertia: the AmRisc restructuring (48 months' notice, minimum 4 years, no expiration date) materially reduces the single-point-of-failure risk that justified the risk premium; and KBRA raised AmCoastal to A and the issuer to BBB; but the three-quarter receivables build, the covenant blocking 2027 refinancing, and 100% Florida concentration add back exactly what was subtracted. Terminal weight: 37% — below the 60% fragility threshold.

Model 3 (bull). The cycle turns in 2028 (Florida soft markets have historically lasted 3-4 years), ACES Specialty gets its AM Best rating in Q4 2026 and scales E&S beyond $50 mil., and the share count keeps falling 3%/year. OE $85 mil., g1 +4%, r 11.5%. Terminal weight rises to 43% — still acceptable, but this is the model where most of the value is an assumption about year 11.

Model 4 (EPV Greenwald). Sustained earning power, with not a dollar of growth: $70.0 mil. / 12% = $583.3 mil. enterprise value, $12.03/share. I don't add excess cash, for the reasons under λ above. Translated into a multiple: EPV implies 1.67x book value of $7.21. The company trades today at 1.27x.

Model 5 (own multiples). I reconstructed the company's own P/B over 13 quarters, from XBRL equity and closing price at the balance-sheet date: 1.23x (Q1 2023) · 1.56 · 2.43 · 2.43 · 2.27 · 2.06 · 1.90 · 2.50 · 2.02 · 1.75 · 1.60 · 1.82 · 1.64 (Q1 2026). Median 1.90x, minimum 1.23x, maximum 2.50x. Those multiples were paid on a 25-36% ROE. The sustainable ROE implied by $70.0 mil. on $340.8 mil. equity is 20.5%, and shrinks as capital accumulates. So I apply not the median, but a point near the bottom of the own-range, 1.55x → $11.18/share. Even so, that's 22.3% above the price: the market today pays 1.27x, below the 13-quarter minimum.

Range, not a point: MOS from −21.6% to +129.2%, median +31.6%.

DEEPDCF;ACIC;-21.6;31.6;129.2;5 DEEPDCFM;ACIC;1;DCF FCFE bear;-21.6 DEEPDCFM;ACIC;2;DCF FCFE baza;41.1 DEEPDCFM;ACIC;3;DCF FCFE bull;129.2 DEEPDCFM;ACIC;4;EPV Greenwald;31.6 DEEPDCFM;ACIC;5;Multipli P/B 13 trimestre;22.3

Monte Carlo — 20,000 scenarios

Reference price: 9.14 · median intrinsic value: 12.96

Percentile	P5	P10	P25	P50	P75	P90	P95
MOS (%)	-14.7	-4.5	+14.9	+41.8	+74.2	+111.2	+137.0

• Probability the stock is undervalued (MOS > 0): 87.0%
• Probability of a margin of safety above 30%: 61.0%
• Central assumptions: OE 70 · g1 0.00% · r 12.00% · gt 2.00% · net debt 0 · shares 48.5 mil. · FX 1
• Dispersions: OE ±25% (lognormal), g1 σ=3.0%, r σ=1.0%, gt σ=0.50%
• Access to net cash (λ): not applied — nd = 0, so there's no net cash to cut; the conservatism is already in the input

Interpretation — what the width says, not the median. The P10–P90 interval spans from −4.5% to +111.2%, i.e. 116 percentage points. This isn't the distribution of a predictable company. What widens it: owner-earnings dispersion (±25%), which on a $70 mil. base means scenarios between $52 and $94 mil. — and that range is realistic, because the company's own guidance ($85-100 mil. pretax) doesn't include an above-normal hurricane season, and first-event retention is $23.5 mil. with the second at $25 mil. and an aggregate deductible. A year with three moderate storms moves OE from $70 to under $30 mil. without anyone calling the event "catastrophic." Sensitivity to g1 (σ=3%) contributes just as much, and it's honest: nobody knows when the softening stops.

What P(undervalued) = 87.0% means for position sizing — and what it doesn't mean. It doesn't mean "87% chance of a gain." It means that, if sustainable owner earnings really is $70 mil. ±25% and r really is ~12%, then in 87% of plausible configurations the $9.14 price is below value. The whole statement hangs on the initial condition, and the initial condition is a single-state, single-channel company with hurricane tail risk that isn't captured in any parameter's standard deviation in the model. P10 at −4.5% is the practical anchor: in one out of ten parametric scenarios you lose, but the scenario that really hurts — three storms in one season — isn't among the 20,000, because the model is a DCF, not a catastrophe model. That's why sizing stays at 1-2% of portfolio, not a function of P(undervalued).

DEEPMC;ACIC;-4.5;14.9;41.8;74.2;111.2;87.0;20000

DEEPIN;ACIC;70;0;0.12;0.02;0

EXPO;ACIC;uragan-florida;1.0;100% din primele brute sunt proprietati comercial-rezidentiale in Florida, cu TIV $78 mld. si retentie de prim eveniment $23,5 mil.; un singur sezon coreleaza ACIC cu orice alt asigurator sau reasigurator expus in Florida din plan, iar sectorul "Financial Services" si tara "US" nu separa asta

---

Pre-mortem (why the thesis could be wrong — 3 concrete scenarios)

Scenario 1 — A year with three moderate storms, not one big one. The thesis assumes a normalized catastrophe load implicit in the company's guidance. The reinsurance structure makes that number fragile on the upside, not the downside. GAAP retention on the first event is $23.5 mil. from 08.01.2026 (down from $49 mil.), on the second $25 mil., on the third $2 mil., calculated on three $100 mil. events. The program covers a 1-in-286-year return on a single occurrence — but a season with three moderate events could cost $50-60 mil. retained without the main tower ever being touched and without a single headline about a "catastrophic hurricane." That's nearly a full year of owner earnings. AmCoastal's statutory surplus is already 18.7% below the 12.31.2025 level ($257.4 mil.), after the increased dividend to the holding, and TIV has grown toward $78 bn. What would invalidate the thesis: a quarterly report with retained catastrophe claims above $30 mil., or two consecutive quarters with operating margin below 5%.

Scenario 2 — The softening doesn't stop in 2027 and operating leverage works in reverse. This is the scenario management itself flags. Net price fell 24% year over year; net earned premiums −11.1% in Q2; operating expenses +2.9%. If rates fall another 15% in 2027 and 8% in 2028, net earned premiums reach ~$230 mil. on an operating-expense base of ~$130 mil., and the combined ratio crosses 85% even with a perfect underlying loss ratio. Owner earnings falls toward $45-48 mil. — Model 1, which gives MOS −21.6%. Then the stock isn't cheap, it's fairly valued or expensive, and the discount to book is the signal, not the opportunity. What would invalidate the thesis: two consecutive quarters with net margin below 18%, or a quarter with revenue falling over 22% year over year — i.e. below the floor of the company's own guidance.

Scenario 3 — Receivables don't retreat and the seasonal explanation dies. My central hypothesis on the point that commissioned this report is that the $119.9 mil. balance as of 06.30.2026 is a calendar peak amplified by wider commercial terms, and that 09.30.2026 will bring it back below $50 mil., as every year ($18.6 mil. in 2024, $30.3 mil. in 2025). If that doesn't happen, only one interpretation remains: ACIC bought volume in a soft market by extending trade credit, i.e. converted part of written premium into financing for the insured. Not fraud, and probably not a source of credit losses — but it structurally lowers float yield, adds ~$70 mil. of permanent working capital, and makes the 20.5% ROE I use in Model 5 too high. What would invalidate the thesis: premiums receivable above $50 mil. as of 09.30.2026, or dso (the checker's metric) above 250 days for two consecutive quarters.

---

Verdict versus the tracker's GBL score (convergence/divergence and why)

Prior GBL score: 60.2% — MONITOR (Graham 6.5 / Buffett 7.0 / Lynch 5.0 = 18.5 of 30, F-Score 7.0, denominator 59.0), evaluated 09.06.2026, source rescor_dovezi. Indicators as of 09.16.2026: price $9.14, RSI14 35.0, below EMA50 and EMA200, BVPS $7.302 (yfinance; the filing gives $7.21), disc_gn 0.9683, disc_consens 0.4223, consensus target $13.00, ROE 0.3165, net margin 0.3031, D/E 0.4476.

The convergence — and it matters more than the divergence. Both verdicts say the same thing about sizing: "MONITOR" at 60.2% and "small accumulation, 1-2%, not a conviction position" are the same sentence in two languages. The convergence is also at the source of the penalty: Lynch 5.0 of 10 is the weakest component and it's correct — Lynch scores growth, and ACIC is a company with declining revenue, with guidance cut from $335-365 to $300-320 mil. and with rates under pressure through 2027. Not a model error; it's exactly what my own analysis says at Model 2 when it sets g1 = 0.

The divergence, and where it comes from. My triangulation gives a median +31.6% MOS and Monte Carlo +41.8%, while the tracker's disc_gn gives 0.9683 — an enormous Graham discount. The two don't contradict each other: both say "very cheap on assets and on current earnings." What pulls the score down to 60.2% are three mechanical penalties a catastrophe insurer incurs by construction:

1. Negative FCF yield (−7.2%), which is a calendar artifact of post-Milton settlement, demonstrated in the valuation chapter via the 3.5-year average of +$66.1 mil./year. The tracker has no way to know this.
2. Negative revenue growth, which is real and which I don't dispute.
3. Implied volatility and technical position (HV30 0.3156, below both moving averages, RSI 35) — which describe a stock in a downtrend, which is true: $13.06 52-week high, $9.095 low, today $9.14, i.e. practically at the low.

Where I disagree with the tracker. nr_analisti = 1 and a consensus target of $13.00 are too thin a base for any inference; the research brief finds between 3 and 8 analysts depending on the source, with an average target of $14.00-14.28, and couldn't even reconcile the two price-target cuts attributed to the same Raymond James analyst (14→13 and 16→15). I don't use the consensus anywhere in the valuation. bvps = 7.302 from yfinance differs from the filing value ($340,788 thousand / 47,271,828 shares = $7.2093); I used the filing figure throughout.

The final verdict, and exactly what this report changed versus the one from 08.16.2026.

INTERESTING — small accumulation maintained (1-2% of portfolio), with a verifiable condition added.

What the re-analysis changed:

1. The dso falsifier that triggered was measuring something other than what the thesis believed — premiums receivable PLUS reinsurance recoverable, not premiums receivable. It had been triggered since the day it was written, on data that hadn't moved since. The thesis doesn't break on it, and the new falsifier set recalibrates it to the checker's own units, not mine (threshold 250 days, two consecutive quarters — see below why exactly there).
2. I withdraw the claim that the receivables signal "remains isolated." It isn't isolated: it's the fourth consecutive rising observation and the third level shift relative to unearned premiums (13.9% → 28.2% → 29.8% → 35.7%). Not fatal — the credit allowance is zero, coverage is 2.8x, half-year operating cash flow is positive — but it's a soft-market concession, not an accident.
3. I add a structural positive the prior report had no way to see: the 09.14.2026 restructuring of the AmRisc agreement removed the 01.01.2029 expiration and replaced it with auto-renewal, 48 months' notice, and a minimum four years from notification. The firm's largest single-point-of-failure risk was pushed from "27 months" to "at least 2030." Plus the 07.21.2026 KBRA upgrade (AmCoastal to A from A−, issuer to BBB from BBB−).
4. I lower g1 from +1.0% to 0.0% based on management's explicit statement that rate pressure holds through 2027, revenue guidance cut ~10%, and the E&S estimate lowered to ~$50 mil. r stays at 12%: AmRisc de-risking exactly offsets the receivables build and the 2027 maturity.
5. Estimated value is practically unchanged, but the price fell 7.3% — so the median margin of safety rose from +20.0% (5 models, 08.16) to +31.6%, and the Monte Carlo one from +36.4% to +41.8%. The improvement comes from price, not the business. Worth stating explicitly, so it doesn't look like I found something new of value.

Arguments for, briefly: 4.6x trailing profit and 1.27x a book value that grew 20.2% year over year; a stable 10.8%-of-gross-premiums underlying loss ratio; underwriting profit every year since inception; the first buyback in seven years, ongoing, with $30.6 mil. authorized remaining; distribution now on a permanent contractual footing; insiders at 50.7%.

Arguments against, equally concrete: 100% Florida on $78 bn TIV with $23.5 mil. first-event retention; 97% from a single channel; statutory surplus 18.7% lower right at the start of hurricane season; $150 mil. due in 15 months with no ability to issue new debt due to the covenant; the reserve-release cushion exhausted, with development already unfavorable in Q2; premiums receivable doubled with no explanation; and a pricing cycle management itself says has at least a year more of decline.

The condition added, to apply on November 4, 2026: if premiums receivable as of 09.30.2026 stays above ~$50 mil. (the seasonal norm being $18.6 mil. in 2024 and $30.3 mil. in 2025), the calendar explanation falls and the position closes or drops to observation status, no matter how large the discount looks — a large discount is often the symptom, not the opportunity.