KINS — Kingstone Companies, Inc.

Deep-value report: KINS — Kingstone Companies, Inc. (REFRESH)

Date: 2026-09-04 · Price: … (close 03.09.2026, yfinance) · Market cap: …M · Shares outstanding: 14,476,133 (10-Q cover, 04.08.2026) · Sector: Personal P&C Insurance (Nasdaq Capital Market)

Regime: REFRESH over the reference thesis from 2026-07-27-deep-KINS.md (38 days). Mechanical trigger: the shares falsifier. Major new filing: 10-Q Q2 2026 (filed 07.08.2026), plus the 8-Ks from 07.05, 16.07, 23.07, 06.08, 07.08 and 24.08.2026 and a SCHEDULE 13D from 12.03.2026 — all downloaded and read for this report (the SEC folder had no 2026 10-Q before this session; the standard downloader only pulls fiscal-year-closing quarters, so I manually grabbed king-20260630.htm and king-20260331.htm from EDGAR).

What I inherit unchanged from the 27.07.2026 report (I did not re-read the full dossier): the 2020–2023 crisis and recovery history, the corporate and distribution structure (700+ independent producers, top 25 brokers = 39% of GWP), the moat assessment, management and board biographies, the pre-13D shareholder structure, the crisis-debt timeline (12… notes, fully repaid by 25.02.2025), and the NY regulatory risk profile (the November 2025 DFS hearing, the legislative bills from the 10-K).

What I fully re-derive (moved by the new filing): normalized earning power, the FCF bridge, all five valuation models, the Monte Carlo simulation (the old one was corrupted — see below), the balance-sheet chapter, the Quality of Earnings chapter and the CEO profile chapter (both missing from the prior report), red flags, and the verdict.

Three process errors, corrected here, flagged explicitly before any figure:

1. The shares falsifier is NOT broken — it’s a units error in the mechanical triage. The delta pack reports the metric shares = 14,476,133 against the threshold > 16.5 and declares the thesis broken. The threshold is expressed in MILLIONS (the assumption “dilution beyond ATM above 16.5 million shares”), the metric in units. The actual share count outstanding at 04.08.2026 is 14,476,133, i.e. 14.48 million — below the threshold, so the claim holds. Moreover, the direction has reversed: in H1 2026 the company issued no shares through the ATM (remaining unused capacity …) and started buying back. Redoing the analysis was still useful, because the Q2 10-Q changes the thesis materially — just not for the reason flagged by the triage.
2. The Monte Carlo simulation in the delta pack is corrupted by a units error. The prior run was given --oe … (thousands of dollars), while mc_dcf.py interprets oe in MILLIONS — so it simulated a company with … billion of owner earnings. Result: median intrinsic value …/share and median MOS ** …**, with “undervaluation probability 100.0%.” That figure must not be read as information. I re-ran from scratch, in millions. See the valuation chapter.
3. The web research brief delivered for this session (research-KINS-20260904.md) is about a different company. The smaller model analyzed Kinsale Capital Group (NYSE: KNSL) — an … billion E&S insurer, priced ~… with a 75.5% combined ratio and a short-seller report from The Bear Cave — not Kingstone Companies (Nasdaq: KINS, …M market cap, a New York homeowners insurer). No figure from the brief is used in this report. The market and consensus context here comes from SEC filings and the Q2 2026 press release (Exhibit 99.1 to the 8-K from 06.08.2026), read directly.

Executive summary

Kingstone Companies underwrites homeowners insurance (homeowners, dwelling fire, co-op/condo, tenants, umbrella) and physical damage for delivery fleets, through its subsidiary Kingstone Insurance Company (KICO, founded 1886, the 11th-largest homeowners writer in New York in 2025). 98.6% of direct premium in H1 2026 comes from New York — geographic concentration remains the defining feature of the business, despite entering California in June 2026 and licensing the Kingstone America Insurance Company (KAIC) subsidiary in Connecticut on May 1, 2026.

What changed since the July 27 thesis. The prior report normalized the combined ratio to 90% starting from an “88.3% ex-cat underlying” read off Q1 2026 in isolation, and concluded the stock was fairly valued (median MOS …, verdict “monitor”). The Q2 2026 10-Q shows that normalization was too pessimistic by ~10 percentage points. The figures now reported, with explicit non-GAAP reconciliation in the press release: underlying combined ratio (excluding catastrophes AND prior-year reserve development) of 73.7% in Q2 2026 and 80.7% for H1 2026 — versus 82.0% in H1 2025. GAAP combined ratio in Q2: 70.2%, the most profitable quarter in company history, with record net profit of …M and annualized ROE of 50.8%. The difference between H1 (90.2%) and Q2 (70.2%) is entirely catastrophe-driven: …M of losses from the January–February 2026 winter storms, 12.0 points on the H1 loss ratio, dominated by an extended period of sub-zero February temperatures that produced a wave of frozen-pipe claims.

The second driver, equally important: the personal-lines quota-share cession dropped from 16% to 5% effective January 1, 2026 (30% for California only, where the company itself recognizes higher risk). The effect is mechanical and large: net earned premium rose … in Q2 and … for H1, even though direct written premium rose “only” 18.7%/…%. The company now keeps nearly all the underwriting margin it used to hand to reinsurers in exchange for a ceding commission (ceding-commission income halved, from …M to …M for H1). Further out: the quota-share treaty expires January 1, 2027 and provides zero coverage for the period January 2 – June 30, 2027 — in the first half of 2027 the company retains 100% of personal-lines risk, subject to XOL and catastrophe protection.

Capital allocation has changed qualitatively. On May 19, 2026, the board authorized the first buyback program in company history — up to 1,000,000 shares (6.9% of capital) over two years. In June, 19,446 shares were bought at an average of … — near the bottom of the range, not at highs. On July 23, the quarterly dividend was raised 20%, from … to … a year after reinstatement. The ATM was not touched at all in 2026. On July 1, 2026, a …M unsecured revolving line was opened at Citizens Bank (SOFR…, maturity 2028, undrawn) — needed because, as of June 30, 2026, KICO could upstream only … to the holding company without DFS approval, and the holding had just capitalized KAIC with …M.

New risks. The auditor changed for the second time in 17 months: Marcum → CBIZ CPAs (April 2025) → Deloitte & Touche (August 18, 2026). CBIZ had issued an ADVERSE opinion on internal control as of 12.31.2025, for lack of a SOC 1 Type 2 report at the premium-rating platform and at the general-ledger system. Management states in the 10-Q that the weaknesses were remediated in Q2 2026, but that remediation effectiveness will only be demonstrated by 12.31.2026, and disclosure controls remain declared ineffective as of June 30, 2026. Separately, the Fortunoff family (Gregory + Scott) moved from a 13G to a SCHEDULE 13D on March 12, 2026, with ~1.09 million shares held jointly (~7.5%), bought with personal funds for …M — “investment purpose,” but the 13D preserves activist optionality.

Estimated value and verdict. Triangulating five models gives a range of … – … per share, with a median of … (… margin of safety) versus … and a DCF base case at … (…). Monte Carlo over 20,000 scenarios (correctly re-run) gives a median MOS of ** …**, with an … probability of undervaluation, but also a P10 of … — the dispersion is real, not a model error, and comes from the fact that a Northeast-coast-concentrated homeowners insurer swings 15–25 points of combined ratio between a good year and a bad one.

Verdict: MONITOR, leaning toward accumulation on weakness — an improvement over the pure “monitor” from July, but not a firm buy at … The case for: normalized earning power is ~30% higher than I estimated in July, reserve development is favorable and rising (not falling, as I had flagged), dilution has stopped and reversed, and the price is practically unchanged. The case against: with 98.6% NY concentration, aggressively rising retention (5%→0% quota share), a material internal-control weakness not yet demonstrated remediated, and a third auditor in 17 months, a P/B of 2.27x leaves no room for error. The level at which I’d move from monitor to buy: …–17 (close to the conservative EPV and to the price at which the company itself bought back).