A round trip through September, and a quiet October rebound

Allstate closed at $247.88 on 16 October 2026, 10.6% below its 52-week high of $277.22 set in July and about 32% above its 52-week low of roughly $188 from late 2025. On a dividend-adjusted basis the shares are up about 18% since September 2025 month-end, slightly ahead of the S&P 500 (SPY) at about 16%. The path was anything but smooth: the stock rose about 29% between the May and July month-ends after a 47% first-quarter earnings beat and a run of lighter storm months, then fell 14% in September (including a 5.2% one-day drop on 22 September) as heavy July and August catastrophe losses ($682M and $748M) met growing talk of softer auto pricing. October has recovered about 12% as September catastrophe losses came in light ($220M).

Exhibit 1 · Price & relative performance

Ahead of the market over twelve months, with a September air pocket

ALL vs. S&P 500 (SPY) · dividend-adjusted month-end close · rebased to 100 at Sep 2025

Allstate S&P 500 (SPY)

Source: Alpha Vantage (monthly adjusted close); latest point = 16 Oct 2026 close · earnings dates: 5 Nov 2025, 4 Feb, 29 Apr, 5 Aug 2026 · next: 4 Nov 2026

Exhibit 1 shows a stock that has tracked catastrophe headlines more than fundamentals. The two big up-moves (November 2025 and May to July 2026) followed reports in which adjusted EPS beat consensus by 36% and 47%, and the September slide came with no change in reported numbers, only the monthly cat releases and the September FOMC hike. Over the full twelve months, trailing adjusted EPS rose from $21.05 (four quarters to June 2025) to $45.12 while the shares returned about 18%, so the trailing multiple compressed from about 10x to 5.5x. The market is treating the earnings jump as cyclical, which in our view is the right instinct; the debate is about how far it reverses.

Price = 16 Oct 2026 close ($247.88). TTM = four quarters to 30 Jun 2026 (revenue $70.1B; net income to common $13.19B; adjusted net income $11.89B; P-L premiums earned $59.0B; underwriting income $11.56B). Share basis: 253.5M common shares outstanding at 30 Jun 2026 (the latest reported count; repurchases have continued since). Market cap = shares × price. "Enterprise value" adds $7.49B debt and $2.05B preferred liquidation value; for an insurer EV and EBITDA multiples are not meaningful because debt funds the holding company, not operations, and the investment portfolio backs claim reserves. Book value per share is the company-reported figure; ex-AOCI adds back $193M of accumulated other comprehensive losses; ex-goodwill removes $3.12B of goodwill only. Underlying combined ratio is premium-weighted across the four quarters. Beta computed from 60 monthly returns (Oct 2021 to Sep 2026, correlation with SPY 0.09); it understates the stock's real risk, which is catastrophe and pricing-cycle risk the market index does not capture. Cost of equity ~9.5–10.5% is an assumption (10-yr Treasury 5.26% plus a 4–5 point premium), not CAPM. Consensus EPS and analyst counts: Alpha Vantage estimates (21–23 analysts), shown in amber. Rating and targets: Barchart (24 analysts: 10 Strong Buy, 10 Hold, 4 Strong Sell) and Benzinga, September 2026; ranges differ by source. Institutions: 13F filings, Q1 2026 (other sources show 59% to 85%). Short interest: Equibles and Benzinga, Aug–Sep 2026. Holding company capital: management on the Q2 2026 call. Statutory surplus, RBC and dividend capacity: 2025 Form 10-K (company estimates at 31 Dec 2025; AIC = Allstate Insurance Company, the main operating insurer; ACL = authorized control level). Operating cash flow is shown without a yield: for an insurer it includes premiums collected before claims are paid.

The numbers everyone quotes, and where they mislead

1. A 5x P/E at a cyclical peak. GAAP TTM P/E is 4.9x and adjusted 5.5x. Both rest on a year with heavy reserve releases and below-normal catastrophes. On consensus 2027 earnings the multiple is 8.8x, and Allstate's own 2022 and 2023 show what the trough looks like: adjusted losses. Low multiples on peak earnings in a cyclical business are a warning, not a signal. 2. "Revenue +11.8%" is mostly not premium. Q2 total revenue included $1.06B of net investment gains, which swing with markets. Property-liability premiums earned grew 4.0%, and premiums written, the leading indicator, grew 2.6%; auto written premiums grew only 0.4%. 3. "Adjusted" removes investment marks but keeps reserve releases. Adjusted EPS of $8.99 in Q2 excludes $4.07 of investment gains and $0.57 of pension gains (good), but includes about $2.09 a share of favorable prior-year reserve development (our estimate), which is real profit but not repeatable at will. 4. Even the "underlying" auto ratio is flattered. The underlying auto combined ratio of 87.6 in Q2 excludes prior-year reserve changes, but the company disclosed it still benefited from 2.4 points of favorable development on claims reported in Q1 2026; adjusted for that, it was about 90.0. 5. Operating cash flow is float, not owner earnings. Operating cash flow less capex was $12.2B over four quarters, but an insurer's operating cash flow includes premiums collected before claims are paid; it is not distributable cash, and we do not quote a yield on it. The cash measures that matter are statutory earnings, what the insurance subsidiaries can pay up to the parent, and what the parent returns (Section 3). 6. Policy growth has a catch-up in it. Total policies in force rose 3.8% to 215.9M, but the June count includes 2.9M policies relating to September 2025 to March 2026 that had not been reported before, mostly in Protection Services. Property-liability policies grew 2.6%. 7. Data-feed artifacts. Generic feeds show a "gross margin" and a current ratio for Allstate; neither means anything for an insurer, and one feed maps Q3 2025 net income at $2.04B in its cash-flow table versus the reported $3.75B.

Scale, brand and a homeowners franchise, in a business where price is the product

What it does. Allstate insures cars, homes and other personal property across the US, mainly through the Allstate Protection segment (2025 premiums written: auto $38.6B, homeowners $16.6B, plus specialty and commercial lines), sold through exclusive Allstate agents, independent agents (the Custom360 product, now in 41 states) and directly. A smaller Protection Services segment (about $0.9B of revenue a quarter) sells device and appliance protection plans through retailers, roadside assistance, dealer products, identity protection and Arity, a telematics and lead-generation business. In 2025 Allstate sold its Employer Voluntary Benefits and Group Health businesses, booking about $1.6B of disposition gains, so it is now almost purely a property-liability company. Tom Wilson leads the company; Christian Lown became CFO on 3 August 2026.

Revenue drivers, decomposed (Q2 2026 vs. Q2 2025). Auto (about two-thirds of property-liability premium): written premium +0.4% on policies in force +2.8%, so average premium fell. Annualized premium per auto policy was $1,486, down slightly, and management said rate changes had a net impact of zero in Q2 after 2.6% in 2025. New auto business rose 8.8%, credited to "affordability initiatives", distribution and marketing. This is volume bought with price. Homeowners: written premium +8.1%, from a 5.8% rise in Allstate-brand average premium (rate increases and higher replacement costs) and 2.9% more policies, with new business +16.4%. This is price plus volume. Investments: net investment income +33.8% to $1.01B, with market-based income +14.2% on larger balances and higher yields.

Exhibit 2 · Revenue & profit trend

Premiums grew a few percent; profit roughly tripled

Quarterly property-liability premiums earned ($B, bars) · adjusted net income to common ($B, line)

P-L premiums earned Adjusted net income

Source: company earnings releases (Q3 2024 to Q2 2026) · adjusted net income is company-defined (excludes investment gains, pension marks, intangibles amortization, disposition gains)

Exhibit 2 is the whole story in one picture. Premiums earned rose 9% over two years while adjusted net income went from about $1B to $2.3B–$3.8B a quarter. That is operating leverage from price catching up with claim costs, not from a bigger business, and it is the part of the cycle that does not last. Earnings have already rolled over from the Q4 2025 peak ($3.79B, helped by a 1.4-point catastrophe quarter) to $2.33B in Q2 2026 as storms returned.

Competitive position. Allstate is the third-largest US personal lines writer behind State Farm and Progressive by premium. It is gaining share in many states: management cited share gains in auto and homeowners on the Q1 call, and new business is growing at high single to double digits. But rivals are doing the same with deeper margins or bigger ad budgets: Progressive grew policies 6.3% in Q3 at an 88.7 combined ratio, State Farm cut auto rates an average 10% across 40 states, and GEICO's combined ratio rose to 91.2 in Q2 from 83.5 as it cut prices. When every large writer is cutting price at once, customers shop more and retention becomes the battleground.

Moat & pricing power: what the numbers show. Narrow, and the evidence sits in underwriting results rather than in size. Over the past ten years Allstate's auto combined ratio averaged 95.2 versus 100.3 for the industry, and homeowners 92.0 versus 102.2 (company figures, Q2 2026 investor presentation): a durable 5 to 10 point margin advantage, largest in homeowners, where catastrophe modelling, pricing segmentation and reinsurance buying matter most. The cost side is less distinctive: the property-liability expense ratio was 21.4 in 2025, of which advertising was 3.6 points (up from 1.3 in 2023), against GEICO's 14.6 in Q2 2026 (definitions differ by company). Growth is being bought: auto new business rose 8.8% in Q2 while policies in force grew 2.8% and written premium 0.4%, with annualized premium per auto policy at $1,486. Allstate does not publish a numeric retention ratio; management described retention on the Q2 call as having "stabilized". Bundling penetration is not disclosed. On the evidence available, the advantage is in risk selection and homeowners underwriting rather than in a lower cost base or in pricing power. Auto is a regulated, rate-filed product that customers can compare in minutes, and pricing power has appeared only when the whole industry was short of margin, as in 2023 to 2024.

Peers on the same page

n/d = not disclosed in the sources reviewed. Allstate: Q2 2026 release and 2025 10-K. Progressive: September 2026 monthly release (combined ratio and policies in force; personal auto = agency + direct auto) and Q2 2026 results via Zacks (book value per share $59.05, ROE); P/B = $218.49 close on 16 Oct 2026 ÷ $59.05. GEICO: Berkshire Hathaway Q2 2026 10-Q as summarized by actuary.info. State Farm: 2025 results release (26 Feb 2026); auto combined ratio derived from $52.6B incurred claims plus $14.1B expenses on $71.3B earned premium. Periods differ by company and are shown in each cell.

The table supports the margin-cushion argument, with a caveat. Allstate's recorded Q2 auto ratio was 5 to 10 points better than the three largest competitors, but 6.6 of those points were prior-year releases; excluding them it was about 90, still better than GEICO's 91.2 but close to Progressive's company-wide 88.7, while Progressive is growing auto policies at more than twice Allstate's rate. Progressive trades at 3.7x book on a 32.5% ROE; Allstate at 2.0x on a higher but more reserve-dependent ROE.

Concentration. No customer concentration, but heavy concentration by product and peril: US personal auto is about 65% of property-liability premium, and homeowners carries most of the catastrophe exposure (in Q2 2026, $1.41B of the $1.72B catastrophe total was homeowners). Severe convective storms (wind and hail) drive most losses; 2026 Q3's $1.65B came from dozens of events, about half of August's from one storm. The January 2025 California wildfires made Q1 2025 the heaviest catastrophe quarter in the period shown in Exhibit 3 (15.7 points).

Real earnings, but a fifth of them came from re-estimating old claims

How to read an insurer's quality of earnings. The combined ratio splits into three pieces: the underlying margin on this year's business, catastrophe losses (volatile, largely weather), and prior-year reserve development (changes to estimates of claims from earlier years). Underlying margin is the repeatable part. Exhibit 3 separates them.

Exhibit 3 · Earnings quality

The underlying margin improved, then stalled; releases and calm weather did the rest

Property-liability combined ratio (lines, left axis) · catastrophe and favorable reserve effects (bars, pts, right axis)

Recorded CR Underlying CR Catastrophe pts Favorable prior-year pts Q3'26 cat pts (est.)

Source: company earnings releases (reconciliation of combined ratio to underlying) · favorable prior-year = non-catastrophe reserve re-estimates (negative = strengthening) · Q3'26 cat points = $1.65B disclosed losses ÷ ~$15.0B estimated premiums earned (newsletter estimate)

Exhibit 3 makes three points. First, the underlying combined ratio fell from about 83 in 2024 to a low of 76.6 in Q4 2025 and has since drifted back to 79 to 80: the largest benefit from prior rate increases appears to have been realized, and the underlying margin is now stabilizing or normalizing rather than improving. Second, reserve releases went from slightly adverse in 2024 to 5.4, 6.9 and 4.6 points in the last three reported quarters. Allstate lowered prior-year auto reserves by $838M in Q1 2026 alone as 2023 to 2025 claim costs came in below estimates; auto releases total about $1.5B so far this year, and the company says about 51% of the auto injury releases related to accident years 2023 and 2024 and 30% to 2025. Over the past four quarters, non-catastrophe releases were worth about $2.9B pre-tax, roughly $8.70 of the $45.12 in trailing adjusted EPS. Third, catastrophes were unusually light in Q3 and Q4 2025 (3.8 and 1.4 points) and have returned: Q2 2026 was 11.5 points and Q3 2026 will be about 11 on disclosed losses. Through September, catastrophe losses ($4.61B) are broadly comparable with the prior-year period ($4.75B), but that does not by itself establish a normal burden: full-year catastrophe ratios were 11.6, 9.2 and 8.6 points in 2023, 2024 and 2025, and rising property values, rebuilding costs and peril mix mean a simple historical average may understate the forward load. What is clear is that the comparison with last autumn's 3.8 and 1.4 points is hard.

The releases are not a red flag in themselves. They reflect severity that came in below the cautious assumptions set during the inflation spike, and they are audited re-estimates, not accounting choices made in the quarter. But they are finite. Reserves for claims fell from $43.1B at September 2025 to $41.0B at June 2026 while premiums grew, which is consistent with releases and with paying down 2025 wildfire claims. Consensus models generally assume none. If they stop, adjusted earnings lose roughly $2 a share per quarter at the recent pace.

Source: Q2 2026 and Q4 2025 earnings releases (GAAP to adjusted reconciliations). Memo rows are newsletter estimates: reserve effect = disclosed points × premiums earned × (1 − 21%) ÷ diluted shares; catastrophe losses × (1 − 21%) ÷ diluted shares. Columns may not sum exactly due to rounding.

The table shows Allstate's adjustments are narrow and mostly sensible: they strip out investment marks, pension marks and one-off sale gains, and they do not add back catastrophes or stock compensation. In 2026 the adjustments actually lowered earnings (adjusted $19.65 vs GAAP $21.73) because equity markets were strong. The memo rows are the part a reader needs: catastrophes take $9 to $15 a share in a normal year, and reserve releases have been adding $5 a share a half-year.

Exhibit 4 · Cash vs. accrual

Cash flow tracks earnings, with float flattering the quiet quarters

GAAP net income to common vs. adjusted net income vs. operating cash flow · $B · quarterly

GAAP net income Adjusted net income Operating cash flow

Source: company releases (net income, adjusted); Alpha Vantage cash flow statement (operating cash flow) · Q3'24 to Q2'26

Exhibit 4 passes the basic cash test: over the last four quarters, operating cash flow was $12.5B against $11.9B of adjusted and $13.2B of GAAP net income. For an insurer that comparison flatters slightly, because growing premiums are collected before claims are paid, and it says nothing about the adequacy of reserves, which is where insurers' earnings quality problems hide. Favorable development indicates that prior accident-year loss estimates have generally proved conservative relative to subsequent claims emergence, although this does not establish that current reserves are similarly conservative.

Margin bridge. Allstate does not report a gross margin; the combined ratio is the margin, and the full-year 2025 split was a 63.8 loss ratio and a 21.4 expense ratio. TTM underwriting margin was about 19.6% (combined ratio ~80.4): roughly 21 points underlying, minus ~6 points of catastrophes, plus ~5 points of releases. On the underlying side, auto is the swing factor (87.6 recorded underlying in Q2, about 90 after the disclosed current-year benefit), while homeowners underlying rose 2.9 points to 61.5 on higher loss costs even as average premiums rose 5.8%. The cost pressure is visible in macro data: the September CPI shows motor vehicle maintenance and repair up 5.8% a year while motor vehicle insurance prices fell 5.0%. Expense discipline helps: the "Affordable, Simple, Connected" products and AI tools are designed to cut acquisition and claims costs.

Working capital. Not a meaningful signal for an insurer. The items to watch are premium installment receivables ($11.9B, growing with premiums) and the reserve-to-premium relationship discussed above.

Exhibit 5 · Balance sheet & returns

Book value per share up 75% in seven quarters; returns now rolling over

Book value per common share ($, bars) · adjusted ROE, trailing twelve months (%, line) · quarter-end

Book value per share Adjusted ROE (TTM)

Source: company earnings releases · adjusted ROE uses common equity excluding unrealized gains and losses

Balance sheet stress points. Few on the liability side. Exhibit 5 shows book value per share rising from $70.35 to $123.38 since September 2024, almost entirely from retained earnings. Debt is $7.49B (18% of capital, 23% including $2.05B of preferred) and interest coverage is about 44x; statutory capital is covered in Section 3. The exposure is on the asset side: an $87.8B portfolio of which $60.8B is fixed income and $11.2B public equities (up from $8.4B at year-end).

Rates: a simplified duration sensitivity. Allstate's 10-K shows a +100bp parallel rise in risk-free rates would have cut the fair value of its interest-sensitive assets by about $3.0B pre-tax at year-end 2025 (duration 5.1 including derivatives), and a +100bp widening in credit spreads by about $2.1B. Duration has since lengthened, to about 5.7 years in Q1. The 10-year Treasury rose about 64bp from the end of June to late September and about 82bp to early October. Scaling the company's disclosure linearly gives roughly $1.9B to $2.8B pre-tax, or $6 to $9 a share after tax. This is not an estimate of the actual Q3 change in book value: it ignores curve shape (the move was not parallel), credit spreads, convexity and mortgage prepayment behaviour, hedges, portfolio turnover, security-level tax effects and the equity book. It shows the order of magnitude: rates alone could offset a large part of a quarter's earnings in reported book value, while leaving the equity base used for adjusted ROE (which excludes unrealized gains and losses) largely unchanged. The other side is higher reinvestment yields, already visible in market-based investment income (+14% in Q2).

Catastrophe reinsurance: what Allstate keeps before protection starts

The program renewed on 1 April 2026 protects the balance sheet against a single very large event far more than against a heavy year of mid-sized storms, and that is the distinction that matters for the normalization debate.

Sources: Artemis and Reinsurance News on the 1 April 2026 renewal (30 Apr–1 May 2026); Allstate 2025 10-K (reinsurance recoverables, probable maximum loss); Q2 2026 10-Q (recoverables at 30 June). The new US aggregate's terms are as reported by trade press and have not been confirmed in a company filing. Florida's separate program renews mid-year and is not detailed here.

Two implications follow. First, severe convective storms, which drove most 2026 losses, arrive as many events that each fall below the $1.0B occurrence retention, so they are largely retained; Artemis estimates losses in the aggregate risk period since April at about $3.15B through August (roughly $3.4B with September added), still short of the $4.78B aggregate attachment. Second, the tail is well covered: Allstate puts its modelled 1-in-100 probable maximum loss for hurricane, earthquake and wildfire at about $3.1B net of reinsurance (year-end 2025), but that figure excludes convective storms. A heavy storm year therefore flows mostly into earnings, which is why the catastrophe sensitivity in Section 6 matters more than the size of the tower. Litigation over claims handling and auto bodily injury trends remain industry risks; GEICO reported bodily injury severity up 10% to 12% in the first half.

Excess capital going back through buybacks, sized to a peak year

ROE vs. cost of equity. For an insurer, return on equity is the right analogue of ROIC: equity is the capital regulators and rating agencies require to write premium. Exhibit 6 shows how cyclical it is. Adjusted ROE was 16.9% in 2021, negative in 2022, about 1.5% in 2023 (our estimate), then 26.8%, 38.3% and 44.2% on a trailing basis. The five-year average is about 16.4%, against a cost of equity we put at 9.5% to 10.5% given a 5.26% 10-year yield. That spread is positive through the cycle but far smaller than today's figure suggests.

Exhibit 6 · ROE vs. cost of equity

A through-cycle spread of about 6 to 7 points, not 34

Adjusted ROE by year (%) vs. ~9.5% cost of equity (assumed) and 5-yr average

Adjusted ROE Cost of equity (est.) 5-yr avg 16.4%

Source: company releases (2021, 2022, 2024, 2025, LTM Jun 2026); 2023 newsletter estimate (adj. EPS $0.95) · cost of equity assumed

Exhibit 7 · Capital returned & share count

Buybacks restarted in 2025 and quadrupled in a year

Repurchases + common dividends ($M, stacked bars) · common shares outstanding (M, line) · quarterly

Buybacks Dividends Shares out

Source: Alpha Vantage cash flow statement (repurchases, dividends); company releases (shares outstanding at quarter-end)

Buybacks & dividends. Allstate essentially stopped repurchasing in 2023 and 2024 to rebuild capital, then restarted: $1.2B in 2025 and $1.7B in the first half of 2026, including $1.05B in Q2. The board approved a $4.0B program in February 2026 after the prior $1.5B program finished, and $2.6B remained at 30 June. A straight-line pace to February 2028 would be about $500M a quarter, so the $1.4B used in roughly four and a half months is ahead of that pace; the CEO said on the Q2 call the company is "committed to completing" it. Exhibit 7 shows the share count falling 3.9% in a year, net of dilution. The heaviest quarter (Q2) came with the stock at roughly $205 to $237 between the March and June month-ends, below today's price, and the quarterly dividend was raised 8% to $1.08 in February. Cash returned over the last twelve months was about $3.6B, 31% of adjusted net income and a 5.8% trailing shareholder yield; dividends plus the remaining authorization spread evenly to February 2028 would be about 4.2% a year.

Capital strength: how much can come back through a bad year

The useful question is not whether Allstate has capital but how much it can return while keeping its ratings (Allstate Insurance Company's financial strength rating is A+ at A.M. Best, outlook stable), its regulatory capital, and enough surplus to grow premiums through a stressed catastrophe year. The disclosed pieces:

Sources: Allstate 2025 10-K (Regulation; Capital Resources and Liquidity); Q2 2026 call. Surplus and RBC are company estimates at 31 Dec 2025; "~4.9x" is our division. Statutory net income is inferred: the 2026 ordinary-dividend formula uses the greater of 2025 statutory net income or 10% of surplus, and $7.98B is far above 10% of surplus. Rating agency capital models (A.M. Best BCAR, S&P) are not public.

An illustrative stress, using our estimates: a 2023-like catastrophe year (11.6 points) instead of the 9.5-point base adds about $1.3B of pre-tax losses on 2027 premiums; adding a modelled 1-in-100 hurricane, earthquake or wildfire event ($3.1B net) takes the shock to about $4.4B pre-tax, roughly $3.5B after tax. That is about 15% of statutory surplus and well under the holding company's ~$9.5B of deployable assets, so on disclosed figures the remaining $2.6B buyback looks fundable even in a bad year. Two things the disclosures do not show: the capital rating agencies require to keep the A+ rating, which is the binding constraint in practice, and exactly how much surplus premium growth will absorb (roughly $1B a year at 4% to 5% growth and a 2.6x premium-to-surplus ratio, our estimate).

M&A and portfolio moves. The 2021 acquisition of National General (non-standard auto, independent agents) left $3.1B of goodwill and has been folded into the independent-agent strategy; the legacy National General homeowners book has been shrunk. Allstate exited life and annuities in 2021 and employer benefits and group health in 2025, simplifying toward property-liability. These have been value-sensible: capital was pulled from lower-return lines and redeployed into the core during its worst years. Reinvestment runway. Incremental capital funds premium growth (each dollar of new premium needs surplus), advertising and technology. Growth at the current ~20% underwriting margin is very profitable; growth bought with lower prices as the market softens will earn less.

No earnings guidance, six large beats, and the beats came from the same places

Allstate does not give EPS or combined-ratio guidance. Instead it publishes catastrophe losses monthly, which lets analysts update estimates in real time. So the "say/do" test has two parts: consensus vs. actual, and stated strategy vs. delivery.

Exhibit 8 · Delivered vs. expected

Reserve releases explain 39% to 90% of each beat; the rest was still positive

Quarterly adjusted EPS · consensus at report date (amber) vs. actual, split into favorable reserve development and everything else · $ · labels = beat excluding releases

Consensus Actual excl. reserve releases Reserve releases, after tax (est.)

Source: Alpha Vantage (consensus; Q3 2026 estimate, 21 analysts, range $4.99–$7.90); company releases (actuals, reserve points, premiums earned) · reserve effect = favorable non-cat points × P-L premiums earned × (1 − 21%) ÷ diluted shares (newsletter estimate)

Newsletter estimates. Diluted share counts approximated as shares outstanding plus ~5M where not reported. Consensus at report date from Alpha Vantage; other providers show somewhat different figures for some quarters (for example Q3 2025 at $7.55), which do not change the sign of any beat.

The decomposition changes the read. Consensus usually assumes no reserve development, so favorable releases explain between 39% and 90% of each beat. Excluding them, Allstate still beat in every quarter, by 5% to 25%. That residual mixes three things the data cannot fully separate: catastrophe losses that differed from what analysts modelled (Q3 and Q4 2025 were exceptionally light at 3.8 and 1.4 points), higher investment income (net investment income rose from $949M in Q3 2025 to $1.01B in Q2 2026), and genuine underlying improvement. The smallest residual came in Q1 2026 (+5%), the quarter with the largest release.

Consensus at report date from Alpha Vantage; other data sources show slightly different figures for some quarters (for example Q3 2025 at $7.55), which do not change the direction of any beat. Buyback used = $4.0B minus $2.6B remaining (management, Q2 call); straight-line pace = $4.0B over 24 months.

Say vs. do. On strategy, management has done what it said: policy growth returned, buybacks are running ahead of a straight-line pace and the portfolio is simpler. On earnings, the decomposition above says the beats came mainly from the components analysts do not model, reserve releases and catastrophe variance, with a smaller but consistent residual. Narrative vs. numbers: the "affordability" message (lower prices for 7.8 million customers in 2025) is consistent with flat auto rates and falling premium per policy, and the margin that funds it is real. What the narrative does not stress is that auto written premium is now growing more slowly than auto policies.

Steady insider selling into the rally, pay tied to cat-adjusted ROE

Insiders. Insiders own roughly 0.5% to 1% of the shares. The Form 4 record over the past twelve months:

Sources: Form 4 filings (Wilson, filed 6 Jan 2026; Rizzo, filed 12 Aug and 3 Oct 2026); Alpha Vantage insider transactions for totals. 10b5-1 status is shown only where stated in the filings named; totals are approximate.

Institutions & positioning. 13F filers hold about 71%, led by BlackRock (8.5%), Vanguard entities (~11.9% combined) and State Street (4.8%), with little change quarter to quarter. Short interest is low at about 2.5% to 2.7% of float and ~4.5 days to cover (August to September), down from earlier in the year. Sell-side views have cooled: UBS and HSBC moved to Hold in July, Argus to Hold and KBW to Underperform in August (KBW target cut to $232 in October), while Raymond James sits at $300.

Compensation. The 2026 to 2028 performance stock awards pay 60% on average performance net income ROE, with catastrophe losses collared to a minimum and maximum and equity measured excluding unrealized gains, and 40% on relative total shareholder return (2026 proxy). That is better aligned than revenue or EBITDA targets: it rewards per-share returns over three years and stops management from being paid for lucky weather. Two caveats: reserve releases still count toward the ROE metric, and adjusted measures also feed the annual incentive.

Cheap on peak earnings; the price sits between through-cycle and current returns

At $247.88, Allstate trades at 5.5x trailing adjusted EPS, 7.0x 2026 consensus and 8.8x 2027 consensus, and at 2.01x book ($123.38 at 30 June). For an insurer whose earnings swing from losses to records, P/E history is close to useless: year-end P/E was 8.7x in 2021, not meaningful in 2022 (loss) and 2023 (147x on $0.95), 10.5x in 2024 and 6.0x in 2025. Price to book is the steadier anchor, provided the return and the book value are measured on the same basis.

Three bases, kept consistent

GAAP and adjusted ROE as reported by Allstate for the twelve months to 30 June 2026. Book excl. AOCI adds back $193M of accumulated other comprehensive loss using the company’s diluted book-value share denominator (approx. 256.9M); at 30 June the two book values differed by under 1%, but the gap will widen if Q3 bond marks are large. Statutory surplus excludes deferred acquisition costs, goodwill and non-admitted assets and sits in the operating insurers, not the holding company, so market cap ÷ surplus is shown for scale only. In the justified-multiple work below, return assumptions are adjusted ROE applied to book value excluding AOCI ($124.13), so numerator and denominator match.

Exhibit 9 · Multiple vs. own history

Price to book sits below its five-year median, but the median was set on depressed book

Price / book value per share at year-end and now · x

Year-end P/B Now 2021–25 median (2.34x)

Source: Alpha Vantage (December month-end close); company releases (book value per share: $81.52, $58.07, $59.39, $72.35, $108.45; $123.38 at 30 Jun 2026)

Versus its own history. Exhibit 9 shows year-end P/B between 1.44x (2021, before the loss years) and 2.66x (2024, when book was still recovering and earnings were inflecting). Today's 2.01x is 14% below the 2.34x median, but that median flatters the comparison: in 2022 and 2023 book value was depressed by underwriting losses and unrealized bond losses, so the market paid a high multiple for a temporarily small denominator.

What the price implies

For an insurer the multiple of book a stock deserves depends on the ROE it can sustain: P/B = (ROE − g) ÷ (CoE − g). Two of the three inputs are assumptions. Cost of equity (CoE) is set at 9.5% to 11.0% (10-year Treasury 5.26% plus a 4 to 6 point premium); CAPM with the measured 0.13 beta would give an implausibly low ~6%. Long-run growth (g) should be tied to the business: book value has to grow roughly with premiums to keep premium-to-surplus stable, and g = ROE × the share of earnings retained. At a 15% ROE, 4% growth means retaining about 27% of earnings and distributing 73% through dividends and buybacks. Allstate distributed 31% of adjusted earnings over the last year because earnings are at a peak; over a full cycle, payout has to rise or book value outgrows the premium base. A 2% to 5% range brackets nominal premium growth over a cycle.

Newsletter estimates. Value = justified P/B × $124.13 book value per share excl. AOCI (30 Jun 2026).

The price implies a sustainable adjusted ROE of roughly 14% to 20%, centred near 17% (CoE 10.5%, g 4%). That is a range rather than a point, and the cost-of-equity choice moves it more than growth does.

Normalized earnings: three cases and a through-cycle anchor

Newsletter estimates. Other personal and commercial lines underlying at 82 (bear), 80 (base), 79 (bull). Line weights from 6M 2026 premiums earned (auto 64.6%, homeowners 28.1%, other 7.3%); 0.3 pts of intangibles amortization added. Pre-tax income = underwriting income + net investment income less ~$160M of other items (Protection Services, interest, corporate; calibrated so the method reproduces TTM adjusted net income of $11.9B); 21% tax; ~$110M preferred dividends. Through-cycle case applies Allstate's ten-year average property-liability combined ratio of 94.8 (Q2 2026 investor presentation) to 2027 premiums with today's investment income. Book path uses FY2026 consensus earnings, then dividends and buybacks near the current pace. Consensus FY2027: $28.11 (22 analysts, $23.55–$34.35).

Exhibit 10 · Normalized earnings range

Trailing EPS sits above even the bull case; consensus sits on the base case

2027 adjusted EPS by scenario (newsletter estimates, amber) vs. FY2027 consensus and trailing adjusted EPS · $

Scenario EPS (est.) FY27 consensus $28.11 TTM adjusted $45.12

Source: newsletter estimates (assumptions in table above); Alpha Vantage (consensus); company releases (TTM)

The base case lands on consensus, which is a cross-check rather than a target: it needs auto underlying margins about a point worse than today and a catastrophe load between 2025's 8.6 points and the nine-month ten-year average. The through-cycle anchor is the more conservative lens: if Allstate simply matches the last decade's average combined ratio with today's investment income, it earns about $23 a share and about 15% to 16% on equity, at the low end of the 14% to 20% the price implies. The base case (~19%) sits at the high end. On these numbers the share price embeds returns somewhere between a full cycle and a mild softening.

Catastrophe sensitivity

Newsletter estimates. Each point of catastrophe load is worth about $1.8 (TTM) to $2.0 (2027) a share after tax. Reference points: Allstate's full-year catastrophe ratio was 11.6 (2023), 9.2 (2024) and 8.6 (2025); its ten-year average for the first nine months of the year, the heaviest part of the year, was 10.3 as of 2025, so a full-year ten-year average would be lower.

No single number is "normal". Rebuilding-cost inflation, higher property values, a peril mix shifting toward convective storm and wildfire, and Allstate's own geographic shifts all argue that history may understate the forward load, which is why the table brackets it rather than picking one.

Other measures

Dividend plus buyback yield: 5.8% trailing ($3.6B returned on a $62.8B market cap), about 4.2% forward if the remaining $2.6B is spread evenly to February 2028. Normalized P/E: 8.9x the base case and 11.0x the through-cycle anchor. Peer P/B: Progressive trades at 3.70x book on a 32.5% ROE and auto policy growth of about 7%; Allstate's 2.0x prices a lower sustainable return. Sell-side: consensus is "Moderate Buy" across about 24 analysts, with mean targets of roughly $265 to $274 depending on the source and a range of $226 to $300.

Late in the auto cycle, early in a rate cycle that cuts both ways

Cycle. Personal auto is moving from a hard market to a soft one. The evidence is broad: the BLS motor vehicle insurance index fell 5.0% year on year in September, its fifth straight year-on-year decline after the first in May; Progressive's monthly combined ratio drifted from 86.8 in July to 90.0 in September; GEICO's Q2 combined ratio rose 7.7 points; and the five largest Florida writers, Allstate included, filed average 8% cuts for 2026. Allstate's margin is still wide (auto recorded combined ratio 83.3 in Q2), so it can compete on price for a while, but the direction of margin is down. Homeowners is earlier in its cycle: rates are still rising mid-single digits to keep up with replacement costs and storms.

Macro. The Fed raised its target range by 25bp to 3.75%–4.00% on 16 September (12–0), its first move since cutting to 3.50%–3.75% in December 2025, citing elevated inflation; the median projection implies one more hike by year-end. September CPI, released 14 October, was 3.7% headline and 2.6% core, above expectations, and one market report put odds of an October hike near 47% afterwards. For Allstate the effects split: higher yields raise investment income on new money (positive), mark down existing bonds and book value (negative near term) and raise the discount rate on its equity (negative for the multiple). Energy-led inflation can also reduce miles driven and claim frequency, a modest potential offset, while repair costs (+5.8%) keep severity rising.

Near-term catalysts. (1) FOMC decision, 28 October. (2) Q3 2026 results, 4 November (call 5 November): cats are known, so the variables are the underlying auto combined ratio, whether reserve releases continue, the size of the AOCI hit to book value, auto rate actions and retention, and buyback pace against $2.6B remaining. (3) October CPI, 10 November, for the motor vehicle insurance and repair indexes. (4) October catastrophe release, typically mid-November. (5) Progressive's October results, 18 November, the best real-time read on industry pricing. (6) End of hurricane season, 30 November; Q3 was a light hurricane quarter for Allstate (September cats $220M).

Second-order tells. Progressive's monthly combined ratio and policy growth; GEICO's Q3 results inside Berkshire Hathaway's filing in early November; the BLS repair, parts and used-car indexes (used cars −1.5% a year in September, a mild help to physical damage severity); state rate filings; and Allstate's monthly policies-in-force data.

What has to go right, and what breaks it

Definitions. Fiscal year = calendar year; Q2 2026 = quarter ended 30 June 2026. TTM = Q3 2025 to Q2 2026. "Adjusted" net income and EPS are company-defined (excluding net investment and derivative gains and losses, pension remeasurement, purchased-intangible amortization, disposition gains and related tax); they include catastrophe losses and prior-year reserve development and do not add back stock compensation. Combined ratio = claims, claims expense and operating costs ÷ premiums earned; underlying combined ratio excludes catastrophes, prior-year non-catastrophe reserve re-estimates and intangible amortization. GAAP ROE uses average common equity; adjusted ROE uses common equity excluding unrealized gains and losses; justified P/B pairs adjusted ROE with book value excluding AOCI. Reserve-release and catastrophe per-share effects apply a 21% tax rate to disclosed points × premiums earned. Q3 2026 catastrophe ratio uses disclosed $1.65B over estimated ~$15.0B premiums earned. Rate sensitivity scales the company's disclosed +100bp sensitivity linearly (simplified; see Section 2). Justified P/B = (ROE − g) ÷ (CoE − g); implied ROE = g + P/B × (CoE − g). Scenario EPS: line-level underlying ratios plus catastrophe and reserve assumptions, investment income, ~$160M other pre-tax items, 21% tax, ~$110M preferred dividends. Beta from 60 monthly adjusted returns vs SPY. Year-end P/B uses December month-end closes. 2023 adjusted ROE is estimated from adjusted EPS of $0.95. RBC multiple, statutory return and the stress illustration are newsletter calculations from company disclosures.

Company sources. Q2 2026 earnings release and Form 8-K (5 Aug 2026), Q2 2026 10-Q, Q2 2026 investor presentation (ten-year combined ratios; accident-year mix of auto injury releases, via Investing.com, Aug 2026), Q1 2026 release (29 Apr 2026), Q4 and FY2025 release (4 Feb 2026), 2025 Form 10-K and annual report (statutory surplus, RBC, dividend capacity, holding company assets, interest-rate and spread sensitivity, reinsurance recoverables, expense ratios, probable maximum loss), Q3 2025 release (5 Nov 2025), Q4 2022 and Q4 2024 releases (book value history), 2026 proxy statement (DEF 14A), monthly catastrophe releases for July (20 Aug), August (17 Sep) and September 2026 (15 Oct), Q3 2026 earnings date announcement (5 Oct 2026), dividend announcement (8 Jul 2026), Form 4 filings (Wilson, 6 Jan 2026; Rizzo, 12 Aug and 3 Oct 2026), Q2 2026 call summaries (GuruFocus, MarketBeat).
Market, peer & macro sources. Alpha Vantage (income statement, balance sheet, cash flow, earnings history and estimates, monthly adjusted prices for ALL and SPY, PGR quote, weekly 10-yr Treasury yield, insider transactions; retrieved 17 Oct 2026); Artemis and Reinsurance News on the April 2026 reinsurance renewal (30 Apr–1 May 2026) and aggregate-period losses (Sep 2026); Federal Reserve FOMC statement (16 Sep 2026); BLS CPI for September 2026 (14 Oct 2026, via InvestingLive and Stock Titan); Progressive September 2026 results (via Finviz, 14 Oct 2026) and Q2 2026 results (Zacks, 14 Aug 2026); GEICO Q2 2026 from Berkshire Hathaway's 10-Q (via actuary.info and Coverager, Aug 2026); State Farm 2025 results release (26 Feb 2026); Florida Office of Insurance Regulation (5 Mar 2026); S&P Global Market Intelligence 2026 P&C outlook (Jan 2026); Tickeron and TradingKey on the 22 Sep 2026 decline; Barchart, Benzinga and StreetInsider for ratings and targets; Equibles and Benzinga for short interest; VCP Scanner and Simply Wall St for institutional ownership (13F, Q1 2026).