Everest Group, Ltd. (NYSE: EG) — The Discount That Paid You to Trust the Reserves Has Gone

Independent fundamental research. Report date: 2026-07-18. Primary sources: SEC filings (10-K FY2020–FY2025, Q1-2026 10-Q, the 64-filing 8-K corpus, DEF 14A, and the full Form 3/4 corpus), Q4-2024 through Q1-2026 earnings calls, company disclosures of Munich Re, Swiss Re, Hannover Re, SCOR and US/Bermuda peers, and broker renewal reports (Guy Carpenter, Howden Re, Gallagher Re, Aon).

⚡ Claude’s Take

This block is the author’s own subjective opinion. It is general information, not investment advice, and you should do your own work before acting on anything here. The analysis that follows takes no position, carries no price target, and confines itself to embedded-expectations and scenario analysis.

Verdict: AVOID here — a fairly-valued commodity underwriter whose margin of safety has just been spent. Accumulate only back toward ~$325–340 (≈0.85–0.88× book), which is precisely where management itself was buying. Fair ≈ $370–390 (≈0.97–1.02× book). Emphatically not a short. Conviction: medium.

Everest is not a bad company, but it is not a good business, and the distinction matters. Over eleven years its ROE has averaged 9.1% against a cost of equity I estimate at ~10% — it earned less than its cost of capital across a full cycle, including at the very top of the hardest reinsurance market in thirty years, printing 9.3% and 10.0% in 2024 and 2025 while peers printed records. That is not a franchise with a moat; Greenwald’s diagnostic puts a sustained 9% ROE squarely in the “advantages absent” band. The structural reason is clean: reinsurance capacity is money, and a cat bond prices in weeks — a capital cycle with no supply lag cannot confer durable excess returns on anyone. Everest’s own mix makes it worse. Its largest single line is property pro rata at 36.3% of reinsurance premium, nearly double its cat XOL at 18.3% — that is balance-sheet rental, where you assume the cedent’s loss ratio and can express alpha only through cedent selection you cannot independently audit. And it owns no distribution: two program administrators supply ~24% of insurance premium, two brokers ~36% of total. Zero goodwill on a $62.5bn balance sheet is the accounting expression of a strategic fact — the customer relationship is not owned. That is exactly why AIG could buy the renewal rights to ~$2bn of premium in a single day for 15% of one year’s premium.

So a ~1.0× book multiple is not an anomaly to be arbitraged; it is roughly what a business earning its cost of capital is worth, and the cohort agrees — across nine reinsurers, P/B tracks delivered ROE almost linearly. The reserve story then removes the margin of safety rather than creating one. The deficiency has migrated forward three times in three reviews (FY2023 blamed AY2016–19 → FY2024 hit AY2019–23 → FY2025 hit AY2022–24), with AY2024 turning adverse on barely one year of maturity — this is not a legacy cohort being quarantined, it is the same excess-casualty book repeatedly under-reserved on a rolling basis. Cumulative adverse development FY2020–25 of $2,379M equals 29.8% of all reported net income over those six years. The $1.2bn adverse development cover is narrower than advertised: its in-the-money first layer was 100.2% consumed within three months, leaving ~$1.0bn of live cover, none of it protecting the $22.7bn Reinsurance reserve — whose current-year loss pick management has left unchanged at 57.3% versus 57.6% in 2023. Meanwhile the one engine still working is repricing down hard: property-cat rate-on-line is −16% year-to-date, the steepest annual decline since the late 1990s, with alternative capital at a record $141bn.

What stops this being a short is that the balance sheet is genuinely strong (zero goodwill, 18.8% debt/capital, no maturities inside five years), the retrenchment is executing, and — the single best bull evidence — everyone with inside information has been buying, at prices far below today’s. The company repurchased $331M in Q1-2026 at an average $330.01 against $393.02 of book ex-AOCI (~0.84×), raising the quarterly floor to $300M. CEO Jim Williamson bought 1,000 shares personally at $337.97 in June 2025 and has never sold. Directors Galtney and Levine bought $3.50M and $0.95M at ~$306–307 on 29 October 2025 — one day after the second charge. 2025 was the largest insider-buying year in the five-year record, and net insider flow is +$3.68M. That is a costly, falsifiable, and genuinely encouraging signal.

But it is also the argument against paying $383. Every one of those informed buyers transacted between $306 and $338; the stock is now 13–25% above all of them, having re-rated from ~0.86× to ~1.01× book in under a month — and roughly the last month of that move is sector-basket beta at a +1.9 z-score, not company-specific news. You are being offered the security at a materially worse price than the people who know it best were willing to pay. Note too that at 1.008× book EG sits at the 59th percentile of its own P/B history: this stock has usually traded below book, so “one times book” is not cheap versus itself. Framing: not a falling knife and not a compounder — a correctly-priced commodity underwriter after a recovery rally, where the discount that compensated you for uncapped casualty tail has been paid away.

What flips me bullish: two to three more quarters of clean development plus an upward revision to the Reinsurance segment’s current-year loss pick (which would signal the picks are finally conservative rather than merely untested), or a pullback to ~0.85× book. What flips me bearish: any adverse development on accident year 2025 — booked at $1,698M ultimate and 85.9% IBNR — or a third strengthening reaching the uncovered Reinsurance book. Tag: “You’re no longer being paid to trust the reserves.”

📈 Stock Price Action — Five-Year Event Map

Over five years Everest has round-tripped a hard-market boom and a self-inflicted reserve bust: from ~$237 in July 2021 to an all-time closing high of $414.59 in November 2023, down to $304.91 in October 2025 after two rounds of US casualty strengthening, and back to $382.57 today — a fresh 52-week high, but still 7.7% below the 2023 peak. The 52-week range is $304.91–$382.57 and the stock trades above all three moving averages (21-EMA $361, 50-EMA $351, 200-EMA $338). Five-year total return is +72.8% (+11.6% annualized) against a price-only +55.0% — the ~$8.00 dividend accounts for the wedge — with a peak-to-trough drawdown of −23.4%.

Cycle narrative. (1) The stock drifted for fifteen months and bottomed at $245.79 on 26 September 2022, days before Ian made landfall; Everest pre-announced $730M of Q3-22 catastrophe losses on 19 October (8-K, Item 7.01). (2) The post-Ian capacity shortage repriced the complex — Everest added ~60% in five months. (3) The re-rating carried to an all-time closing high of $414.59 on 27 November 2023; the May-2023 sale of 4.14M shares at $360.00 cost ~2.5% on pricing day. (4) 2024 was a stalemate as the market began probing US casualty. (5) Hurricane Milton triggered the worst non-reserve day of the window. (6) January 2025 brought the wildfires ($442M), Juan Andrade’s resignation and Jim Williamson’s elevation (8-Ks of 8 and 22 January), and decisively the 27 January 8-K pre-announcing $1.5bn of prior-year plus $229M of current-year strengthening. (7) Twelve months later the problem recurred: Q3-25 carried $478M of further development (Group combined ratio 103.4%) alongside a $1.2bn adverse development cover and the sale of ~$2bn of retail commercial renewal rights to AIG — an 11.4% single-day fall to the five-year trough. (8) The recovery since is fundamental in part: Q1-26 delivered a 91.2% combined ratio and 16.7% annualized operating ROE — though, as shows, 12.2 of the 13.1 points of loss-ratio improvement came from catastrophes rather than underlying attritional performance. Price moves are FACT (AZI price CSV); attributed drivers are INTERPRETATION cross-checked to prints, 8-Ks and renewal commentary. No price target or recommendation here — that judgment lives in Claude’s Take above.

1. Executive Summary

Everest Group is a Bermuda-domiciled global reinsurer (roughly two-thirds of premium) attached to a specialty and E&S insurance platform (roughly one-third) that it built organically from 2015 and has spent the last two years dismantling. FY2025 gross written premium was $17.7bn, down 3.1% — the first shrinkage in the window — and Q1-2026 GWP fell 18.5% year over year as the AIG renewal-rights sale and the casualty exit took effect.

The business does not earn its cost of capital. Across eleven years ROE has averaged 9.1%, clearing 15% exactly once. Against a cost of equity estimated at ~10% (built up as 4.3% risk-free + 5.0% ERP + a 1.0–1.5pt reserve/tail premium; a CAPM reading off the 0.42 beta yields ~6.4% and is not credible), Everest has destroyed or barely preserved value through a full cycle — including at the top of it, printing 9.3% (2024) and 10.0% (2025) while Munich Re earned 18.3%, Swiss Re 19.6%, Hannover Re 21.4% and Arch 17.1%. Everest has run a combined ratio above 100% in three of the last six years.

The reserve problem is the thesis, and its shape is worse than its size. Cumulative unfavorable prior-year development FY2020–FY2025 totals $2,379M — 29.8% of all reported net income over those six years. More important than the total is the migration: the deficient cohort has moved forward in each of three successive reviews (AY2016–19 → AY2019–23 → AY2022–24), with accident year 2024 developing adversely on barely one year of maturity. The three “clean” years that preceded the 2024 confession (FY2021–23 net development of −$9M, −$2M, −$5M) were a netting artifact: the FY2023 10-K discloses $397M of favorable mortgage and short-tail development offsetting $392M of unfavorable 2016–2019 casualty. This is not a quarantined legacy book. It is the same excess-casualty portfolio repeatedly under-reserved on a rolling window.

The protection purchased is narrower than the headline. The October-2025 adverse development cover is presented as $1.2bn. Its in-the-money first layer of $1,250M had $1,253M ceded against it by 31 December 2025 — 100.2% consumed within three months — leaving roughly $1.0bn of genuinely live excess cover. It attaches only to North American Insurance and Other liabilities, accident year 2024 and prior, excluding asbestos and environmental. The $22.7bn Reinsurance reserve is entirely uncovered — and that segment’s current-accident-year loss pick has been left essentially unchanged at 57.3% versus 57.6% in 2023, despite US casualty being the acknowledged problem. Everest also surrendered 50% of any favorable development up to $625M under the profit commission, truncating the upside as well as the downside.

The cycle is turning against the one engine still working. Property-cat rate-on-line is down 12% at 1 January 2026 and 16% year-to-date — the steepest annual decline since the late 1990s — with total reinsurance capital at a record $790bn and alternative capital at a record $141bn. Everest’s own disclosures track it: −10% at 1/1/26, −13% at 4/1/26, “maybe in the mid-teens zone” at 6/1/26, with the decline accelerating. Casualty offers no refuge: US excess and umbrella rate of +6.3% to +10.7% now runs below a 12–15% loss-cost trend, which is margin erosion despite a positive sign.

Against all that, the balance sheet and the capital allocation are genuinely sound. There is zero goodwill on a $62.5bn balance sheet — book value is 100% tangible and carries no impairment risk. Debt is 18.8% of total capital with nothing due inside five years. Prior-year paid losses have been stable at 23.7–24.2% of opening reserves, so the bear argument that cash flow signals reserve stress is simply wrong. And management has repurchased aggressively and cheaply: $797M in FY2025 at ~0.95× book and $331M in Q1-2026 at ~0.86×, with the quarterly floor raised from $200M to $300M.

The investment tension is therefore narrow and specific. At $382.57 the stock trades at 1.008× stated book (0.997× Q1-2026 book) — the cheapest multiple in a nine-name reinsurance cohort, but also the lowest ROE in that cohort, and the 59th percentile of Everest’s own P/B history, meaning this stock has typically traded below book. Across the cohort, price-to-book tracks delivered ROE closely enough that Everest’s multiple is roughly what its ~10% return deserves. The entire bull case reduces to one proposition: that normalized ROE is materially above 10%. This article takes no position and sets no price target; the sections below frame the embedded expectations and the falsification tests that resolve it.

2. Business Overview

What Everest does. Everest assumes reinsurance globally and writes specialty and excess-and-surplus insurance directly. Through FY2025 it reported two segments — Reinsurance and Insurance — plus an “Other” run-off bucket created in Q4-2024. Effective 1 January 2026 it re-segmented again into Reinsurance Treaty, Global Wholesale & Specialty, and Legacy, following the AIG disposal. FY2025 gross written premium was $17.7bn, split roughly $12.8bn Reinsurance and $4.8bn Insurance.

How it makes money. Two engines, of which only one has worked. The first is underwriting: premium less losses and expenses, measured by the combined ratio. Everest’s has been 102.9% (FY20), 97.8%, 96.0%, 90.9%, 102.3% and 98.6% (FY25) — an underwriting loss in three of six years. The second is investment income on float: $643M (FY20) rising to $2,124M (FY25) on a $45.4bn portfolio. The investment engine has carried the company. Net investment income now materially exceeds cumulative underwriting profit over the period.

The line-of-business mix is the most under-appreciated fact about Everest. Its single largest line is property pro rata at 36.3% of reinsurance GWP — nearly double property catastrophe excess-of-loss at 18.3%. This matters enormously and is rarely discussed. Proportional (pro rata) reinsurance is balance-sheet rental: the reinsurer takes an agreed share of the cedent’s premium and its loss ratio, paying back a ceding commission. There is no independent pricing act. Underwriting alpha is expressible only through cedent selection — and under the doctrines of “utmost good faith” and “follow the fortunes,” the reinsurer reserves on the cedent’s own data, which it cannot independently audit. Everest is correspondingly under-weighted in catastrophe excess-of-loss, which is the one place in reinsurance where genuine modeling alpha lives and where RenaissanceRe has built its franchise.

Distribution is rented, not owned. The FY2020 10-K discloses that “two program administrators accounted for approximately 12% of the Company’s gross written premium each” — roughly 24% combined — with the largest single program at 7% of overall GWP. Broker concentration is similar: the ten largest brokers supply ~55% of GWP, with Marsh & McLennan at ~20% and Aon at ~16%. Everest has never acquired an MGA, a program administrator or a broker; the terms “managing general agent” and “MGA” return zero hits in the FY2025 10-K.

Recurring vs. non-recurring. Reinsurance treaties renew annually and are re-tendered through brokers each year. There is no contractual lock-in, no multi-year revenue backlog, and no switching cost of consequence. Roughly all revenue is “recurring” in the trivial sense that it renews, and none of it is recurring in the sense that matters — that the customer would find it costly to leave. The demonstration is empirical: when Everest declined to match price, GWP fell 18.5% in a single quarter.

The third-party capital platform is immaterial. Mt. Logan cell income was $7M in 2025 and $8M in 2024. RenaissanceRe, by contrast, earned $328.9M of fee income on ~$7.6bn of partner capital. Everest has essentially no capital-management franchise and none of the gross-to-net ROE leverage that lets RNR be a must-have size in placements while committing less of its own balance sheet. Any comparison that treats Everest and RenaissanceRe as similar business models is wrong.

Verdict. A large, competently-run, broker-distributed reinsurance balance sheet with a heavy proportional tilt, no owned distribution, no meaningful fee annuity, and an insurance platform now being sold and run off. It is easy to understand, and what one understands is that the economics belong to whoever supplies the capital, not to a franchise.

3. Industry Dynamics

Reinsurance is the canonical capital-cycle industry — with one fatal modification. Marathon’s framework holds that capital destruction creates scarcity, scarcity creates pricing power, pricing power draws capital, and returns mean-revert. That cycle rewards incumbents because supply is lumpy and slow: three years to build a ship, nine to open a mine. Reinsurance capacity is money, and a cat bond prices in weeks. A capital cycle with essentially no supply lag cannot confer a durable excess return on anyone — which is the structural explanation both for Everest’s 9.1% eleven-year ROE and for the violence with which the whole cohort’s returns mean-revert.

Where we are: past the peak and descending quickly. The hard market topped at the January-2025 renewals. The evidence for the down-slope is now overwhelming and comes from five independent sources:

• Guy Carpenter’s Global Property Catastrophe Rate-on-Line Index fell 12% at 1 January 2026 and is down 16% year-to-date after the April and June/July renewals, with APAC down 19%. This is the largest annual decline since the late 1990s — steeper than any single year of the 2010s soft market, though the index remains ~32% above its 2017 trough.
• Howden Re put 1/1/26 property-cat at −14.7% (the largest since 2014), retrocession at −16.5% and direct-and-facultative at −17.5%. Gallagher Re reported a −10% to −20% range (Global/North America −15%, UK −20%, France −15%, Australia −15%, Germany −12%).
• Everest’s own disclosures track it precisely: −10% at 1/1/26, −13% at 4/1/26, and guided to “maybe in the mid-teens zone” at 6/1/26. The rate of decline is accelerating. Asked how far pricing could fall, CEO Jim Williamson declined to answer: “I don’t want to give anybody any ideas.”
• Swiss Re’s January-2026 renewal disclosure is the single cleanest datum in the file: USD 12.4bn of premium renewed flat versus expiring, with price +0.3% but loss assumptions +4.6% — a net price decrease of 4.3%. Headline rate was roughly flat while expected losses rose; margin was given away in a form a naive “rates are flat” reading misses entirely. Rate adequacy, not rate change, is what matters.
• Both European majors are guiding FY2026 down. Munich Re guides a P&C reinsurance combined ratio of 80% against 73.5% delivered; Swiss Re guides net income of USD 4.5bn against USD 4.76bn delivered and a P&C Re combined ratio below 85% against 79.4% delivered. Hannover Re guides below 87% against 84.0% delivered and has raised its large-loss budget to EUR 2.3bn. When the three largest and best-capitalized reinsurers in the world all guide down, the softening is not a matter of opinion.

Capital is flooding back — the textbook late-cycle signal. Total reinsurance capital reached a record $790bn at Q1-2026, with alternative capital the sole driver of growth at a record $141bn (+4% in a single quarter, per Aon). Catastrophe bond issuance set records: 83 transactions in H1-2026, and Q2-2026 at $11.3bn was the first quarter ever above $11bn, with monthly records in February and March. Benign catastrophe experience compounds it — Hannover Re’s FY2025 large losses came in EUR 375M under budget and Swiss Re’s nat-cat losses were USD 813M against a USD 2.0bn budget, so 2025’s excellent industry results reflect light losses rather than durable underwriting improvement.

Casualty offers no refuge, and this is the point most often missed. The industry is bifurcated — property softening, casualty hardening — which invites the conclusion that a diversified reinsurer can rotate. The numbers refute it. US excess and umbrella rate rose +6.3% to +10.7% in H1-2026, decelerating from +12–16% a year earlier, against a loss-cost trend of 12–15%. Positive rate below loss trend is margin erosion. A casualty writer congratulating itself on “still-positive rate” is losing ground. Industry casualty adverse development hit a record $15.8bn, and even W.R. Berkley — the cohort’s most respected reserving culture — reported adverse development in its primary Insurance segment on auto and umbrella, naming social inflation. The drivers are structural: third-party litigation funding, nuclear verdicts, forum shopping and the erosion of tort reform. Everest’s own 8-K names all of them.

Structure and barriers. Property-cat excess-of-loss is a genuine oligopoly gated by rated balance-sheet capital, catastrophe-modeling capability and cedent relevance. Proportional casualty reinsurance — Everest’s largest exposure — is materially more fragmented and commoditized. The A.M. Best A+ rating is a ticket of entry, not a barrier: every meaningful competitor holds A+ or better, and a qualification shared by all significant competitors is a cost of doing business, not an advantage. Bermuda’s 15% corporate income tax took effect 1 January 2025, ending the zero-tax era; its 2023 enactment produced Everest’s $578M one-time deferred-tax benefit.

Verdict: a structurally mediocre industry, currently at the wrong point in its cycle. Better than commodity manufacturing on barriers, but fundamentally an industry where the scarce input is capital, capital is infinitely mobile, and the supply response is measured in weeks. The three-year window in which returns looked attractive is closing, and the closing is accelerating. This is a bad time to own a levered claim on reinsurance pricing, and Everest is on the wrong side of both halves of the bifurcation: retrenching from casualty (where rate is below trend) into property (where rate is falling 16%).

4. Competitive Position

The honest answer: there is no durable competitive advantage. This is the central verdict of the report, it is uncomfortable given Everest’s size and history, and the evidence supports it plainly.

The financial test comes first, because of our framework requires that a moat surface in financial outcomes. Everest’s ROE by year: 2015 11.9% · 2016 11.0% · 2017 5.1% · 2018 0.9% · 2019 10.2% · 2020 4.9% · 2021 12.4% · 2022 5.0% · 2023 19.1% · 2024 9.3% · 2025 10.0%. The mean is approximately 9.1%, and exactly one year in eleven cleared 15%. Against a ~10% cost of equity this is not an excess return; it is a shortfall. And the timing is damning: the 2023–2025 window was the hardest reinsurance market in thirty years, and Everest earned 19.1%, 9.3% and 10.0% — while Munich Re earned 18.3%, Swiss Re 19.6%, Hannover Re 21.4%, Arch 17.1% and W.R. Berkley 20.6% in 2025 alone. A company that cannot earn an excess return at the top of its own cycle does not have a moat; it has a balance sheet. Even the 2023 high point is overstated: $578M of it was the one-time Bermuda deferred-tax benefit, which reduces normalized FY2023 EPS from $60.19 to roughly $46.37 and ROE from 23.3% to 17.9% on the recomputed average-equity basis.

Greenwald’s taxonomy, applied honestly.

Customer captivity — absent. Reinsurance treaties are re-tendered annually through brokers. There is no habit, no search cost, no switching cost. The empirical demonstration is Everest’s own: the moment it declined to match price, GWP fell 18.5% year over year. Business that leaves in one renewal cycle was never captive.

Network effects — none. No participant’s value rises with the number of other participants. Reinsurance is bilateral risk transfer intermediated by brokers.

Economies of scale — present but insufficient, because they fail the mandatory conjunction. Greenwald is explicit that scale confers advantage only in combination with customer captivity; scale without captivity simply invites competitors to build the same scale. Everest does have one genuine cost advantage: a reinsurance other-underwriting-expense ratio of 2.5%, which is best-in-class. But it is competed away through a 25.2% commission ratio — the consequence of the proportional mix — and never reaches the ROE. A cost advantage that cannot be traced to a financial outcome is not a moat; per our own standard, we should say so, and we do.

Intangibles / rating — a ticket, not a barrier. Everest holds A.M. Best A+. So does essentially every meaningful competitor. Worse, the rating has become a vulnerability rather than a protection: A.M. Best revised the outlook to negative on 29 October 2025 (affirming A+), citing “elevated uncertainty surrounding the group’s business profile and enterprise risk management capabilities.” S&P moved negative in February 2025. A rating agency questioning enterprise risk management capability — not merely reserve levels — attacks the only quasi-barrier the business has. This is the mechanism by which a reserve problem becomes a franchise problem.

The market-share stability test fails at both ends. Greenwald’s most practical diagnostic is whether shares are stable: stable shares imply barriers, volatile shares imply their absence. Everest’s GWP ran $9.1bn (2019) → $18.2bn (2024) → $17.7bn (2025) → a Q1-2026 run-rate near $14.4bn. Doubling in five years and then surrendering a fifth within eighteen months is the signature of a balance sheet rented at the clearing price, not a defended franchise. Share was bought on the way up by accepting price, and lost on the way down by declining it.

The accounting corroborates the strategic verdict from an independent direction. Everest carries zero goodwill and zero intangible assets — verified by exhaustive full-text search of six 10-Ks. That is genuinely good for book-value quality, but it also encodes a strategic fact: the customer relationship is not owned. Everest rents ~24% of insurance premium flow from two program administrators and ~36% of total GWP from two brokers. This is exactly why AIG would pay $301M for the renewal rights to ~$2bn of premium, and exactly why the transaction could be signed and closed in a single day — there was no entity, no licence and no reserve transfer to negotiate. A franchise whose distribution can be sold in a day, for 15% of one year’s premium, is by definition not a franchise with customer captivity.

Direct comparison. The cohort, all at 17 July 2026 close, using operating ROE where available:

Methodological note that must travel with this table: Munich Re, Swiss Re, Hannover Re and SCOR report under IFRS 17, which permits discounting of claims reserves and flatters the combined ratio by roughly 4–8 points versus undiscounted US GAAP. Munich Re’s 73.5% is not 25 points of better underwriting than Everest’s 98.6%. The ROE comparison is the fair one — and Everest is last.

Two further asymmetries deserve emphasis. First, three of the four US/Bermuda peers were releasing reserves while Everest was strengthening them — RNR harvested $1,091M favorable (roughly 11 points of its combined ratio, essentially all Property) and Arch released ~$600M. Reserve releases are a discretionary choice that flatters peers’ ratios; Everest’s charge is the same choice in the opposite direction. This cuts both ways and honesty requires saying so: part of Everest’s optical inferiority is peers harvesting redundancy. But the direction of Everest’s own choice, three years running, is what it is. Second, SCOR is the most instructive comparable in the table. It printed a better FY2025 combined ratio than Hannover Re and a 19.2% ROE, yet trades at 1.34× against Hannover’s 2.38× — because its book value and economic value have not compounded since 2022, following its own 2023–24 reserve problem. The market prices the track record, not the quarter. That is the empirical answer to how long a reserve-credibility discount persists: for SCOR, at least two years of clean results and counting.

Verdict: no durable competitive advantage — a top-five global reinsurance balance sheet, competently operated, in an industry whose capital cycle has no supply lag and therefore cannot confer durable excess returns. Everest is the weakest franchise in the cohort on ROE, book-value compounding and reserve quality simultaneously. It is correctly, not punitively, priced at approximately book value.

5. Growth History and Forward Opportunities

The growth history is a single arc: aggressive expansion into a line that did not pay, followed by retreat. Gross written premium ran $9.1bn (2019) → $10.5bn (2020) → $13.1bn (2021) → $14.0bn (2022) → $16.6bn (2023) → $18.2bn (2024) → $17.7bn (2025), with Q1-2026 down 18.5% year over year (−6.4% excluding the Legacy segment). That is a doubling in five years and then a sharp reversal.

The growth was real but low-quality, and the filings let us prove it. Everest built its insurance platform organically from 2015 “through the investment in key leadership hires,” launching Lloyd’s Syndicate 2786 de novo and forming Everest Insurance (Ireland). There were no acquisitions; the growth was genuinely organic. But organic growth into a softening-loss-cost line is not a virtue. The Insurance segment’s combined ratios were 100.5% (2023), 130.7% (2024) and 114.6% (2025), producing underwriting results of −$18M, −$1,097M and −$541M. Not one profitable year, and roughly $1.66bn of value destroyed in 2024–25 alone. Everest grew GWP 19.2% in 2023 against its own stated 10–15% target band — and premium growth was, as documents, a named pay-linkage measure.

Marathon’s asset-growth anomaly is directly on point: rapid balance-sheet growth in a financial business is among the most reliable predictors of subsequent poor returns, because growth in an intangible-free, price-taking business is achieved by accepting terms others declined. Everest doubled its balance sheet from $32.7bn (FY2020) to $62.5bn (FY2025). The reserve charges are the bill.

The forward opportunity set is deliberately smaller. The strategy is retrenchment, and it is being executed with unusual decisiveness:

• Non-renewed 33% of the US casualty portfolio’s GWP in 2024, resetting the forward loss-trend assumption for GL, excess/umbrella and commercial auto to ~12%.
• Sold the sports and leisure business (Q4-2024) — the source of $403M of FY2024 adverse development.
• Sold ~$2bn of commercial retail renewal rights to AIG (October–December 2025) for $301M plus a $30M origination fee and $90M of transition-services fees, at approximately 15% of one year’s premium.
• Agreed to sell Everest Insurance Company of Canada to Wawanesa for C$410M (March 2026), closing 2H-2026.
• Re-segmented into Reinsurance Treaty / Global Wholesale & Specialty / Legacy effective 1 January 2026.

Q1-2026 shows the surgery working, and the honest bull case rests here. Legacy GWP fell 80.3%; Reinsurance Treaty ran an 87.2% combined ratio on $2,405M of net written premium; Global Wholesale & Specialty grew net written premium 5.6% to $692M at a 96.8% combined ratio; the Group combined ratio was 91.2% with a 16.7% annualized operating ROE. A cleaner, smaller, more reinsurance-weighted Everest is genuinely emerging.

But three qualifications must be applied to that improvement. First, 12.2 of the 13.1 points of Q1-2026 loss-ratio improvement came from catastrophes and catastrophe-related prior-year releases — cats were $130M against $472M in Q1-2025. The underlying attritional improvement was 1.9 points, not 13. Second, book value per share rose 1.0% in the quarter while total equity fell 1.1% — the per-share gain was a pure share-count effect from the $331M buyback, not value creation. Third, every disposal has been structured the same way: sell the going concern, retain the tail. Mt. McKinley (2015), sports and leisure (2024), the AIG renewal rights (2025) and Everest Canada (2026, via an intra-group loss portfolio transfer) all leave the reserves with Everest. The franchise leaves; the liabilities stay.

Where can growth come from? Realistically: property-cat and specialty treaty reinsurance, into a market repricing down 16%; and the Global Wholesale & Specialty platform, which is small ($692M of quarterly NWP) and growing 5.6%. There is no adjacency, no fee business of scale (Mt. Logan earns $7M), no owned distribution to leverage, and no acquisition currency at 1.0× book. Everest’s realistic forward growth rate is negative in the near term and low-single-digit thereafter.

Verdict: low-quality historical growth, correctly being reversed — but with no visible engine to replace it. The retrenchment is the right decision, executed decisively, and deserves credit. It is nonetheless a smaller company earning a similar return in a softening market. Growth is not part of this investment case; the case, if there is one, is entirely about whether the balance sheet is worth stated book.

6. Financial Quality

A preliminary warning on data. Third-party aggregators are unusable for this name and the report relies on filings throughout. ROIC.ai returns gross_margin: 100 with no underwriting lines and models pretax income as the negative of “non-operating income”; its ROE series reconciles to neither ending nor average equity; its book_val_per_sh reports FY2022 BVPS of $310.39 against an actual $215.88. Its EV/EBITDA is meaningless for an insurer and is reported nowhere in this article. Every figure below is recomputed from the 10-K or 10-Q.

6.1 Underwriting

Segment combined ratios (FY2023 / FY2024 / FY2025): Reinsurance 86.4% / 89.7% / 91.7% — deteriorating steadily; Insurance 100.5% / 130.7% / 114.6%.

Everest ran an underwriting loss in three of the last six years, at the top of the best reinsurance market in a generation. That is the plainest available statement of business quality and it sits badly beside peers printing 73–91% in the same window.

The loss-ratio decomposition contains the report’s most falsifiable bear point. The consolidated current-accident-year attritional loss ratio has been essentially flat at ~60% across FY2023–25 (59.9% / 59.8% / 60.4%). All of the damage arrived through prior-year development, not through current-year picks. The Insurance segment did respond, raising its current-year pick 5.1 points from 63.3% to 68.4%. But the Reinsurance segment’s current-year pick is essentially unchanged at 57.3% versus 57.6% in 2023 — despite US casualty being the acknowledged source of the problem, and despite $22.7bn of reserves standing behind that segment. If the loss-cost environment deteriorated enough to require ~$2.0bn of strengthening on prior years, it is hard to argue the same environment leaves Reinsurance’s current picks untouched. Either the assumed book is genuinely different in mix from the primary book that failed, or its current picks remain optimistic and a future strengthening is being seeded now.

A comparability warning. Everest re-segmented three times in about 26 months: A&H realigned from Reinsurance to Insurance in Q4-2023; the “Other” run-off bucket created in Q4-2024 (restated 2023–25 only); and the three-segment structure effective 1 January 2026, with a formal FY2025 10-K recast filed 3 June 2026. FY2021–22 segment ratios are on the old basis. Each recast coincided with a period of segment underperformance and each moved loss-making business into a run-off bucket. Any multi-year segment trend must carry this footnote, and readers should understand that part of the improvement in “continuing” ratios is reclassification rather than performance.

6.2 Book value — and what actually drove it

Book value per share: $243.25 (2020) → $258.21 → $215.54 → $304.29 → $322.97 → $379.83 (2025) → $383.75 (31 March 2026). Book value plus dividends compounded at 11.29% over five years. Tangible book value equals book value — there is zero goodwill, so the multiple means exactly what it says, and there is no impairment risk concealed in equity. On a $62.5bn balance sheet that is a genuine and unusual quality marker, and it materially strengthens the reliability of price-to-book as the valuation anchor here.

But 47% of FY2025’s book-value growth was bond marks, not earnings. AOCI swung from −$1,138M to −$52M, contributing +$1,086M, or +$26.68 per share of the +$56.86 gain. Ex-AOCI, FY2025 book growth was approximately 9.3%, not the headline 17.6%. The FY2025 equity bridge ties exactly: $13,876M opening + $1,591M net income − ~$335M dividends − ~$798M buybacks + $1,086M AOCI + $40M APIC = $15,460M.

This tailwind is now exhausted, and so is the second one. At −$52M, AOCI has essentially completed its pull to par and cannot repeat. Simultaneously, the reinvestment tailwind has peaked: net investment income growth decelerated to +8.7%, only $91M of the $170M increase came from fixed maturities (+6.1% on a $34.6bn book), and book yield actually fell from 4.9% to 4.8%. Both of the forces that flattered FY2024–25 book-value growth are spent. From here, book value must be grown by underwriting and investment income alone — which is precisely the ~10% ROE the record suggests.

6.3 ROE quality and the cost of equity

Recomputed on average equity: 5.45% (FY20) · 13.88% · 6.43% · 17.92% · 10.14% · 10.85% (FY25). Six-year average 10.78%; five-year average 11.84%; the eleven-year average is 9.1%.

FY2023 must be normalized. The Bermuda Corporate Income Tax Act of 2023 produced $578M of net deferred tax benefits — which is why FY2023 shows a $363M tax benefit on $2,154M of pretax income, and why the auditor elevated it to a Critical Audit Matter. Normalized FY2023 EPS is roughly $46.37, not $60.19, and ROE 17.9% rather than 23.3%. FY2023 should not anchor any normalized-earnings estimate: it carries both the tax windfall and the peak of the reserve illusion.

Cost of equity: ~10%, range 9–11%. The 0.42 beta should emphatically not be run through CAPM, which yields ~6.4% and is not credible. The low beta is a real diversification artifact — catastrophe frequency and casualty reserve adequacy are genuinely uncorrelated with equity-market factors — but beta measures covariance, not the probability of permanent capital impairment, and Everest’s risks are fat-tailed and one-sided. Idiosyncratic volatility is 16.9% against ~24% total, so roughly 42% of variance is stock-specific, and both worst days of the five-year window (−11.4% and −8.5%) were single-name events on quiet market days. The lifetime maximum drawdown is −46.7%. A 0.42 beta here means diversifying, not safe, and any argument resting on the low beta as evidence of low risk should be rejected. A build-up of 4.3% risk-free + 5.0% equity risk premium + 1.0–1.5 points for reserve and tail risk gives ~10%.

The punchline: Everest earns approximately its cost of equity and no more. A franchise earning exactly its cost of capital is worth approximately book value. This is a complete and sufficient explanation for the 1.008× multiple without any appeal to reserve fear at all — and it is the single most important analytical point in this report.

6.4 Reserves — the thesis

Net loss and LAE reserves stand at $30,597M against ~$15.5bn of equity — reserve leverage of 1.98×. A 10% reserve error is therefore roughly 20% of book value. Gross reserves as a multiple of GWP rose from 1.48× to 1.94×. IBNR climbed from 65.3% of gross reserves (2023) to 69.0% (2024) to 70.3% (2025), and 73–74% within Insurance — meaning an increasing share of the balance sheet is estimate rather than reported claim.

Net prior-year development ($M, positive = unfavorable):

The three “clean” years were a netting artifact. FY2023’s headline $5M favorable decomposes into $397M of favorable mortgage and short-tail development offsetting $392M of unfavorable 2016–2019 casualty. For three consecutive years, short-tail redundancies were harvested to offset long-tail deterioration that was already emerging. An investor reading only the net line saw three benign years; the components were already breaking. The same mechanism appears inside the confession quarter itself: the FY2024 charge included $684M of adverse Reinsurance US casualty exactly offset by $684M of favorable property/mortgage, producing a net zero Reinsurance print.

The accident-year migration is the decisive finding. The FY2025 10-K’s Insurance-Casualty triangle discloses prior-year development by accident year:

The deficient cohort has moved forward in each of three successive reviews: FY2023 blamed AY2016–19; FY2024 hit AY2019–23; FY2025 hit AY2022–24 (+458 of the +474). Meanwhile the old problem years turned slightly favorable, and accident year 2024 developed adversely on barely one year of maturity. This is not a quarantined legacy cohort being cleaned up — it is the same excess-casualty book being repeatedly under-reserved across a rolling window. Each “final” charge addressed the years that had already emerged while the then-current years were still being written short. Accident year 2025 is booked at $1,698M ultimate and is 85.9% IBNR — the next cohort to watch, and the cleanest forward test available.

A disclosure-hygiene note, stated fairly. Four different figures circulate for the 2024 charge: $2,187M gross casualty strengthening, $1,704M management “reserve action” (the 8-K of 27 January 2025, Exhibits 99.1/99.2), $1,500M pre-announced prior-year development, and $1,337M audited net development per Note 4. FY2025 repeats the pattern: Note 4 leads with +$389M, excluding $146M of Other and the $122M ADC loss, against a true total of $657M. All are disclosed plainly in footnotes and none is improper — they measure different things, and the bridge between $1,704M and $1,337M is the ~$684M of offsetting property/mortgage releases and the current-year component. But the consistent practice of leading with the smallest defensible number is worth naming.

Reserve sensitivity. The most defensible anchor is management’s own disclosed range: gross reserves of $31,743M to $36,880M, i.e. ±7.5%. The high end equates to $53.20 per share, or 14.0% of book value. Analyst-assumed sensitivities on the ~$19.9bn assumed casualty base give: 5% adverse = $20.61/share (5.4% of BVPS); 10% = $41.21/share (10.9%); 10% net of remaining ADC = $24.64/share (6.5%). These assume roughly 50% of the Reinsurance reserve is casualty; if the casualty share is higher, they understate the risk.

At 1.008× book, the market is paying par for a balance-sheet figure carrying a company-disclosed ±14%-of-book error bar, with a demonstrated three-year bias toward the adverse side of that range.

6.5 The adverse development cover — narrower than the headline

Effective 1 October 2025 (agreements signed 26 October), Everest Reinsurance Company and Everest Reinsurance (Bermuda) ceded to State National Insurance Company and MS Transverse Insurance Company, both retroceded to Longtail Re, an affiliate of Stone Ridge Capital; Mitsui Sumitomo provides a parental guarantee for MS Transverse. Gallagher Re structured it.

Subject business: North American Insurance and Other segments, accident year 2024 and prior, excluding asbestos and environmental. Subject statutory reserves were $5,369,448,704 at 30 September 2025.

Four facts make this materially narrower than “a $1.2bn cover.”

1. The in-the-money first layer was consumed almost immediately. At 31 December 2025, total covered losses ceded to State National were $1,253M against a $1,250M first-layer limit — 100.2%, within three months of inception. Remaining unexpired limit was $597M + $400M at year-end and $598M + $400M at 31 March 2026. Everest bought protection above the in-the-money tranche and burned the bottom $1.25bn of it on day one. That first layer was never protection; it was a funded transfer of reserves that had already developed adversely. Roughly $1.0bn of genuinely live cover remains.
2. Everest paid a ~9–10% premium over carried reserves. Consideration of $1,372M against $1,250M of ceded reserves produced an immediate $122M pre-tax loss ($105M Insurance, $17M Other), which added 11.1 points to the Q4-2025 Insurance combined ratio. The 10-K describes the write-off as “excess compensation for the uncertainty of future claims development.” That premium is the market’s explicit price for Everest’s own reserve uncertainty.
3. The upside was surrendered too. Under the profit commission Everest gives back 50% of any favorable development below 100% of carried reserves, capped at $625M, plus 15% of the MS Transverse premium on a loss-free commutation within 60 months. Investors hoping future redundancy on the North American book will drive a re-rating should understand half the first $625M of it now belongs to Longtail Re. The cover truncates both tails.
4. The largest exposure is entirely uncovered. The ADC does not touch the $22.7bn Reinsurance reserve (which took $684M adverse in 2024 and $456M in 2025), nor accident year 2025 onward, nor A&E.

And it substitutes counterparty credit for reserve risk. State National holds only a $250M funds-withheld trust against a $1,253M recoverable, and unlike MS Transverse its obligations carry no parental guarantee. That leaves roughly $1.0bn of largely uncollateralized exposure to a fronting structure whose ultimate risk-taker is an alternative-capital vehicle. Total reinsurance recoverables rose to $5.1bn from $3.1bn. This substitution is not stress-tested in the recoverable disclosure and belongs in the risk matrix as a distinct exposure.

Two further facts complete the picture, and both are absences.

First — and it is the single most important negative in this report — there is no independent third-party actuarial reserve review. Across all six earnings calls from Q4-2024 to Q1-2026, no external actuarial firm is ever named: not Milliman, not Oliver Wyman, not EY, Deloitte or WTW. The phrases “independent review,” “third-party actuarial,” “external actuary” and “external review” do not appear at all. What exists instead is (i) internal reserve studies, (ii) the Longtail Re cover used rhetorically as validation, and (iii) “outside strategic advisers” — engaged for the retail exit, explicitly not for reserves. For a company that has mis-reserved by roughly $2.2bn gross across two charges and asks the market to accept that the third estimate is right, the absence of an independent actuarial opinion is a conspicuous gap. It is also the single cheapest, most credible action available to management, and it has not been taken.

Second, the reserve-review cadence was never changed. Asked directly, Williamson: “Our reserve deep dives are still conducted on an annual basis, with obviously our quarterly process still in place.” Kociancic: “There’s no difference in the cadence of the reserve reviews. I would just say there’s a heightened awareness and alertness on the U.S. casualty lines.” Only the loss-trend review frequency increased, and the global studies were later pulled forward into Q3-2025. A process that produced three successive misses was retained substantially unaltered — the response was more conservatism inside the same process, not a different process. The one structural change came later and from outside: a new Group Chief Actuary and Group CRO hired externally in December 2025 — though note the new CRO reports to the Group CFO, not to the CEO or directly to the Board Risk Committee, which is a weaker control line than best practice at a company in this position.

Management’s own account of the sizing is the most revealing disclosure in the file. Asked on the Q3-2025 call how the $478M charge was determined, Williamson said it “was all about getting ultimately to an ADC that would create finality,” and on sizing the cover: “it’s less about… a certain percentage of the actuarial best estimate or a certain standard deviation. It’s more about putting this out in the tail so that people don’t have to worry about it anymore.” By management’s own account the charge was sized to the reinsurance structure they wished to buy, not derived from an actuarial best estimate. That inverts the normal causal order — reserves should determine the cover, not the cover the reserves. It is candid, and it is also an admission that the internal estimate was not trusted by its own authors. There is a legitimate opposing reading, and it should be stated: sophisticated third parties, advised by Gallagher Re, were willing to attach at $5.4bn, which is real external validation of the attachment point for AY2024-and-prior North American reserves. But it validates a level, not a process, and it says nothing about the segment holding three-quarters of the reserves.

6.6 Investments, capital and cash flow

Investments: $45.4bn, 98.1% investment grade, duration 3.4 years against 4.0-year liabilities. Conservative and well matched. One quality-of-earnings flag: $5.5bn of limited-partnership and COLI alternatives (12% of invested assets), carried at GP-supplied marks on a one-month-to-one-quarter lag. FY2025 recognized $401M of income against only $195M of cash distributions — roughly $169M of non-cash NAV appreciation, about 8% of net investment income. The private-credit sub-allocation is undisclosed. Modest, but real, and worth an IR question given the cohort-wide drift into private credit.

Capital: conservative, with two nuances worth stating. Debt is 18.8% of total capital and the term debt is excellent: $2.4bn of senior notes issued at a blended ~3.3% coupon, with $2.0bn of it 30-year money raised in October 2020 and October 2021 at the bottom of the rate cycle. That debt now carries at $2,352M against a fair value of $1,689M — roughly $663M of unrecognized economic gain that GAAP book value does not reflect, a genuine and rarely-noted offset to the reserve risk on the other side of the balance sheet.

The two nuances. First, the apparent deleveraging from 26.8% (FY2022) to 18.8% (FY2025) is an AOCI story, not debt reduction — absolute debt has been flat at ~$3.59bn since FY2024 while AOCI recovery added ~$1.1bn to equity. Second, roughly $1,019M of Federal Home Loan Bank borrowing sits inside that total at an effective ~4.7% — more expensive than all $2.4bn of senior notes combined — with $719M rolling within twelve months. Excluding FHLB, leverage is 14.3%. Neither point threatens solvency; both mean the leverage trend flatters. No leverage target is disclosed in any filing.

Cash flow — and an honest correction to a plausible bear argument. Operating cash flow fell 38% to $3,068M, which invites a reserve-stress reading. That reading is wrong. The ~$1,372M ADC consideration flows through operating activities; underlying operating cash flow was approximately $4.4bn, down ~11%. Decisively, prior-year paid losses as a percentage of opening reserves are stable at 23.7% / 23.5% / 24.2%. There is no payout acceleration. The reserve problem is an estimation problem, not a liquidity problem, and the article should say so plainly.

Verdict: ADEQUATE, NOT HIGH. Economics do not improve with scale here, and 2021–24 is the proof: gross written premium grew 74% and the reward was a 130.7% Insurance combined ratio and ~$2.0bn of adverse development. Book-value growth flatters the record — 47% of FY2025’s gain was bond marks and roughly a third of FY2023’s ROE was a tax entry. The balance sheet is strong enough to absorb the disclosed reserve range and carries zero goodwill; the earnings stream is not good enough to justify a premium to book.

7. Capital Allocation

The scorecard on buybacks is good, with one clear and costly exception.

FY2020 (the COVID crash, 0.88×), FY2025 (0.95×) and Q1-2026 (0.86×) were well executed. FY2024 was not: $200M spent at 1.19× book — 1.09× even on ex-AOCI book — executed February through September 2024, in the twelve months immediately before the reserve charge. Buying stock at 1.19× a book value about to be revealed as ~$1.3bn overstated is the single worst capital-allocation decision in the window.

The 2023 equity raise is a near-complete round trip. On 19 May 2023 Everest sold 4,140,000 shares at $360.00 for net proceeds of $1,445M ($349.03/share) — 1.52× the 31 March 2023 BVPS of $229.92. Issuing well above book is good execution and was accretive: +$11.4/share, +5.0% to BVPS. The stated use was “long-term reinsurance opportunities and continuing build out of the global insurance business.” Everest then repurchased 3,933,748 shares across FY2024–Q1-2026 for $1,328M ($337.60/share) — 95.0% of the shares issued, bought back 3.3% below the net issue price, within 34 months.

The arithmetic of the round trip is roughly neutral to slightly positive. The strategic verdict is not. The capital was raised to build out the global insurance business, and that business required a $1.7bn reserve correction twenty months later, has since had its renewal rights sold to AIG, its Canadian operation sold to Wawanesa, its back book capped by a $1.2bn ADC, and its remnant re-segmented into a bucket labelled “Legacy.” The raise was individually accretive; the strategy it funded destroyed far more than the 1.52× issuance created. Capital allocation must be judged on where money went, not only on the price at which it was raised.

The compressive number. Cumulative unfavorable prior-year development FY2020–FY2025 of $2,379M equals 29.8% of cumulative reported net income ($7,971M) and 26.1% of opening equity. Roughly thirty cents of every dollar of reported profit over six years was subsequently taken back. That is the capital-allocation verdict in a single statistic, and it is why a “cheap on trailing earnings” argument cannot be trusted here: a material share of those earnings did not exist.

M&A: essentially none, and that is genuinely good. There is zero goodwill and zero intangible assets across six years of filings — no ASC 805 business combination, no purchase-price allocation, no impairment risk. Everest built its platform through key leadership hires, launching Lloyd’s Syndicate 2786 de novo in 2015 and internalizing its managing agency in August 2025. But every corporate transaction in the ten-year window runs in one direction: out. Mt. McKinley sold to Clearwater/Fairfax with a 100% retrocession (2015, cap $440.3M, raised to $450M in 2019); sports and leisure sold (Q4-2024); AIG renewal rights (2025); Everest Canada (2026). Everest is an organic underwriter in reverse gear.

The AIG transaction, properly bridged. The headline $127M “gain from sale of renewal rights” is a net figure that the filing never bridges. Gross gains were $204M (ROW) + $55M (EU) = $259M; charges were $83M of capitalized-software impairment + $28M severance + $21M legal/M&A fees = $132M; $259M − $132M = exactly $127M. The separately-received $30M origination fee sits outside it, and $47M was deferred. Cash proceeds were $331M. There is also a clawback: if premium renewed with AIG through 31 December 2027 falls below 80% of FY2025 aggregate premiums, Everest reimburses up to $70M. A further $81M of net transaction expenses hit Q1-2026.

The sports and leisure disposal deserves a specific flag. Everest recognized a $40M gain, but the buyer is never named, no purchase price is disclosed, no 8-K was filed, and no proceeds line appears in the FY2024 cash-flow statement — verified across the FY2024 and FY2025 10-Ks, the Q2/Q3-2024 10-Qs, the Q4-2024 earnings 8-K and the full 8-K index. The $40M gain is cosmetic against $403M of FY2024 adverse development on that same book (accident years 2019–2023). Recording the gain in other income while the loss sits in incurred losses splits one economic event across two lines and flatters the presentation.

Dividend. Frozen at $2.00 per quarter since Q2-2024 — nine consecutive quarters — ending a roughly 6%/year growth streak. It is very safe: 21% payout, 9.2× covered by operating cash flow. Management has decisively chosen buyback (FY2025 buyback was 2.4× dividends; Q1-2026 4.1×), which is defensible at a sub-book valuation. The timing of the freeze is the tell: it began in Q2-2024, before the charge was public, coinciding precisely with the period in which casualty was breaking internally.

Compensation — genuinely better designed than expected, then overridden. Fairness requires giving credit where due. Management is paid on ROE, not premium growth: 60% of the bonus on a single metric, Adjusted Net Operating Income ROE, with 40% discretionary. Critically, reserve development is not excluded from the pay metric — the carve-outs are closed (investments, FX, a catastrophe collar). That is materially more honest than the attritional combined ratio Everest markets to investors, which does strip out 8.8 points of 2024 development.

But there is no reserve metric anywhere in any Everest incentive plan. An exhaustive search of the FY2025 and FY2026 proxies for reserve development, reserve adequacy or prior-year development in any incentive context returns nothing. The sole permitted adjustment to the pay metric is for catastrophes — and in 2025 actual catastrophe losses fell inside the range, so no adjustment was even triggered. Reserves appear in the compensation philosophy only as a virtue to be shielded: management should act long-term “even when such actions may temporarily reduce short-term profitability,” with “reserving methodologies and reserve positions” given as the example. There is no symmetric mechanism penalizing reserve deficiency. Everest’s most-promoted headline metric, the attritional combined ratio, is defined to exclude prior-year development by construction.

Four defects then hollow out the good design:

1. The formula took the hit honestly; discretion gave it back. FY2024 Adjusted Operating ROE printed 8.7% against a 17.0% target, cutting the financial leg to 24.2% of target — whereupon the Committee awarded the individual leg at 100% of maximum. Had it paid at target instead, Williamson’s bonus would have been $687,300 rather than $903,300: discretion added $216,000, +24%, in the year of the $1.7bn charge, and did so at the statutory cap. The individual bucket has exceeded target in both charge years.
2. The FY2025 bar was quietly reset downward. The ROE target was cut 17.0% → 15.0% and the maximum from ≥25.0% → ≥18.0%. The 12.4% FY2025 print paid 67.5% of the financial component on the reset curve; on the prior curve it would have paid roughly 44%.
3. The 2023 PSU award — whose performance period spanned both reserve charges — paid out at 102.6% of target. Two mechanisms neutralized the charges: the strong 2023 ROE tranche (18.7% → 157.3%) was banked before the charges and is unrecoverable, because the ROE sleeve is a one-year look-back inside a three-year award; and the relative-TSR sleeve paid 142% precisely because peers were also weak, insulating the award from Everest-specific reserve failure. No PSU was ever reduced, adjusted or forfeited on account of reserve development.
4. The clawback cannot reach this. It has exactly two triggers — an accounting restatement and material willful misconduct. Neither occurred; no non-reliance determination was ever made. Reserve deficiency absent a restatement is simply not a clawback trigger. Separately, premium growth is a named Item 402(v) pay-linkage measure (“Gross Written Premium Annual Growth Rate”) with no threshold, cap or penalty; Everest grew GWP 19.2% in 2023 against its own 10–15% band.

There was no clawback, no reduction and no negative discretion. The Committee framed the charge three times as “decisive action… to position the Company for sustainable profitability,” and awarded the CFO his maximum individual bonus citing “managing a comprehensive loss reserving strategy” — in the year that strategy required a $1.7bn correction. A further $5.0M of purely time-vested retention stock, carrying no performance or reserve-adequacy condition, was granted in February 2025 — the same Committee meeting that set the depressed formulaic bonus, eight weeks after the charge. For Williamson the retention grant alone was 1.9× his entire annual restricted-stock award. Meanwhile his package was reset upward: base $900,000 → $1,250,000 (+38.9%), target bonus 140% → 200%, target equity 217% → 420%; FY2025 total compensation $8,952,500. Departing CFO Kociancic’s FY2025 total rose 52% to $6,445,543 in the year he was told he was leaving, with his bonus fixed at target ($1,645,000) overriding a formula that would have paid ~$666,000, inside an aggregate transition package of roughly $9.35M. Incoming CFO Habayeb received ~$10.7M of inducement against ~$5.0M of annual target pay, of which the $4.9M retention RSU is not attributable to any disclosed forfeiture.

Shareholders have noticed, mildly: say-on-pay support fell from 94.06% (FY2024) to 91.95% (FY2025), and Compensation Committee member John Amore drew the weakest support of any nominee in both 2025 (83.1%) and 2026 (89.6%). No proxy adviser is named in any proxy, there is no shareholder-engagement-feedback section, and the compensation consultant was switched from Mercer to Meridian for 2025 with no reason given.

The finding is not that the metrics were wrong; it is that the Committee declined to let them bite.

Insider behaviour is the strongest bull evidence in this report — and it required correcting an initial misreading. A first pass of the Form 4 corpus concluded that no insider bought the reserve dip and that only directors bought. A rigorous re-parse of all 176 filings (231 transactions; the files carry .xml extensions but are SEC-rendered HTML, which defeats naive tag-based parsing) overturns that. The itemized record reconciles exactly to ten open-market purchases in five years totalling 22,255 shares, and the conclusion is materially more positive than first thought.

2025 was the largest insider-buying year in the five-year corpus — $4.79M — and the buying came after the charges, not before. The CEO himself bought 1,000 shares at $337.97 in June 2025, four months after the $1.7bn charge, and has never sold; his holding rose from 11,749 to 29,636 shares. Galtney’s October purchase was 3.5× his November-2024 buy at a materially lower price, and new director Levine bought roughly 3× his entire granted position four months into his tenure — both one day after the $478M Q3-2025 charge, a deliberate open-window purchase into bad news. Against this, sales across 2025–26 total just three transactions and $1.10M, each seller disposing of only 6–11% of position while retaining 8,000–12,000+ shares. Net insider flow is +$3.68M. Notably, retiring CFO Kociancic made zero purchases and zero sales despite a known exit, and former Chairman Taranto’s 293,524 shares (~$100M) are untouched.

The honest verdict: the insider tape does not support a “nobody would touch it” reading, and materially strengthens the bull case. Two caveats keep it from being decisive. Everest grants no options, so there is no exercise-and-sell channel and the corpus contains zero Table II rows — which makes the buying cleaner but also means the base rate of insider activity is low. And former CEO Andrade filed no Form 4 after 18 September 2024 while holding 57,529 shares; his reporting obligation lapsed on departure, so post-exit disposals are unobservable — a genuine blind spot, not evidence he held.

Static alignment nonetheless remains thin. Directors and officers hold 0.7% of shares, down from 1.1%, against $68.5M of grants over the period. The stock-ownership guideline counts unvested stock toward compliance — which is not an ownership guideline, it is a restatement of the equity grant. And no filing in the corpus references a 10b5-1 plan, so all sales are presumptively discretionary.

Verdict: MIXED, TRENDING BETTER — but the record does not support a premium. Individual transactions have been executed competently: issuing at 1.52× book, repurchasing at 0.86–0.95×, selling renewal rights at 15% of premium. The strategic allocation was poor: capital raised in 2023 to expand a book requiring a $1.7bn correction, $200M repurchased at 1.19× an overstated book, and ~30% of six years’ reported earnings subsequently reversed. Governance compounds it — a well-designed pay formula was overridden by discretion in precisely the year it should have bitten.

8. Changes and Headwinds — Last Two Years

The management turnover is near-total, and the sequence is the finding.

The head of the division that generated $1,278M of the charge resigned eight weeks before it was announced — and his March-2024 agreement provided that resignation before 1 January 2025 would be treated as “Good Reason,” a protection that lapsed on that date; he was paid his target bonus of $1,120,000 on that basis. The CEO resigned four weeks later. The charge was disclosed five days after his successor was made permanent. This chronology is circumstantial, not proof of anything, and is presented as sequence rather than inference — but a reader is entitled to it.

Two clarifications on Andrade. The 8-K discloses no severance, separation agreement or compensatory arrangement whatsoever and does not name a destination — but the destination is externally verifiable: USAA announced him as President and CEO on 8 January 2025, effective 2 April 2025. He forfeited his unvested equity, received a $0 FY2024 bonus, and his FY2025 Summary Compensation Table total was $54,089. He is never mentioned by name on any of the six earnings calls from Q4-2024 through Q1-2026 — and, remarkably, not one analyst asked about the departure.

A point of accountability that materially qualifies the “new management inherited the problem” reading. Williamson joined Everest in October 2020 as COO, added Head of Reinsurance in May 2021, and in March 2024 his remit was expanded to lead both global reinsurance and insurance. By his own account on the Q3-2025 call, he ran the remediation from July 2024. He is therefore not an outsider cleaning up someone else’s book: he had operational ownership of US casualty for roughly two to three quarters before the $1.7bn charge, and for the entire period preceding the $478M second charge. That cuts both ways — it makes his personal share purchase more informative, and it makes the second charge harder to attribute to a predecessor.

Five new directors joined between March and August 2025, the board expanded to eleven with a full committee reorganization, and a new General Counsel, Reinsurance CEO, Global Wholesale & Specialty CEO and CFO were installed. Eight of twelve Form 3s post-date the charge. Of the five named executive officers in the 2023 proxy, four are gone and the fifth departs in July 2026 — only Williamson remains, and he changed seats. This is a governance overhaul, not routine refreshment.

The most important layer of turnover is invisible in the SEC filings entirely, and an investor reading only EDGAR would miss it. Everest’s 10-K contains no “Information about our Executive Officers” section; Part III is incorporated by reference. The strings “Chief Actuary” and “Chief Claims” appear zero times across the FY2022–FY2025 10-Ks, and no individual holding the Chief Underwriting Officer, Group Chief Actuary, Group Chief Risk Officer or Chief Claims Officer role is named in any Everest proxy or 8-K, ever. Everest does not designate these as Section 16 officers, so the entire turnover of its actuarial and risk-control function through the reserve-charge period was never disclosed to shareholders. From trade press:

• Mike Mulray, President of North American Insurance, exited August 2024 — in a reported “management shake-up,” four months before the charge. Not SEC-disclosed.
• Ari Moskowitz, Group Chief Risk Officer since 2022 and a twelve-year Everest veteran, left for Selective Insurance around June 2025 — four months after the $1.7bn charge. The role then sat vacant for roughly six months. Not SEC-disclosed.
• Bill Hazelton, Mulray’s successor, exited within days of the AIG renewal-rights sale in October 2025. Not SEC-disclosed.
• On 3 December 2025, six weeks after the second charge, Everest announced Katy Bradica (ex-AXA XL Chief Pricing Actuary) as Group Chief Actuary — remit explicitly covering “pricing, reserving and analytics,” reporting directly to Williamson — and Attila Kerényi (ex-CRO of Swiss Re P&C Reinsurance) as Group CRO, both effective January 2026. Neither announcement names a departing incumbent.
• Melissa McDermott, hired as Insurance-division Global Chief Actuary in August 2023 specifically as part of the remediation Williamson later cited on the Q4-2024 call, left for CNA by June 2026.

Two observations follow. First, the roughly six-month CRO vacancy (June–December 2025) spans exactly the window in which the Q3-2025 charge was determined and the $1.2bn Longtail Re adverse development cover was designed — the two most consequential risk decisions of the period were taken without a permanent Group Chief Risk Officer. Second, the remediation hires themselves have now left: the chief actuary brought in to fix the insurance division departed after both charges landed on her book. The FY2025 10-K’s Enterprise Risk Committee roster corroborates the restructuring from inside the filings — it newly adds a Group Chief Underwriting Officer and a Chief Executive of Legacy Operations, control and run-off roles that did not previously exist.

This is not a CEO succession story. It is a three-wave clear-out running along the reserve fault line: underwriting leadership of the offending book (August–December 2024, both before the announcement), then CEO and CFO, then the entire risk-and-actuarial control layer — rebuilt from outside. Whether one reads this as a credible reconstruction of a failed function or as evidence of how deep the failure ran, it is material, and it is not in the filings.

One further comparability warning. The 8-K of 3 June 2026 recast the FY2025 10-K onto the new three-segment structure — and the recast reaches Footnote 4 (Reserve for Losses and LAE) and Schedule III, meaning the reserve-development disclosures themselves have been re-cut along new segment lines. This breaks comparability of the historical development triangles against everything published before June 2026. The filing carries two audit-firm consents (KPMG and PwC), reflecting the auditor transition the recast period spans. The “Legacy” segment is the reporting counterpart to the ADC and the AIG sale: the mechanism by which the problem casualty book is ring-fenced out of the ongoing business.

The disclosure gap is the sharpest single piece of evidence. The Q3-2024 earnings 8-K, filed 30 October 2024 — nine weeks before the charge — reported net income of $509M, a year-to-date net-income ROE of 17.8%, a Group combined ratio of 93.1% and an Insurance combined ratio of 97.1%, with no indication of casualty reserve stress. Everest had also published its annual Global Loss Triangles for FY2023 on 9 August 2024, five months before the charge. What those triangles showed in AY2020–2023 US casualty that the Q4-2024 review then repriced remains an open question.

Guidance was withdrawn alongside the charge. Everest adopted a “new target objective to deliver a mid-teens total shareholder return over the cycle” and stated it “will no longer be providing detailed forward guidance.” Withdrawing guidance while announcing a $1.7bn charge is a credibility reset, not a routine framework change — and it removes the most direct instrument by which management could rebuild credibility.

The de-risking promise and its falsification. Management asserted adequacy across three consecutive calls before being falsified:

• Q4-2024 (4 Feb 2025), Williamson: “Everest’s decisive reserve action added $1.7 billion to net reserves including over $200 million of additions to our 2024 loss picks,” promising “a level of prudence that we plan on continuing in 2025.” The 8-K quantified $481M of risk margin above actuarial central estimates.
• Q1-2025 (1 May 2025), CFO Kociancic, asked directly: “the bookings that we made are holding well… we’re quite — very comfortable with how we’ve progressed three months later after the charge.”
• Q2-2025 (31 Jul 2025), Williamson: “In Insurance, we remain consistent with our booked position.”
• Q3-2025 (28 Oct 2025), Kociancic: “We took further action to fortify our U.S. casualty reserves, strengthening reserves by $478 million on a net basis or 12.4 points on the combined ratio” — accident years 2022–2024, driven by “an acceleration of large loss activity, particularly in excess casualty and management liability.”

The entire $481M risk margin was consumed within three quarters. And the second charge struck accident years after those addressed by the first — which is materially worse than a single legacy clean-up, because it implies the pricing and reserving process was still producing deficient picks while the company was publicly asserting prudence.

Rating-agency pressure. A.M. Best’s negative outlook cites “elevated uncertainty surrounding the group’s business profile and enterprise risk management capabilities.” An agency questioning ERM capability — not merely reserve levels — attacks the only quasi-barrier a reinsurer possesses. S&P moved negative in February 2025. No rating has been downgraded and A+ is affirmed; this is pressure, not impairment.

Catastrophe experience. Hurricane Ida and European floods $635M (Q3-2021); Hurricane Ian $600M of a $730M Q3-2022 total; FY2024 cats $672M; the January-2025 California wildfires $442M — “the highest level of Q1 catastrophe losses in over a decade”; FY2025 cats $757M; and a benign Q1-2026 at $130M. Note that Everest pre-announced catastrophe losses via standalone 8-Ks in 2021 and 2022 but did not for the 2025 wildfires.

A tail item flagged but not booked: Williamson noted “a few tens of millions of dollars of incremental loss reserve” on the Baltimore Bridge, not yet reflected.

Verdict: these changes materially weaken the thesis, then partially repair it. The reserve recurrence, the risk-margin consumption, the rating outlook cuts and the near-total management turnover are all adverse. Against them, the retrenchment is genuine and decisive, the new leadership inherited rather than created the problem, the board has been substantially rebuilt, and Q1-2026 is a clean quarter. On net the last two years have destroyed credibility faster than they have rebuilt it — but the direction of travel in 2026 is right.

A note on news-flow coverage. The AZI news feed is unusable for this name — ten articles over a two-month lookback, one scored, and that single scored row mis-tagged to another issuer. No sentiment skew can be honestly reported, and this section rests on the 8-K corpus instead. The only clean consensus datum is a July-2026 cluster of seven price-target raises with ratings unchanged at mostly Neutral/Equal-Weight — the signature of a show-me name, where the sell side is marking to market rather than upgrading conviction.

9. Risk Analysis

The risks are not independent, and that is the real danger. A large catastrophe year would consume earnings and pressure the rating at the same time as casualty development and a softening property market. The correlation structure — not any single line — is what could turn a fair valuation into a poor outcome. Conversely, risks 14 and 15 are genuinely low: this is not a balance sheet under stress, and the cash-flow evidence directly refutes a liquidity reading.

10. Valuation Discussion

Start by discarding the headline multiple. At $382.57 the screen shows a trailing P/E of 7.78×, which looks arrestingly cheap. It is an artifact. The trailing-twelve-month window (Q2-2025 through Q1-2026, EPS $49.17 — which ties to the cent against two independent sources) excludes both disaster quarters of the last three years: Q4-2024 (−$13.96/share, the reserve charge) and Q1-2025 ($4.90, the California wildfires). It is the cleanest four consecutive quarters Everest has printed. The 7.78× multiple measures what Everest earns when nothing goes wrong, in a business whose entire economic purpose is to absorb things going wrong.

Honest multiples at $382.57:

On a two-year average EPS of $34.31 the stock trades at 11.2×, not 7.8× — against a cohort where earlier work on Arch put ACGL near 9× and RNR near 5×. Everest is not conspicuously cheap on earnings once the window is normalized. The genuine valuation question is entirely price-to-book.

And book value is unusually trustworthy here. Zero goodwill and zero intangibles mean price-to-book equals price-to-tangible-book exactly — no purchase accounting, no impairment risk. Among a cohort of serial acquirers this is a real advantage: the multiple means what it says. The offsetting caveat is that the liability side carries a company-disclosed ±7.5% range on gross reserves, equal to ±$53.20 per share, or 14.0% of book.

The own-history percentile inverts the naive reading. Everest’s P/B of 1.008× sits at the 59th percentile of its own multi-year range — the upper-middle, not the bottom. This stock has spent much of the last decade trading below stated book. “One times book” is not a distressed multiple for Everest; it is slightly rich versus its own history. (The P/E percentile of 31.7 is contaminated by the same TTM artifact and should be ignored; the P/B percentile is the trustworthy read.)

The cohort relationship is the central valuation evidence. Across nine reinsurers, price-to-book tracks delivered ROE closely:

(European names report under IFRS 17; combined ratios are not comparable to US GAAP, but ROE is the fair basis. The “price per point” column is a crude normalization, not a valuation model.)

Everest’s 1.008× is almost exactly what its ~10% ROE deserves on the cohort’s own pricing relationship. It is not obviously mispriced. This is the analytical heart of the report: a business earning ~10% against a ~10% cost of equity is worth approximately book value, and that is where it trades.

Embedded expectations — what must be true at $382.57. Using a simple justified-P/B framework, P/B = (ROE − g) / (COE − g). At a 10% cost of equity and 3% sustainable growth, the current 1.008× implies a normalized ROE of essentially ~10.1% — i.e. the market is underwriting exactly what Everest has delivered over eleven years (9.1%) and over the last two (10.1%, 10.9%), and nothing more. The market is not pricing a recovery to peer-level returns, and it is not pricing a reserve catastrophe. It is pricing continuation.

That framing is clarifying, because it tells you precisely what you are betting on:

Assumes COE 10%, g 3%, book $379.83. These are illustrative scenario mechanics, not forecasts, and no scenario is a price target.

The distribution is roughly symmetric, which is itself the finding. A genuine value opportunity offers asymmetry — limited downside, meaningful upside. Here the bear case (−36%) is about as large as the bull case (+42%), and the base case is zero. You are being asked to accept a symmetric distribution on a business with no competitive advantage, entering the down-slope of its cycle, with a reserve process that has been falsified three times.

Three further considerations bear on the valuation.

First, the two tailwinds that flattered recent book-value growth are exhausted. AOCI has pulled to −$52M from −$1,138M and cannot repeat (it contributed 47% of FY2025’s book growth, ~$26.68/share). The reinvestment tailwind has peaked — book yield actually fell from 4.9% to 4.8%. Forward book growth must be earned entirely by underwriting and investment income, which is precisely the ~10% ROE the record implies.

Second, SCOR is the empirical answer to “how fast does a reserve discount close?” SCOR printed a better FY2025 combined ratio than Hannover Re and a 19.2% ROE, yet trades at 1.34× against Hannover’s 2.38× — a ~44% multiple discount to a peer it out-underwrote — because its book value and economic value have not compounded since 2022 following its own reserve problem. The market prices the track record, not the quarter. A thesis assuming Everest re-rates promptly on one or two clean quarters is contradicted by the closest real-world analogue, where the discount has persisted for at least two years of clean results.

Third, the Q1-2026 print that drove the re-rating is weaker than it looks. Of the 13.1-point loss-ratio improvement, 12.2 points came from catastrophes and cat-related releases — the underlying attritional improvement was 1.9 points. Book value per share rose 1.0% while total equity fell 1.1%: the per-share gain was share-count arithmetic from the buyback, not value creation.

What the market is pricing correctly, and what it may not be. Correctly: the ~10% return profile, the absence of a moat, the reserve uncertainty at roughly its disclosed magnitude. Possibly incorrectly, in either direction: (a) the market may be under-weighting that the Reinsurance segment’s current-year pick is unchanged while its $22.7bn reserve sits outside the ADC — a genuine, uncapped tail; or (b) it may be under-weighting that a smaller, reinsurance-weighted Everest running an 87.2% Reinsurance Treaty combined ratio, buying back 8%+ of its shares annually below book, could compound book value at low-double-digits without any re-rating at all. Both are live. The evidence does not decisively favour either, which is why the honest conclusion is that the stock is fairly valued rather than cheap or expensive.

11. Variant Perception

Consensus. The sell side is agnostic and marking to market. A July-2026 cluster of seven price-target raises came with ratings unchanged at mostly Neutral/Equal-Weight — the signature of a show-me name. Consensus holds roughly that Everest is a cheap, reserve-damaged reinsurer whose new management is executing a credible retrenchment, and that the stock is worth owning only once development proves clean. An earlier assessment three weeks before this report put Everest at ~0.86× book with the note “reserve-troubled; discount is earned.” Since then the stock has re-rated ~17%.

The strongest bull case. Everest is a good reinsurance business that was attached for a decade to a value-destroying insurance business, and that insurance business is now being surgically removed. The evidence is real: Reinsurance Treaty ran an 87.2% combined ratio in Q1-2026; Legacy GWP fell 80.3%; the AIG sale removed ~$2bn of the worst premium at 15% of one year’s revenue; the ADC caps the North American back book; a new CEO, CFO, CRO, Group Chief Actuary and Group CUO have been installed. Meanwhile the company is buying back stock at 0.86× book with the quarterly floor raised to $300M — retiring roughly 8% of shares annually below book, which mechanically adds several points to book-value-per-share growth — and insiders have been net buyers of $3.68M, with the CEO buying personally at $337.97 and directors buying $4.45M the day after the second charge. The retrenchment is also larger than generally appreciated: Everest has shed more than $1.2bn of reinsurance casualty premium since January 2024, and on the primary side non-renewed 40–50% of casualty premium coming up for renewal in each of five consecutive quarters, at rate increases of 14–27% on what it kept. And management’s single best datapoint is genuinely good: on the Q3-2025 call Williamson stated that “80% of that development in U.S. casualty came from policies that were eliminated from our portfolio during the course of the remediation” — i.e. the deterioration is concentrated in business Everest has already exited, while it takes no credit for that in its go-forward picks. On this view, a cleaner, smaller Everest earning 12–13% on a shrinking share count at 1.0× book compounds attractively without needing any re-rating at all, and any re-rating toward the 1.3–1.4× of RNR and Arch is free optionality.

The strongest bear case. The reserve process, not the reserve level, is the problem — and it has not been shown to be fixed. The deficient cohort has migrated forward in three successive reviews and reached accident year 2024 on barely one year of maturity; AY2025 sits at $1,698M ultimate and 85.9% IBNR. The $481M of “risk margin above actuarial central estimates” announced in January 2025 was fully consumed within three quarters. Management sized the second charge, by its own account, to “getting ultimately to an ADC that would create finality” rather than to an actuarial estimate. The ADC’s first layer was 100.2% burned within three months, leaving ~$1.0bn of live cover, none of it over the $22.7bn Reinsurance reserve — whose current-year loss pick management has left unchanged at 57.3%. Cumulative development has already reversed 29.8% of six years’ reported earnings. And all of this is happening as the one working engine reprices down 16% into record capital, with A.M. Best questioning enterprise risk management capability.

The five assumptions that actually matter.

1. That the Reinsurance segment’s casualty reserves are adequate at an unchanged 57.3% current-year pick. This is the largest single uncertainty in the security and the least examined.
2. That normalized ROE is materially above 10%. The entire valuation case rests here; eleven years of 9.1% is the base rate against it.
3. That property-cat rate declines of 15%+ do not compress Reinsurance Treaty’s 87.2% combined ratio faster than mix improvement helps.
4. That the AOCI and reinvestment tailwinds’ exhaustion is understood. Both flattered FY2024–25 book growth; neither repeats.
5. That management’s reserving process — not merely its reserve level — has been rebuilt. The external hiring of a Group Chief Actuary and CRO in December 2025 is evidence for; the fact that the 2023 remediation chief actuary has since left is evidence against.

What would falsify each side. The bull case dies if AY2025 develops adversely, or any strengthening reaches the uncovered Reinsurance book, or Reinsurance Treaty’s combined ratio drifts above ~92% as rates soften. The bear case dies if Everest posts three-to-four consecutive quarters of neutral-to-favourable development and raises the Reinsurance current-year pick (which would signal genuine conservatism rather than an untested assumption), while sustaining a sub-90% Reinsurance Treaty combined ratio through the soft market.

The factor-positioning read supports a specific and useful conclusion. Everest’s empirical loadings are Insurance industry +0.949, Market +0.575, P&C Insurance Leaders +0.491, Value +0.479, BetaFactor −0.373, Momentum +0.106 — and Quality at +0.022, effectively zero. This is not the classic “momentum carrying a low-quality business” setup, which requires negative quality. It is a low-volatility value insurer whose recovery the model attributes almost entirely to Value and industry exposure, with no quality signature whatsoever. For a franchise that impaired its own reserve credibility twice in twelve months, the absence of a quality loading is itself the finding: the market is buying Everest as a cheap asset, not as a good business — which is exactly what the fundamental work concludes independently.

Two positioning caveats follow. First, Everest’s three largest style exposures — Value, negative Beta and Low Volatility — are all currently in favour (Value +14.3% over 252 days at z+1.7; low beta rewarded), a tailwind independent of fundamentals that could reverse. Second, the P&C Insurance Leaders basket Everest loads +0.49 on sits at a 21-day z-score of +1.9, a near-statistical extreme that mean-reverts on average. A meaningful share of the last month’s move is basket beta, not company-specific alpha — which is the most likely explanation for a 17% re-rating unaccompanied by any 8-K.

And the low beta must not be misread as safety. Idiosyncratic volatility is 16.9% against ~24% total — roughly 42% of variance is stock-specific — and both worst days of the five-year window (−11.4% and −8.5%) were single-name events on quiet market days. Lifetime maximum drawdown is −46.7%; the three-year annualized return is +4.9% with a Sharpe of 0.12, which is the honest scorecard: the reserve cycle consumed three years of compounding. A 0.42 beta here means diversifying, not safe.

Where I think consensus may be offsides. Not on direction — consensus is roughly right that this is a fairly-priced, reserve-damaged name. It may be offsides on timing and on the location of the remaining risk. The market appears to be treating the ADC as having capped the casualty problem; it capped the smaller half of it, burned its first layer immediately, and left the $22.7bn Reinsurance book untouched with an unchanged loss pick. Simultaneously, the market may be under-appreciating the mechanical power of retiring 8% of the share count annually below book. The variant perception here is not a view on value; it is that the risk sits somewhere other than where the recent price action implies it has been resolved.

12. Fact vs. Interpretation

13. Open Questions

1. Are the Reinsurance segment’s casualty reserves adequate at an unchanged 57.3% current-year loss pick, with $22.7bn outstanding and no ADC protection? The single most important unanswered question in the security.
2. Will accident year 2025 — $1,698M ultimate, 85.9% IBNR — follow the AY2022–24 pattern? The cleanest forward test.
3. $317M is unreconciled between the Reinsurance-Casualty triangle-derived FY2025 development (+$139M) and the disclosed +$456M. Candidates: pre-2016 rows, ALAE vs ULAE, differing net basis. A legitimate IR question; not presented as evidence of impropriety.
4. What did the Global Loss Triangles published 9 August 2024 show in AY2020–2023 US casualty that the Q4-2024 review then repriced five months later?
5. What is the private-credit sub-allocation within the $5.5bn LP/COLI book, and what is the new-money yield?
6. Holdco liquidity is not separately quantified in this work.
7. What is the Reinsurance segment’s casualty/property reserve split? The sensitivity analysis assumes ~50%; a higher casualty share would understate the risk.
8. Buyer and price of the sports and leisure business are disclosed nowhere in the filing set — no 8-K, no proceeds line, no counterparty named.
9. The EU tranche of the AIG renewal-rights price was redacted from the 8-K.
10. IRS audit of tax years 2014–2018 remains open to 30 September 2026 with zero unrecognized tax benefit accrued and no quantum disclosed.
11. Bermuda DTA writedown risk is disclosed but unquantified; OECD guidance restricts utilization to ~20% with a grace period through 2026.
12. ADC counterparty credit — roughly $1.0bn of the State National recoverable is uncollateralized beyond the $250M funds-withheld trust, with no parental guarantee. Not stress-tested in the disclosure.
13. Whether David Harris remains Chief Reserving Actuary, and the identity of the Group CUO — Everest does not treat these as Section 16 officers.
14. Was the Feb–Apr 2025 absence of insider buying a closed window or absent conviction? Filings cannot distinguish the two.
15. Andrade’s destination and any severance terms are not disclosed in any 8-K; two independent sources gave different destinations and neither is verifiable from filings.
16. The 2 July 2026 +3.2% day and the 6–10 July sell-side target-raise wave have no corresponding 8-K. Most plausibly benign 1H-2026 catastrophe experience plus July-1 renewal commentary — unverified.
17. Q2-2026 results fall after this report date. The 10-Q is not filed; Baltimore Bridge reserves (“a few tens of millions”) are flagged but not booked.

14. What Must Be True

For the bull case

For the bear case

The single most informative observable over the next four quarters is whether Everest raises the Reinsurance segment’s current-accident-year loss pick without being forced to by adverse development. That one act would distinguish a company whose reserving process has genuinely been rebuilt from one whose current picks are merely untested — and it is the cleanest resolution available to the central question in this security.

15. Source Appendix

See EG_source_appendix.md (delivered as Appendix B of the combined report) for the full source list with URLs and access dates.

This article takes no investment position, makes no recommendation, and sets no price target; the sole exception is the clearly-labelled Claude's Take block at the head of the document, which is the author’s own subjective opinion and is general information rather than investment advice. Report date: 2026-07-18.

APPENDIX A — Standard Diligence Questionnaire

Supplemental appendix to the research memo dated 2026-07-18. Not counted toward the article’s length standard. Answers are grounded in the underlying research and labelled Fact / Interpretation / Assumption where it matters. Where a question does not map to a reinsurer’s business model, the correct sector analog is given rather than a forced answer.

General

What thoughtful questions have other investors asked about this company?

From the Q4-2024 through Q1-2026 earnings calls, the sharpest analyst questions cluster on four themes, and management’s answers to two of them are unsatisfying.

1. “Is the charge final?” — asked in some form on every call since February 2025. Management answered affirmatively three times (Q4-2024, Q1-2025, Q2-2025) and was falsified in Q3-2025. Analysts have since stopped accepting the assurance and now ask instead for the evidence of adequacy, which management has not supplied in the form requested.
2. “How was the second charge sized?” — the most productive question asked. Williamson’s answer is the most revealing disclosure in the file: the strengthening “was all about getting ultimately to an ADC that would create finality,” and the cover was sized “less about… a certain percentage of the actuarial best estimate or a certain standard deviation” than about “putting this out in the tail so that people don’t have to worry about it anymore.” (Fact — quotation.)
3. “How far can property-cat pricing fall?” — asked directly by analyst Motemaden. Williamson declined to answer: “I don’t want to give anybody any ideas.” (Fact.) That is a legitimate competitive reason and also a non-answer to the central question about the company’s remaining profit engine.
4. The reserve-contagion question — whether the primary-book deficiency implies deficiency in the assumed-reinsurance book. This has no clean answer from outside the company, for a structural reason set out below under Business Quality.

Questions investors should be asking but largely are not: (a) why the Reinsurance segment’s current-accident-year loss pick is unchanged at 57.3% versus 57.6% in 2023; (b) what the $317M unreconciled difference between the Reinsurance-Casualty triangle and disclosed FY2025 development represents; and © how much of the State National recoverable is genuinely collateralized.

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low?

Past a cyclical high and descending (Interpretation, strongly evidenced). The reinsurance hard market peaked at the January-2025 renewals. Property-catastrophe rate-on-line is down 12% at 1/1/2026 and 16% year-to-date — the steepest annual decline since the late 1990s (Fact, Guy Carpenter, corroborated by Howden Re at −14.7% and Gallagher Re at −10% to −20%). Everest’s own disclosures track it: −10%, then −13%, then guided “mid-teens,” with the decline accelerating.

But Everest is a special case: its earnings are simultaneously at a cyclical high (industry pricing) and a company-specific low (reserve charges). FY2024 and FY2025 ROEs of 10.1% and 10.9% were depressed by $1,337M and $657M of adverse development respectively, in years when peers earned 17–21%. So “cyclical high or low” has two answers pointing in opposite directions — which is precisely why normalized earnings power is contested.

Driven by the external environment or internal actions?

Both, and separably. The external environment (hard market, then softening; rising rates lifting net investment income from $643M to $2,124M) has been favourable and is now turning. The internal actions — writing US casualty aggressively into 2021–2023 and under-reserving it — are Everest-specific and account for essentially all of its underperformance versus peers. Everest’s problem is not the cycle; the cycle was as good as it gets and Everest still earned 10%.

How stable are revenues?

Not stable, and less so than a reinsurer’s premium base implies. Gross written premium: $9.1bn (2019) → $18.2bn (2024) → $17.7bn (2025) → Q1-2026 down 18.5% year over year. Treaties re-tender annually through brokers with no contractual lock-in. Roughly all revenue “recurs” in the trivial sense of renewing and none recurs in the sense that matters.

Outlook for products/services?

Deliberately smaller. The company has non-renewed 33% of the US casualty portfolio, sold ~$2bn of renewal rights to AIG, sold its sports and leisure book and its Canadian operation, and re-segmented the remainder with a “Legacy” run-off bucket. Forward growth is negative near-term and plausibly low-single-digit thereafter.

How big will this market be — growing, shrinking, domestic or international?

Global and large: total reinsurance capital reached a record $790bn at Q1-2026 (Fact, Aon). The market is growing in capital while shrinking in price — which is the problem. Alternative capital reached a record $141bn and was the sole driver of growth. Catastrophe bond issuance set records (83 transactions in H1-2026; Q2-2026 at $11.3bn was the first quarter ever above $11bn). Demand is also rising, but supply is rising faster.

Business Quality & Competitive Moat

Is the industry getting more or less competitive?

More. This is the core industry finding. Record capital, record alternative capital, record cat-bond issuance, and rates falling 16% simultaneously. Marathon’s capital cycle is running in the direction that destroys returns.

How profitable is the business (ROIC, ROE)?

ROE is the right metric for an insurer; ROIC is not meaningful and should not be quoted (the “invested capital” of an insurer is its float plus equity, and conventional ROIC formulations produce nonsense — third-party aggregators report exactly such nonsense for this name).

Eleven-year ROE: 2015 11.9% · 2016 11.0% · 2017 5.1% · 2018 0.9% · 2019 10.2% · 2020 4.9% · 2021 12.4% · 2022 5.0% · 2023 19.1% · 2024 9.3% · 2025 10.0%. Mean ≈ 9.1%; one year in eleven above 15% (Fact). Recomputed on average equity, FY2020–25 averages 10.78%. Against an estimated ~10% cost of equity, Everest earns approximately its cost of capital and no more (Interpretation).

How profitable is the industry — how many competitors, what barriers to entry?

The industry is more profitable than Everest: Munich Re 18.3%, Swiss Re 19.6%, Hannover Re 21.4%, Arch 17.1%, W.R. Berkley 20.6% in FY2025. Perhaps a dozen globally relevant reinsurers plus the ILS market. Barriers are weak in the way that matters: the scarce input is capital, capital is the most mobile input in finance, and a catastrophe bond can price in weeks. A capital cycle with no supply lag cannot confer durable excess returns on anyone (Interpretation — and the structural explanation for the 9.1% figure above).

Can the business be easily understood?

Yes at the level of model — assume risk, collect premium, invest float — and no at the level that decides the investment case. The reserves are 1.98× equity and 70.3% IBNR: the largest number on the balance sheet is an estimate, and its accuracy has been falsified three times. An outside investor cannot independently verify it.

Can it be undermined by foreign low-cost labour?

Not applicable. Labour is an immaterial input (Everest’s reinsurance other-underwriting-expense ratio is 2.5%). The analogous threat is low-cost capital, and that threat is live and growing: alternative capital at a record $141bn is precisely low-cost capital undermining traditional reinsurance margins.

Do brands matter?

Barely. What matters is the financial-strength rating and the broker relationship. Everest holds A.M. Best A+ — but so does every meaningful competitor, which makes it a ticket of entry, not a barrier (Interpretation). Worse, the rating is currently a source of risk rather than protection: A.M. Best revised the outlook to negative on 29 October 2025 citing “elevated uncertainty surrounding the group’s business profile and enterprise risk management capabilities” (Fact); S&P went negative in February 2025.

What is the nature of competition?

Price and capacity, intermediated by brokers, re-tendered annually. Everest’s own experience is the cleanest demonstration: when it declined to match price, GWP fell 18.5% in a quarter.

Customers’ switching costs?

Effectively zero — and this is the decisive moat finding. There is no captivity, no habit, no search cost, no lock-in. The strongest evidence is the AIG transaction: AIG bought the renewal rights to ~$2bn of premium for $301M (≈15% of one year’s premium), signed and closed in a single day. A book of business that can be transferred to a competitor in a day, for 15% of one year’s revenue, is by definition not a captive customer base (Interpretation). Note also the structural reserving asymmetry: under “utmost good faith” and “follow the fortunes,” a reinsurer reserves on the cedent’s data, which it cannot independently audit — and Everest’s own primary book, which management calls “squarely in the bottom quartile,” was somebody else’s cedent.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet?

Two, both modest. (a) The franchise value of internally-built distribution and the Lloyd’s platform — Syndicate 2786 was launched de novo in 2015 and carries no book value; the Bermuda tax authorities valued Everest’s identifiable intangibles highly enough to generate a $578M deferred-tax benefit, which is an indirect market-style estimate of an asset that appears nowhere on the balance sheet. (b) Deferred acquisition costs are recognized but conservatively. Overall the balance sheet is not carrying hidden value of consequence.

Off-balance-sheet liabilities?

None of significance in the conventional sense. The material off-balance-sheet exposures are: Funds at Lloyd’s of $260M, a £150M letter-of-credit facility maturing 2029, and — most importantly — the profit-commission obligation under the ADC, which requires Everest to return 50% of favorable development up to $625M if the reserves prove redundant. That last item is an unrecognized claim on future upside and is not widely discussed.

How conservative is the accounting?

Mixed, trending toward less conservative than presented. Genuinely conservative features: zero goodwill and zero intangibles across six years (verified by exhaustive search), so book value is 100% tangible with no impairment risk; bond marks run through AOCI rather than income (unlike RenaissanceRe), so earnings are less flattered by investment gains.

Against that, four issues:

1. Reserve estimation has been falsified three times, with the deficient cohort migrating forward each time.
2. Presentation consistently leads with the smallest defensible number. Four figures circulate for the 2024 charge — $2,187M gross casualty, $1,704M “reserve action,” $1,500M pre-announced, $1,337M audited net. FY2025’s Note 4 leads with +$389M against a true total of $657M. All are disclosed; none is improper; the pattern is worth naming.
3. Netting. The FY2024 Reinsurance segment showed net zero development — $684M adverse casualty exactly offset by $684M of favorable property/mortgage. The same mechanism made FY2021–23 look clean.
4. Segment recasts. Three in ~26 months, each moving loss-making business into a run-off bucket, and the June-2026 recast reaches Footnote 4 and Schedule III — breaking comparability of the historical reserve-development triangles.

One further quality-of-earnings flag: $5.5bn of LP/COLI alternatives (12% of invested assets) at GP-supplied marks on a one-to-three-month lag, recognizing $401M of income against $195M of cash distributions — roughly $169M of non-cash NAV appreciation, ~8% of net investment income. Private-credit sub-allocation undisclosed.

How CapEx-hungry is the business?

Not applicable in the industrial sense; physical capital expenditure is immaterial. The correct analog is capital intensity: growth requires balance-sheet capital, and Everest’s assets grew from $32.7bn to $62.5bn in five years. That is the relevant “capex,” and Marathon’s asset-growth anomaly warns precisely against it — rapid balance-sheet growth in a price-taking financial business reliably precedes poor returns, which is exactly what happened.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy?

“Free cash flow” is not the right metric for an insurer — operating cash flow is premium-driven and grows simply by writing more business, so it can rise while economics deteriorate. The correct analogs are operating cash flow, book-value-per-share growth plus dividends, and the combined ratio.

Operating cash flow was $3,068M in FY2025, down 38% — but this is distorted: the ~$1,372M ADC consideration flows through operating activities, so underlying OCF was ~$4.4bn, down ~11% (Fact). Critically, prior-year paid losses have been stable at 23.7% / 23.5% / 24.2% of opening reserves — there is no payout acceleration, and the bear argument that cash flow signals reserve stress is simply wrong (Fact). Book value plus dividends compounded 11.29% over five years.

Philosophy: buybacks over dividend growth. FY2025 buybacks were 2.4× dividends; Q1-2026 4.1×.

Significant acquisitions recently?

None — verified exhaustively. Zero goodwill, zero intangibles, no ASC 805 note, no purchase-price allocation in any of FY2020–FY2025. Everest built its platform through key leadership hires and de novo licensing (Lloyd’s Syndicate 2786 in 2015).

Every corporate transaction in the ten-year window runs the other way — out, and all four share one structure: sell the going concern, retain the tail. Mt. McKinley to Clearwater/Fairfax (2015, with a 100% retrocession capped at $440.3M, raised to $450M in 2019); sports and leisure (Q4-2024, $40M gain, buyer and price never disclosed, no 8-K filed); AIG renewal rights (2025, $301M + $30M origination + $90M transition fees, with a $70M clawback if renewals fall below 80%); Everest Canada to Wawanesa (2026, C$410M, with an intra-group loss portfolio transfer stripping out the reserves). Everest is an organic underwriter in reverse gear.

Buying back shares?

Yes, aggressively and mostly well. FY2020 $200M at 0.88× book; FY2021 $225M at 1.01×; FY2025 $797M at 0.95×; Q1-2026 $331M at ~0.86×, with the quarterly floor raised from $200M to $300M. The exception is FY2024: $200M at 1.19× book, executed February–September 2024 — in the twelve months immediately before the reserve charge. Buying at 1.19× a book about to be revealed as ~$1.3bn overstated is the worst allocation decision in the window.

Issuing large amounts of new shares to insiders?

No. Equity grants totalled $68.5M over five years against a ~$14.8bn market capitalization — immaterial dilution, and Everest grants no options at all (the Form 4 corpus contains zero derivative rows). The one equity issuance was the May-2023 public offering: 4,140,000 shares at $360.00, net $1,445M — 1.52× the then-current book value, which is good execution. Everest has since repurchased 95% of those shares at 3.3% below the net issue price. The round trip is roughly neutral; the strategy it funded — expanding the casualty book — destroyed far more than the issuance created (Interpretation).

Compensation policy of directors/management?

Better designed than expected, then overridden. Fair credit first: the bonus is 60% weighted to a single metric, Adjusted Net Operating Income ROE — not premium growth — and reserve development is not excluded from it (the only carve-outs are investments, FX and a catastrophe collar). That is more honest than the attritional combined ratio Everest markets to investors, which does strip out prior-year development by construction.

But there is no reserve metric anywhere in any incentive plan (Fact, exhaustive proxy search), and four defects hollow out the design:

1. FY2024’s financial leg paid 24.2% of target — and the Committee then paid the 40% individual leg at 100% of maximum, adding $216,000 (+24%) to the CEO’s and CFO’s bonuses in the year of the $1.7bn charge.
2. The FY2025 ROE target was reset down from 17.0% to 15.0% and the maximum from ≥25.0% to ≥18.0%; the 12.4% print paid 67.5% on the new curve versus ~44% on the old.
3. The 2023 PSU award, spanning both charges, paid 102.6% of target — the strong pre-charge ROE tranche was banked and unrecoverable, and the relative-TSR sleeve paid 142% because peers were also weak.
4. The clawback has only two triggers — restatement and willful misconduct. Neither occurred. Reserve deficiency absent a restatement is not a clawback trigger.

There was no clawback, no reduction and no negative discretion. The CFO received his maximum individual bonus citing “managing a comprehensive loss reserving strategy” in the year that strategy required a $1.7bn correction, and $5.0M of purely time-vested retention stock was granted in February 2025 — the month of the charge.

Director pay: $125,000 retainer plus $325,000 in restricted shares. Say-on-pay support slipped from 94.06% to 91.95%; Compensation Committee member John Amore drew the weakest support of any nominee in both 2025 and 2026.

Motivations of management?

The new team inherited rather than created the problem, and its actions are consistent with fixing it: 33% of US casualty non-renewed, the AIG and Canada disposals, the ADC, external hires for Group Chief Actuary and Group CRO, and heavy buybacks below book. Their revealed preferences also show genuine conviction: CEO Williamson bought 1,000 shares personally at $337.97 in June 2025 and has never sold (holding up from 11,749 to 29,636 shares), and directors Galtney and Levine bought $3.50M and $0.95M at ~$306 one day after the second charge. Net insider flow is +$3.68M, with 2025 the largest buying year in the corpus.

Against that, static alignment is thin: directors and officers hold 0.7% of shares, down from 1.1%, and the ownership guideline counts unvested stock toward compliance — which makes it a restatement of the equity grant rather than an ownership requirement.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer?

None of these. Everest Group, Ltd. is a Bermuda-domiciled corporation listed directly on the NYSE as ordinary common shares. It files 10-Ks and 10-Qs as a US domestic filer (CIK 0001095073), issues a Form 1099, not a K-1, and is not an ADR. Bermuda domicile means the 15% Bermuda corporate income tax has applied since 1 January 2025; its 2023 enactment produced a one-time $578M deferred-tax benefit that must be normalized out of FY2023. US shareholders should note the standard considerations applying to foreign corporations, though Everest’s effective tax rate (15.70% in FY2025) is now broadly conventional.

Dividend policy?

$2.00 per quarter ($8.00 annually), a ~2.09% yield — and frozen since Q2-2024, nine consecutive quarters, ending a roughly 6%/year growth streak. It is very safe: 21% payout ratio, 9.2× covered by operating cash flow. The freeze began before the charge was public and coincides precisely with the period casualty was breaking internally — an under-noticed tell (Interpretation). Management has explicitly preferred buybacks at a sub-book valuation, which is defensible.

How profitable is the business?

Answered above: ~9.1% eleven-year average ROE against a ~10% cost of equity. Combined ratio above 100% in three of the last six years (102.9%, 97.8%, 96.0%, 90.9%, 102.3%, 98.6%), at the top of the best reinsurance market in thirty years.

Is net income diverging from cash from operations?

No, in the direction that would matter. FY2025 net income was $1,591M against underlying operating cash flow of ~$4.4bn — cash flow far exceeds net income, which is normal and healthy for a growing-reserve insurer. The headline 38% OCF decline is explained by the ADC consideration, not by deterioration. The decisive test is paid-loss behaviour, and it is clean: prior-year paid losses have been stable at 23.7% / 23.5% / 24.2% of opening reserves. Everest’s problem is reserve estimation, not cash conversion, and the distinction should be made explicitly rather than allowing a cash-flow bear case to be asserted.

Risks & Downside

What factors would cause the stock to decline?

In rough order of probability × impact: (1) further adverse development in the uncovered $22.7bn Reinsurance casualty book, whose current-year pick is unchanged at 57.3%; (2) adverse development on AY2025 — $1,698M ultimate, 85.9% IBNR; (3) property-cat margin compression as rate-on-line falls a further 10–15%; (4) a major catastrophe year; (5) an A.M. Best downgrade below A+, which would impair access to treaty panels; (6) reversal of the Value / low-beta factor regime currently supporting the stock, and mean-reversion of the P&C basket from its +1.9 z-score.

Risk of a catastrophic loss?

Real but bounded, and it is what the business is for. Everest’s exposure runs roughly 1% of large industry catastrophe events (Hurricane Ian: $600M on a ~$55bn industry loss). FY2025 catastrophe losses were $757M against $15.5bn of equity. A 1-in-100 industry year would be painful and would likely consume a year’s earnings; it would not by itself impair the balance sheet.

The more relevant tail is the combination risk: a large catastrophe year occurring simultaneously with further casualty development and a softening market. These risks are not independent, and the correlation structure is what could turn a fair valuation into a poor outcome.

A distinct and under-discussed exposure: ADC counterparty credit. Everest holds a $1,253M recoverable from State National against only a $250M funds-withheld trust, with no parental guarantee (unlike MS Transverse, which is guaranteed by Mitsui Sumitomo). Roughly $1.0bn is effectively uncollateralized exposure to a fronting structure retroceded to an alternative-capital vehicle. Everest has converted reserve risk into counterparty credit risk, and the recoverable disclosure does not stress-test the substitution.

Chance of a total loss?

Very low. Equity is $15.5bn against a company-disclosed reserve range of ±$2.6bn (±7.5% of gross reserves, or 14.0% of book). Debt is 18.8% of total capital with nothing due inside five years. Underlying operating cash flow is ~$4.4bn. Book value is 100% tangible with zero goodwill to impair. A total loss would require a reserve error several multiples of the disclosed range combined with a catastrophe of unprecedented scale. The realistic bear case is a ~35% drawdown from a reserve-driven book impairment and multiple compression, not a wipeout — which is why the article’s framing is “fairly valued, poor asymmetry,” not “avoid at any price.”

Recent News & Events

A data-availability note first: the AZI news feed is unusable for this ticker — ten articles over a two-month lookback, one scored, and that single scored row mis-tagged to a different issuer. No sentiment skew can be honestly reported. This section therefore rests on the 64-filing 8-K corpus and the earnings calls.

Has the business environment changed recently?

Yes, in two opposite directions. Adversely: the reinsurance pricing cycle turned decisively — property-cat rate-on-line down 16% year-to-date, record capital, record cat-bond issuance, and both Munich Re and Swiss Re guiding FY2026 down. Favourably for Everest specifically: the retrenchment is executing, and Q1-2026 delivered a 91.2% combined ratio and 16.7% annualized operating ROE — though 12.2 of the 13.1 points of loss-ratio improvement were catastrophe-driven, with underlying attritional improvement of just 1.9 points.

Significant acquisitions?

None. Four disposals (see Capital Allocation above).

Change in accounting policies?

No change in accounting policy, but three material changes in segment presentation in ~26 months: A&H realigned (Q4-2023); an “Other” run-off segment created (Q4-2024); and the three-segment structure — Reinsurance Treaty / Global Wholesale & Specialty / Legacy — effective 1 January 2026. The 8-K of 3 June 2026 formally recast the FY2025 10-K, and the recast reaches Footnote 4 (Reserve for Losses and LAE) and Schedule III — breaking comparability of the historical reserve-development triangles against everything published before June 2026. It carries two audit-firm consents (KPMG and PwC), reflecting the FY2024 auditor transition from PwC to KPMG. Separately, guidance was withdrawn alongside the January-2025 charge in favour of a “mid-teens total shareholder return over the cycle” objective.

Recent changes — new markets, facilities, management?

Management turnover has been near-total, and the most important layer is invisible in SEC filings. Everest does not designate its Chief Underwriting Officer, Group Chief Actuary, Group CRO or Chief Claims Officer as Section 16 officers — the strings “Chief Actuary” and “Chief Claims” appear zero times in the FY2022–FY2025 10-Ks, and no individual holding those roles is named in any proxy or 8-K, ever.

Disclosed: Insurance division head Karmilowicz resigned 2 December 2024; CEO Andrade resigned 3 January 2025; Williamson became permanent CEO 22 January; the charge was pre-announced 27 January. The General Counsel departed with a $7.25M separation and a waived non-compete; CFO Kociancic retires July 2026, replaced by Elias Habayeb (ex-Corebridge/AIG) on ~$10.7M of inducement. Five new directors joined March–August 2025; Chair Taranto retired.

Not disclosed in filings (trade press only): NA Insurance President Mulray exited August 2024; Group CRO Moskowitz left ~June 2025, leaving the seat vacant roughly six months — spanning exactly the window in which the Q3-2025 charge was determined and the $1.2bn ADC was designed; his successor Hazelton exited October 2025; Katy Bradica (ex-AXA XL) and Attila Kerényi (ex-Swiss Re) were named Group Chief Actuary and Group CRO in December 2025, with neither announcement naming a departing incumbent; and Melissa McDermott — the chief actuary hired in 2023 specifically as part of the remediation — left for CNA by June 2026.

New markets/facilities: none of consequence; the direction is contraction. Everest internalized its Lloyd’s managing agency on 18 August 2025 (Everest Managing Agency Limited), and the FY2025 Enterprise Risk Committee roster newly includes a Group Chief Underwriting Officer and a Chief Executive of Legacy Operations — control and run-off roles created after the charges.

Rating-agency actions: A.M. Best revised the outlook to negative on 29 October 2025 (A+ affirmed), citing “elevated uncertainty surrounding the group’s business profile and enterprise risk management capabilities.” S&P moved negative in February 2025. No downgrade has occurred.

Litigation: No securities-litigation 8-K, no rating-action 8-K and no restatement (Item 4.02) exists anywhere in the five-year corpus. The charge was handled entirely as prospective strengthening; no non-reliance determination was made.

One item flagged but not booked: Williamson noted “a few tens of millions of dollars of incremental loss reserve” on the Baltimore Bridge. Q2-2026 results fall after this report date.

APPENDIX B — Source Appendix

Companion to the research memo dated 2026-07-18. Primary sources listed first, by source reliability. All web sources accessed 2026-07-17/18 unless otherwise noted. Everest Group, Ltd., CIK 0001095073.

A. SEC Filings — Everest Group, Ltd. (primary)

The trailing 60-month corpus was enumerated and mirrored locally to output/EG/sources/ (MANIFEST.csv, 334 rows). 367 filings since 2021-07-01; excluding 424B*/FWP/144 noise: 222 Form 4 · 64 Form 8-K (+3 8-K/A) · 15 Form 10-Q · 12 Form 3 · 5 Form 10-K · 5 DEF 14A · 14 SC 13D/G.

Annual and quarterly reports

Note: the Q2-2026 10-Q was not filed as of the report date. No 1H-2026 filing data exists.

Form 8-K — material events (selected from the 64-filing corpus, 2021-07-28 to 2026-06-03)

Proxy statements (DEF 14A)

DEF 14A filed 2024, 2025 and 2026-04-10 — relied upon for: the 60%/40% bonus structure and Adjusted Net Operating Income ROE definition (p.55 fn.1, p.65); the FY2024 individual-component maximum award language (2025 proxy, p.62, including the Kociancic “managing a comprehensive loss reserving strategy” citation); FY2025 Summary Compensation Table (p.70); employment and severance agreements (p.81); PSU design and the 2023 award payout at 102.6% of target; clawback triggers; stock-ownership guidelines; Item 402(v) pay-versus-performance measures (including “Gross Written Premium Annual Growth Rate”).

Forms 3, 4 and 5

All 222 Form 4s and 12 Form 3s parsed (files carry .xml extensions but are SEC-rendered HTML; parsed from the rendered Table I grids). 176 filings / 231 transactions over 2021-07-06 to 2026-07-02. Zero Table II (derivative) rows and zero code-M — Everest grants no options. Extraction saved to output/EG/2026-07-18/_scratch/eg_form4_parsed.csv.

Note on an internal correction: an initial pass reported 27,940 shares of code-P purchases and concluded no officer bought post-charge. A rigorous re-parse produced an itemized table summing to 22,255 shares and establishing that CEO Williamson bought 1,000 shares at $337.97 on 2025-06-11. The itemized figure is used in the article; the earlier figure was rejected.

B. Earnings-call transcripts (via ROIC.ai MCP; copies saved to output/EG/transcripts/)

C. Peer and industry primary sources

D. Quantitative data services

E. Trade press (used only where SEC filings are silent — lower-tier sources)

Everest does not designate its Chief Underwriting Officer, Group Chief Actuary, Group CRO or Chief Claims Officer as Section 16 officers, so the turnover of the entire actuarial and risk-control layer appears in no SEC filing. The following are the sole sources for that finding and are labelled as secondary throughout the article.

• The Insurer — “Everest parts ways with NA insurance president Mulray in management shake-up” (August 2024)
• The Insurer — “Everest Global Insurance chairman Karmilowicz resigns” (2024-12-04)
• The Insurer — “Selective adds Everest’s Moskowitz as Chief Risk and Reinsurance Officer” (2025-06-25)
• The Insurer — “Hazelton to exit Everest as retail sale to AIG prompts restructuring” (2025-10-28)
• The Insurer — “CNA hires Everest insurance chief actuary McDermott” (2026-06-11)
• Businesswire — “Everest Announces Key Leadership Appointments” (2025-12-03) — Bradica (Group Chief Actuary), Kerényi (Group CRO)
• Reinsurance News — “Everest names Bradica as Group Chief Actuary and Kerényi as CRO”
• Businesswire — “Everest Appoints Jill Beggs as EVP and CEO of Reinsurance” (2025-07-24)
• Carrier Management — “‘Underwriting Choices’ Added to Everest’s Social Inflation Woes: CEO” (2025-01-31)
• Reinsurance News — Hannover Re FY2025 results (2026-03-12); SCOR FY2025 results (2026-03-04)

F. Comparative sources

Same-sector peer disclosures were read for comparative and cycle framing, and are cited in Section C above. The peer set used throughout is RenaissanceRe, Arch Capital, W. R. Berkley, Markel Group, Chubb, RLI, Kinsale and American Financial Group among US/Bermuda GAAP filers, and Munich Re, Swiss Re, Hannover Re and SCOR among IFRS 17 reporters. Every comparative figure is sourced to the relevant company’s own annual report, results release or SEC filing.

No position is held, stated or implied in Everest Group or in any company named in this article.

G. Analytical frameworks

• investment-research-frameworks skill (.claude/skills/) — Greenwald & Kahn, Competition Demystified (barriers to entry as dominant; the three genuine advantage types; the market-share-stability and ROIC tests) applied above; Marathon Asset Management / Chancellor, Capital Returns (supply-side capital-cycle analysis; the asset-growth anomaly) applied above and

H. Known gaps and unverified items

Stated explicitly rather than papered over. Each also appears in the Open Questions section.

1. $317M unreconciled between the Reinsurance-Casualty triangle-derived FY2025 development (+$139M) and the disclosed +$456M.
2. Buyer and price of the sports and leisure business — disclosed nowhere in the filing set; no 8-K, no proceeds line, no counterparty named.
3. EU tranche of the AIG renewal-rights price — redacted from the 8-K, deferred to a 10-K exhibit.
4. Andrade’s destination and any severance — not in any 8-K. Two independent research streams in this engagement returned different destinations (USAA; Ameriprise); neither is verifiable from Everest’s filings and neither is asserted in the article.
5. A&E survival ratio reported as both 7.0 years and 4.7 years by different workstreams; presented as a range, not resolved.
6. Reinsurance segment casualty/property reserve split — undisclosed; the sensitivity analysis assumes ~50% and would understate risk if the casualty share is higher.
7. Private-credit sub-allocation within the $5.5bn LP/COLI book; new-money yield; holdco liquidity — none separately quantified.
8. Hannover Re’s EUR 3.2bn “reserve resiliency” figure — ambiguous between cumulative and incremental in the secondary source; not used as a flow figure.
9. SCOR’s 82.3% FY2025 combined ratio rests on a secondary source (the company release surfaced only the Q4 standalone 80.9%).
10. Munich Re’s current share price (EUR 515.60) is secondary-sourced (Trading Economics); the AZI feed returned nothing for MUV2 variants. Treat as ±1–2%.
11. No clean 5-year IFRS 17 BVPS series exists for any European reinsurer — the standard begins in 2022/2023. ROIC.ai’s SCOR and RNR book-value lines are corrupted (tangible BVPS exceeding total BVPS) and were rejected.
12. European P/B figures use YE2025 book against a July-2026 price — none has reported H1-2026 — so they modestly overstate the multiples.
13. The 2 July 2026 +3.2% day and the 6–10 July sell-side target-raise wave have no corresponding 8-K. Attribution to benign 1H-2026 catastrophe experience and July-1 renewal commentary is unverified interpretation.
14. Q2-2026 results and the Q2-2026 10-Q fall after this report date. Baltimore Bridge reserves are flagged but not booked.
15. IRS audit of tax years 2014–2018, open to 2026-09-30 with zero unrecognized tax benefit accrued and no quantum disclosed; Bermuda DTA writedown risk disclosed but unquantified.