Arch Capital Group Ltd.（ACGL） · Insurance

Arch Capital Group Ltd. Deep Value Research

May 21, 2026·~19 min read

Quick ReadPlain-language overview · read this first

Arch Capital is a three-platform insurer spanning specialty insurance/reinsurance/mortgage, trading at $95.87, 1.45x PB and 7.4x PE. Cautious Buy.

Its moat comes from S&P AA-/A.M. Best A+ high-rating access, a disciplined underwriting culture, and three-platform diversification. In 2025 the insurance combined ratio was 95.2% and operating ROE 17.1%; BVPS has compounded at a 16.2% six-year CAGR, with $7.8 billion of cumulative buybacks shrinking the share count by 11.5%. But 2023-2025 profits may embed a cyclical dividend, and long-tail reserve under-provisioning, underwriting softening and mortgage deterioration could distort the low PE.

Conservative owner earnings of $2.2-2.6 billion imply 13-15x; the DCF's three cases give $81/$119/$172. Ideal buy $80-90 (1.2-1.35x BVPS); in an extreme scenario $42-50, a permanent drawdown of 48-56%, so the margin of safety is insufficient and only small, tranche-by-tranche buying is warranted.

Lead
A high-quality insurer built on disciplined underwriting and multi-platform diversification, with book value per share compounding at roughly 16% CAGR over six years and sustained buybacks accreting per-share value. At about $95.87 it trades near 1.45x book, reasonable but only slightly cheap, with a limited margin of safety. Rated Cautious Buy, with an ideal buy range of $80-90.

Full report

Prices in the article are as of publication; see the valuation band above for the live price.

Conclusion First

Let me lead with the conclusion: Arch Capital Group Ltd. is a high-quality insurer I'm willing to track for the long term and to hold like a "business owner" when the price is right, but at today's price it looks more like a good company you "can build a position in cautiously" than a bargain that is "materially undervalued and worth loading up on." As of the latest available price, ACGL trades at about $95.87, for a market cap of roughly $33.8 billion; against book value per share of $66.19 at the end of Q1 2026, that is about 1.45x PB; against trailing-twelve-month diluted EPS of roughly $13.02, that is about 7.4x TTM PE. On the books it isn't expensive, but insurance profits are cyclical, and the high-earnings phase of 2023-2025 may not represent the long-run norm, so the margin of safety is not obvious.

To avoid "mixing facts and judgments," the whole report follows four conventions: Facts come mainly from the company's annual reports, 10-Ks, 10-Qs, quarterly supplements, proxy statements and the official investor-relations pages; Assumptions are used only for valuation; Inferences are clearly flagged; and Opinions appear only in the rating and the final recommendation. Based on the high-confidence evidence I currently have, my rating on ACGL is: Cautious Buy. It suits long-term value investors who understand the insurance cycle, accept swings in book value, and hold for 10 years or more; it is less suited to those who watch only near-term earnings and try to force a free-cash-flow model onto an insurer as if it were an ordinary manufacturer.

Item	Assessment
Investment rating	Cautious Buy
Core judgment	A good business, but not a no-brainer bargain; a good management team, but an industry that is inherently volatile; high-quality capital allocation, but only a moderate margin of safety at today's price.
Margin of safety at current price	Not obvious
Suitable investor type	Long-term value investors; compounding-oriented investors who understand the insurance cycle
Biggest uncertainty	Softening of the insurance/reinsurance pricing cycle, long-tail reserve error, and deterioration in housing and employment hitting mortgage-insurance profits

My one-sentence investment thesis: ACGL's long-term value comes from "disciplined underwriting + high-quality float investing + opportunistic buybacks + multi-platform diversification," but the current valuation only offers a "reasonable-but-slightly-cheap" entry, and not the kind of deep Buffett-style discount. The biggest uncertainty behind this judgment is whether its high ROE of the past two or three years embeds a strong cyclical dividend; whether, as P&C competition intensifies in the coming years, the company can still sustain a mid-to-high double-digit long-run operating ROE; and how resilient mortgage insurance proves in an economic downturn.

The Business and the Industry

Arch's business model isn't complicated, but its accounting is. The company essentially does three things: first, it takes on risk through insurance and reinsurance contracts, collecting premiums up front and paying claims later; second, it invests the "float" that accumulates before claims are paid into fixed income, equities and other investments; third, in mortgage insurance and GSE credit-risk-transfer business, it earns risk pricing, service fees and investment income. In its 2025 annual report the company explicitly defines itself as an enterprise writing specialty-lines insurance, reinsurance and mortgage insurance/reinsurance globally, and emphasizes that the market is highly competitive and subject to an underwriting cycle.

By customer and revenue model, the insurance business serves corporate clients and broker channels, the reinsurance business serves primary insurers, and the mortgage-insurance business serves lenders, risk-sharing arrangements tied to government-sponsored enterprises, and capital-market participants. Revenue isn't "subscription-based," but it isn't a one-off sale either: premiums are earned pro rata over the policy period, while claims may be paid years later, making this a business of collecting cash up front and performing later. In its 2025 10-K the company writes that insurance-segment net premiums earned are recognized pro rata over the typically 12-month policy period; at the same time it repeatedly cautions that the actual timing of claim payments varies significantly with large losses, the legal environment and catastrophes. In other words, revenue has some recurring character, but cash flow and profit recognition are never as smooth as for consumer goods.

The cost structure is typical of insurance: the core items are losses and loss-adjustment expenses, acquisition costs, and other operating expenses. For an insurer, what really matters is not "gross margin" but the combined ratio, reserve adequacy, investment yield and return on capital. Take the 2025 insurance segment: its combined ratio was 95.2%, still below 100%, meaning underwriting itself still made money. In Q1 2026 the company's insurance-segment combined ratio was 96.5% and its reinsurance-segment combined ratio 75.9%, a marked improvement after the high base set by the California wildfires. The mortgage business depends more on mortgage delinquencies, home prices, unemployment and policy persistency. The company disclosed a persistency rate of 81.8% for its U.S. primary mortgage portfolio at the end of 2025.

Is this a business "I can understand"? The answer is yes, but you can't be lazy about it. What you understand is the economic essence: whether underwriting makes money, whether float quality is high, whether discipline holds through the cycle, and whether capital allocation is rational; you should not be thrown off by complex investment accounting, reserve conventions, and seasonal/catastrophe noise. If the stock market closed for five years, I'd be willing to hold this business, provided I accept that intervening years can swing sharply because of catastrophes, reserves or the housing cycle. Business understandability: 4/5.

On the industry, P&C insurance and reinsurance is not a declining industry but a mature, stable-demand, strongly cyclical one. Long-run demand is stable because both companies and individuals need to transfer risk; but profits swing with the supply of capital, catastrophe losses, interest rates, liability-cost inflation and regulatory change. The company itself cautions in its 2025 annual report that competition had increased as it entered 2026; Chubb publicly stated in Q1 2026 that parts of the property and financial-lines markets were already "soft or softening," and it therefore proactively reduced certain shared-layer property exposures. On mortgage insurance, Fitch believes the U.S. mortgage-insurance industry still has strong capital and high policy persistency to support it in 2026, but housing affordability and macro weakness remain risks to monitor. All told, this is not an "inherently wonderful industry," but it is an industry that can breed good companies. Industry attractiveness: 3/5.

Moat and Management

ACGL's moat is the kind more typical of, and harder to replicate in, insurance, rather than a consumer-brand moat or a platform network effect: ratings and regulatory licenses, a specialist underwriting culture, a diversified platform, long-standing distribution relationships, and capital plus capital-allocation ability. Its main insurance subsidiaries carry relatively high ratings such as S&P AA-, Moody's A1, Fitch AA-, and A.M. Best A+; these are not decorative labels but the infrastructure for underwriting large and complex risks, retaining brokers and corporate clients, and gaining "entry qualification" in reinsurance and capital-market transactions. A rating can be lost overnight but is very hard to rebuild quickly.

The second layer comes from specialist underwriting capability and culture. Across multiple annual reports Arch repeatedly stresses its "cycle management philosophy": holding discipline in soft markets, scaling up underwriting in hard markets, and dynamically allocating capital across businesses and geographies. In its 2023 shareholder letter the company attributes its long-run success to specialty lines, cycle management, and allocating capital to the opportunities with the highest risk-adjusted returns; its 2025 annual report continues to emphasize strengthening risk selection through data and analytics and treats a "culture that attracts top talent" as a source of differentiation. This kind of moat shows up not in patents but in fewer mistakes, disciplined pricing, decisive exits, and preserved float quality.

The third layer comes from platform diversification. Arch runs specialty insurance, global reinsurance and mortgage as three platforms at once, a structure that lets it hedge catastrophe volatility with mortgage profits in some years and capture high-payoff opportunities in a hard reinsurance market in others. In its 2022 annual report the company describes itself as "the only globally diversified insurer of mortgage credit risk"; even though this is management's own phrasing, the underlying fact still holds: ACGL's mortgage-insurance platform is different from most pure-P&C peers, giving it a more distinctive earnings profile and capital-allocation options.

Is this moat widening, stable or narrowing? My judgment is: broadly stable, with some local cyclical stretch and compression. In a hard market, the underwriting culture and capital strength look wider; in a soft market, as price competition intensifies, the moat can "look narrower." For a competitor to replicate it takes not only capital but years of claims, underwriting, broker relationships and rating history; that usually can't be done in two or three years, nor by throwing money at it alone. It has some pricing power, but that power comes from the discipline to exit business that isn't profitable enough, rather than from brand. In an inflationary environment, short-tail lines and newly written business can be repriced fairly quickly, while long-tail liability lines are more dangerous; in a downturn the company can still be profitable, but its earnings elasticity drops noticeably. Moat strength: 4/5.

On management, the strengths are very clear. First, equity alignment is not weak: as of March 9, 2026, CEO Nicolas Papadopoulo held about 1.6599 million shares, CFO François Morin about 738,200 shares, and Maamoun Rajeh about 894,700 shares; all directors and officers together held about 11.6187 million shares, roughly 3.3% of common stock. This isn't a Berkshire-style controlling stake, but it's already enough to make executives genuinely feel the pain. Second, the compensation metrics are broadly sound: the 2026 proxy shows 77% of the CEO's target pay is performance-linked and 68% is long-term incentives; long-term performance shares are anchored to three-year TBVPS growth, with a TSR modifier. All NEOs meet the stock-ownership requirements, and the company also requires executives to retain 50% of net shares acquired until they hit the threshold.

Capital allocation also deserves credit, though not without blemishes. The positives: the company has used buybacks over the long term and stresses an "opportunistic basis"; since the buyback program began through the end of 2025, it has repurchased about 455 million shares in total for roughly $7.8 billion; in 2025 alone it repurchased about $1.9 billion of common stock, and by the end of Q1 2026 the share count had fallen to 352.9 million shares. That is a genuine boost to per-share intrinsic value. On the other hand, two notes of caution are warranted: first, during the 2024 leadership transition the company granted a one-time outperformance equity award, which pushed the 2025 say-on-pay support rate down to 84.7%; second, an insurer buying back heavily in good years is not automatically "smart," and whether buybacks truly create value still depends on how deeply the repurchase price discounts intrinsic value. On balance, my impression of management is: long-term-oriented, broadly rational in capital allocation, with an incentive framework fairly well aligned with shareholders, but the new CEO's buyback discipline still needs continued observation. Management and capital allocation: 4/5.

Financial Quality and Owner Earnings

Let me start with a key principle: an insurer's "free cash flow" cannot be computed by copying the industrial-company formula. The reason isn't that it doesn't generate cash, but that collecting premiums before paying claims makes operating cash flow look very handsome, while a substantial part of it must remain within the system as reserves, solvency capital and rating capital, and cannot simply be treated as "money that can be fully distributed to shareholders today." In its 10-K the company itself explicitly cautions that the timing and amount of claim payments can cause large swings in operating cash flow across periods, and in extreme cases may even require selling investment assets to meet claims.

The table below summarizes the historical measures I find most useful: net income to common, operating cash flow, the very small capex on fixed assets, book value per share, and the change in share count. Here "CFO minus fixed-asset capex" is only a proxy for cash generation, not the owner earnings I ultimately use for valuation. The data are compiled from the company's 2019, 2020, 2021, 2023 and 2025 annual reports and the latest quarterly supplement.

Year	Net income to common	Operating cash flow	Fixed-asset capex	CFO − Capex	BVPS	Approx. year-end shares
2019	$1.595 billion	$2.048 billion	$38 million	$2.011 billion	$26.42	405.6 million
2020	$1.364 billion	$2.887 billion	$40 million	$2.847 billion	$30.31	406.7 million
2021	$2.093 billion	$3.428 billion	$41 million	$3.386 billion	$33.56	378.9 million
2022	$1.436 billion	$3.816 billion	$50 million	$3.766 billion	$32.62	370.4 million
2023	$4.403 billion	$5.749 billion	$52 million	$5.697 billion	$46.94	373.4 million
2024	$4.272 billion	$6.673 billion	$51 million	$6.622 billion	$53.11	376.4 million
2025	$4.359 billion	$6.172 billion	~$44 million	~$6.128 billion	$65.11	359.0 million

This table tells us several very important things. First, book value per share rose from $26.42 to $65.11 between 2019 and 2025, a six-year CAGR of about 16.2%; even accounting for the pause caused by the 2022 capital-market and catastrophe shocks, this long-run curve is still very strong. Second, the share count fell from about 405.6 million to 359.0 million, a shrinkage of roughly 11.5%, showing that management has indeed been returning capital on a "per-share" basis over the long run rather than blindly chasing scale. Third, reported profit and cash flow broadly match, and capex is extremely light, an advantage of an asset-light financial company.

But I won't declare it an unbeatable cash cow just because "CFO is large." What really matters is whether the profit is "real cash profit." My view is that ACGL's earnings quality is broadly high, but two layers of adjustment are needed. The first layer: strip out fair-value swings in equity investments, FX, and certain one-off items, and look at profitability on an operating basis rather than as single-quarter GAAP noise; the company itself habitually communicates in terms of after-tax operating income and operating ROAE. The second layer: from operating profit, further deduct the capital that must be retained for future growth and solvency. Because of this second layer, an insurer's owner earnings are usually materially below CFO minus capex.

For ACGL's owner earnings I use a conservative rather than flattering approach. Fact: common equity was about $23.376 billion at the end of 2025 and about $19.990 billion at the end of 2024; the company reported an annualized operating ROE of 17.1% for 2025. Inference: if the simple average of 2024 and 2025 common equity approximates average common equity for 2025, then 2025 operating earnings attributable to common shareholders were roughly $3.7 billion. Conservative deduction: further subtracting about $40-50 million a year of maintenance capex on fixed assets, plus the $1.1-1.5 billion of growth/safety capital I believe must be retained to sustain mid-single-digit to low-double-digit growth, rating capital and reinsurance volatility, gives conservative owner earnings of about $2.2-2.6 billion. Against a current market cap of about $33.8 billion, that is roughly 13-15x owner earnings. That isn't expensive, but it is by no means a "cigar-butt" bargain.

On the other financial questions you especially care about, my read is as follows: ROE/ROA: for 2025 the company disclosed an annualized ROE of 20.1% and operating ROE of 17.1%; for Q1 2026 annualized ROE/operating ROE were 17.8%/15.4%, still solid. Leverage: at the end of Q1 2026 senior notes were about $2.729 billion, cash $914 million, and total shareholders' equity $24.188 billion; looking only at interest-bearing debt, debt/total capital is about 10% and net debt/common equity about 7.8%, so financial leverage is not aggressive. Net debt/EBITDA, EV/EBITDA: of weak relevance for an insurer, because "liabilities" are largely operating float rather than debt in the industrial sense, and EBITDA itself doesn't capture underwriting and reserve economics well. Inventory: not applicable. Receivables/payables: receivables are mainly premiums receivable and reinsurance recoverables; payables are mainly reinsurance balances payable, contractholder payables, and unearned-premium/loss reserves. At Q1 2026, premiums receivable were $6.526 billion, reinsurance recoverable $9.732 billion, and loss reserves $34.105 billion, all items whose quality and direction of change need continuous monitoring.

On the evidence I've seen so far, there is no direct sign of obvious financial fraud or aggressive accounting. On the contrary, the company discusses the cycle, competition, claims volatility and liquidity constraints fairly candidly in its shareholder letters; its 2023 letter even states explicitly that on reserves it would rather "recognize bad news quickly and good news slowly." Of course, the risk in insurance accounting is never "invisible fraud" but reserve under-provisioning that only surfaces years later. That is why I place the risk of permanent capital loss mainly on long-tail reserve risk rather than on short-term share-price swings.

Valuation and Margin of Safety

I won't offer just a "gut-feel target price"; instead I use owner-earnings discounting, relative valuation, and asset value together to constrain the conclusion. First the overall verdict: if you treat ACGL as a high-quality insurer capable of a normalized operating ROE of 12%-14% over the medium-to-long term, it isn't expensive now; if you demand a clear 20%-30% margin of safety, it isn't cheap enough yet.

Owner-Earnings Discounting

The model below isn't about "predicting the share price" but about making my cautious assumptions on normalized owner earnings explicit. The starting point uses the conservative owner-earnings range of $2.2-2.6 billion above; the discount rate runs between 8% and 10%; growth is 4%-8% in the first five years and 2%-4% in the next five; the terminal value is handled at 11-13x owner earnings. Note: this is itself an assumption, not management guidance. The model's anchors come from the company's current share base, book capital and operating ROE.

Dimension	Conservative	Neutral	Optimistic
Starting owner earnings	$2.2 billion	$2.6 billion	$3.0 billion
First-five-year growth	4%	6%	8%
Next-five-year growth	2%	3%	4%
Discount rate	10%	9%	8%
Terminal multiple	11x	12x	13x
Estimated equity value	~$28.5 billion	~$42.0 billion	~$60.7 billion
Intrinsic value per share	~$81	~$119	~$172

The point of these results isn't that "81/119/172 is the true answer," but that the current price sits roughly above the conservative value and below the neutral value. So it looks more "reasonable-but-slightly-cheap" than "clearly undervalued." If your investment discipline requires a "buy price below conservative intrinsic value or near the conservative value," then it isn't cheap enough today; if your discipline is "start scaling into a high-quality compounding machine at a 20% discount to neutral value," then today is an acceptable entry point.

Relative Valuation

For an insurer, PB, PE, operating ROE and TBVPS growth are more meaningful than EV/EBITDA. The table below compares using Q1 2026 figures and a recent price snapshot. Note: the peers' business mixes are not identical, with RNR skewing toward reinsurance, WRB toward high-quality commercial P&C, and Chubb larger and more globally diversified, so relative valuation can only serve as a "boundary check," not a mechanical template.

Company	Latest price	BVPS	PB	Operating/core ROE	Observation
ACGL	$95.9	$66.19	1.45x	15.4%	Valued below Chubb, far below WRB, above the more "cat-resilient" RNR
RNR	$303.1	$250.48	1.21x	21.8%	Cheaper, but more reinsurance-heavy and more volatile
WRB	$68.5	$26.13	2.62x	21.2%	A clear quality premium; the market awards a high PB
CB	$328.4	$189.93	1.73x	14.0%	A scale-and-stability premium; high quality but more expensive

The relative comparison tells me two things. First, ACGL is not an expensive stock. Second, the market hasn't priced it as a "deep value stock" either, placing it instead within the "reasonable range for a high-quality insurer." If you look only at PE, ACGL looks astonishingly cheap; but once you factor in the 2023-2025 earnings boom and the buyback effect, 1.45x PB is actually closer to "fair-to-slightly-cheap" than to "a fire sale." More broadly, the current U.S. 10-year Treasury yield is about 4.67%, while ACGL's earnings yield on trailing-four-quarter EPS is about 13.6%; meanwhile the S&P 500's forward PE is roughly 22x. Against these two alternatives, ACGL's current static valuation is indeed more attractive, provided you accept that insurance earnings are more cyclical and more dependent on reserve quality than index constituents.

Asset Value

For an insurer like ACGL, the core of the asset- or liquidation-value method isn't plant and equipment but the liquidity of the investment portfolio, the reliability of liability reserves, the share of goodwill/intangibles, and whether common book value is close to true realizable value. At the end of Q1 2026 the company had total investment assets of about $46.786 billion, cash of $914 million, total assets of $81.446 billion, and total shareholders' equity of $24.188 billion, of which goodwill and intangibles were about $1.190 billion; common book value per share was $66.19. This shows the company's "asset floor" is mostly financial assets rather than hard-to-value heavy assets. For ACGL, PB/TBVPS is more meaningful than a net-liquidation discount.

Combining the three methods, I offer the following price ranges: Conservative intrinsic-value range: $75-90. Fair intrinsic-value range: $105-125. Optimistic intrinsic-value range: $150-175. Mapped to trading discipline, my preferred ideal buy range is $80-90, roughly 1.2-1.35x current BVPS; the acceptable holding-price range is $90-115; and if the price is clearly above $125, I'd consider it to have entered "good company, bad price" territory for a conservative long-term investor. At the current roughly $95.9, the market isn't offering a particularly thick margin of safety, so the margin-of-safety conclusion: insufficient, but not nonexistent.

Risks, Counterarguments and Comparisons

The most important risk is permanent capital loss, rather than "volatility." I rank them by importance below.

Reserve risk. This is the risk every excellent insurer ultimately fears most, especially in long-tail liability, medical liability and excess liability; if persistent adverse reserve development shows up years later, today's seemingly cheap PE will instantly distort.
Softening of the underwriting cycle. The company itself says competition increased entering 2026; Chubb has also publicly acknowledged that parts of property/financial lines are already soft or softening. If prices fall and the company can't cut volume fast enough, normalized ROE will retreat.
Catastrophe and climate risk. Reinsurance and commercial property are inherently exposed to hurricanes, wildfires, earthquakes and the like; the Q1 2025 California wildfires are a ready example. Extreme years hit profit, capital and market valuation simultaneously.
Mortgage-insurance risk. The industry is currently supported by high policy persistency and relatively firm home prices, but if unemployment rises, home prices fall, or natural disasters damage the value of mortgaged homes, loss ratios on mortgage insurance and CRT could climb.
Rating and regulatory risk. High ratings are a moat but also a vulnerability; once underwriting deteriorates or capital is impaired enough to trigger a downgrade, competitiveness is systematically weakened.
Capital-allocation risk. The new CEO era's buyback discipline still needs verification; sustained heavy buybacks at cycle highs and un-cheap valuations would dilute long-term value creation. The one-time incentive during the 2024 leadership transition also reminds us that governance, though good, is not perfect.

The strongest counterargument goes roughly like this: the ACGL you see now is too good, so good it's actually abnormal. In 2023-2025 profits benefited simultaneously from a hard reinsurance market, higher investment income from higher rates, low mortgage loss ratios, and the per-share amplification from buybacks. Bears would say today's PE of a bit over 7x isn't "cheap" but "the illusion of peak-cycle profit"; once the market softens, catastrophes multiply and reserves start to be topped up, the market's PB will first fall from 1.45x back to 1.1x-1.2x, earnings will retreat too, and a double whammy appears. This counterargument isn't absurd; I regard it as the short thesis most deserving of serious attention today.

What facts, if they appear, would make me admit my initial judgment was wrong? First, underwriting profit deteriorating markedly for several consecutive quarters after excluding catastrophes, signaling a loss of discipline. Second, persistent and sizable adverse reserve development, especially concentrated in long-tail liability. Third, a rating downgrade or clear pressure on regulatory capital. Fourth, mortgage loss ratios jumping persistently while persistency/new-business quality deteriorate in tandem. Fifth, management continuing to buy back heavily when valuations are elevated and the market is softening, showing capital allocation has begun to serve short-term EPS rather than intrinsic value.

Compared with other opportunities, my conclusion is: at the current price, ACGL is very likely superior to holding long-term Treasuries, and has a chance of beating the broad index, but it isn't necessarily a "must-overweight-over-the-index" large position. Against the 10-year Treasury, ACGL's earnings yield is clearly higher; against the S&P 500, its current valuation is also lower. The problem is that the index's certainty comes from diversification, while ACGL's comes from underwriting culture and governance, which, though real, aren't as "naturally stable" as a high-quality consumer or platform business. If you could hold only five assets, I'd say: it's on the edge of the shortlist, worthy of consideration, but not among my most-certain top five.

Open Questions and Limitations

This research has a few boundaries I deliberately keep. First, I did not directly cite the company's precisely disclosed 2025 after-tax operating income in the final draft, instead making a conservative inference from operating ROE and average common equity. Second, for insurers, EV/EBITDA, net debt/EBITDA and ROIC are of limited help in judging value, so I substitute PB, BVPS/TBVPS, operating ROE and owner earnings. Third, the peer comparison uses Q1 2026 and recent price snapshots, and the business mixes aren't fully comparable, so it can only be used as a boundary check, not a pricing formula.

Investment Checklist and Final Conclusion

First the checklist. Here "pass/fail/uncertain" is my judgment, not the company's self-description.

Checklist	Conclusion
Can I understand this business	Pass
Does it have stable long-term demand	Pass
Does it have a durable moat	Pass, but not an extremely deep moat
Does it have pricing power	Partial pass
Can it generate stable free cash flow	Uncertain
Is its return on capital excellent	Pass
Is management trustworthy	Pass
Is capital allocation rational	Pass, but keep watching buyback-price discipline
Is the balance sheet sound	Pass
Is the valuation below intrinsic value	Pass, but the discount is limited
Is the margin of safety sufficient	Fail
Am I comfortable holding long term	Pass, but you need to understand the insurance cycle
What key facts would make me sell	Deteriorating reserves, a rating downgrade, a loss of underwriting discipline, mortgage credit deterioration
Am I buying just because the price rose or on emotion	Fail

【Final Rating】 Cautious Buy

【One-Sentence Investment Thesis】 ACGL is a high-quality insurer driven by disciplined underwriting, high-quality rated capital, a diversified platform and long-run buybacks, but the current price is only "reasonable-but-slightly-cheap," not "materially undervalued."

【Core Bull Case】

The company integrates specialty insurance, global reinsurance and mortgage into three platforms, giving more diversified earnings sources and more flexible capital allocation.
Book value and per-share value have grown strongly in recent years; BVPS rose from $26.42 in 2019 to $65.11 in 2025.
High ratings, non-aggressive leverage and strong capital; at the end of Q1 2026 it still maintained thick equity and a light interest-bearing-debt burden.
Management incentives are tied to TBVPS/TSR, executive ownership is not low, and long-run buybacks have significantly reduced the share count.
The current valuation isn't expensive relative to peers and the broad market, with TTM PE about 7.4x and PB about 1.45x.

【Core Bear Case】

2023-2025 profits may be above the long-run norm; a low static PE isn't necessarily truly cheap.
Insurance/reinsurance is a strongly cyclical industry; today's underwriting discipline can't fully eliminate tomorrow's price competition.
Long-tail liability reserves and catastrophe losses "come back to bite" years later, an eternal risk for insurance stocks.
Mortgage insurance drags on profit when housing and employment deteriorate; today's low loss ratio is not a permanent constant.
The current margin of safety is limited, not tempting enough for a conservative large-position buyer.

【Key Assumptions】

The company's normalized operating ROE can hold at least in the 12%-14% range over the next decade.
No systematic under-provisioning of reserves.
The mortgage business does not suffer a deep credit recession.
Management continues to keep buybacks opportunistic rather than mechanical.
Ratings stay high, with no downgrade that would damage its competitive position.

【Fair Buy Price】 $80-90. This range roughly corresponds to 1.2-1.35x current BVPS and is closer to the 20%-30% margin of safety I require; if you can only accept a "moderate margin of safety," then below $90-95 you can also begin scaling in in tranches.

【Target Holding Period】 At least 5-10 years, ideally more than 10.

【Expected Annualized Return】

Conservative scenario: about 6%. Assumes BVPS compounds at about 8% over ten years and the exit PB is compressed to 1.2x.
Neutral scenario: about 9%-10%. Assumes BVPS compounds at about 10% over ten years and the exit PB holds around 1.4x.
Optimistic scenario: about 13%. Assumes BVPS compounds at about 12% over ten years and the exit PB rises to 1.6x. These return estimates are my assumptions, not company guidance.

【Maximum Loss Risk】 In an extreme scenario where a "big-catastrophe year + reserve top-ups + market softening + mortgage credit deterioration" occur simultaneously, I think it isn't unimaginable for the stock to fall to $42-50, equivalent to a stress test at 0.8x-0.9x PB on impaired book value, implying a potential drawdown of about 48%-56% from the current price. What truly causes permanent loss isn't volatility itself but reserve error or a rating impairment.

【Tracking Metrics】

The combined ratio, especially the ex-catastrophe combined ratio.
The direction and magnitude of prior-year reserve development.
Operating ROE / operating ROAE.
The annualized growth of BVPS and TBVPS.
Buyback amounts, buyback pace, and the PB/valuation at the time.
Mortgage persistency, delinquency, and the quality of new insurance written.
Rating changes.
Investment-portfolio yield and credit-risk exposure.
Insurance/reinsurance market pricing trends, especially in the property and casualty categories.
Whether the share count changes and per-share value rises in tandem.

【Signals That Would Trigger Reassessment】

Clear adverse reserve development for two or more consecutive years.
The ex-catastrophe combined ratio deteriorating persistently toward or above 100%.
A downgrade of the high rating.
Mortgage loss ratios rising in tandem with unemployment/default pressure.
Management still buying back heavily at elevated valuations during a weakening phase of the cycle.
BVPS growth persistently below mid-single digits while the market still awards a high PB.
A "chasing scale" tendency in capital allocation in the new CEO era.

【Final Recommendation】 Soberly put, ACGL deserves a place on your long-term watchlist, and it's worth scaling in gradually in small tranches, provided you genuinely understand insurance accounting and the underwriting cycle; but if your style is "conservative among the conservative," I'd suggest waiting for a more obvious discount. It's a good company that will most likely keep creating value over the long run; it's just that today's price is closer to "a reasonable buy" than to "buy it in your sleep." For a company like this, what really determines the outcome isn't next quarter's share price but whether, over the next decade, it can keep its underwriting discipline, rated capital and per-share value growth at a high level.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.

Baillie Framework · Ten Questions for Growth Investing

Hunting ten-year five-baggers among great growth stocks — pressing the upside question: "Can it get much bigger?"

How high is its market ceiling? Is it growing a slice of an existing pie, or creating an entirely new market?5/10

Bottom line first: Arch is "growing an existing pie and taking share across the cycle" in a mature, stable-demand market, rather than creating an entirely new one. Its ceiling depends more on the slow expansion of the global risk-transfer pool and on its ability to take share from peers than on exponential new demand. That makes its growth character "high-quality compounding" rather than the "disruptive new-market breakout" Baillie Gifford most cherishes.

Property & casualty (P&C), reinsurance and mortgage insurance are all mature markets that have existed for over a hundred years. The report itself characterizes the industry as "mature, stable-demand, strongly cyclical," and says outright that "this is not an inherently wonderful industry," giving industry attractiveness only 3/5. This differs fundamentally from the "open a whole new category, ten-bagger-over-a-decade" targets that Baillie Gifford's LTGG hunts for; insurance demand grows slowly with GDP, asset values and liability-cost inflation, and a huge new pie won't materialize out of thin air.

By size, in 2025 Arch's full-year gross premiums written were about $22.9 billion (reinsurance about 48%, insurance about 46%, mortgage about 6%), with net premiums written of about $15.4 billion. It still holds only a single-digit percentage of the trillion-dollar-scale premium pool of global commercial insurance and reinsurance, so the theoretical "existing pie it could grow into" is indeed very large; but the pie itself grows at low single digits, and Arch's growth comes more from share migration under cycle management: scaling up in hard markets and disciplined exits in soft markets, taking quality risk from less-disciplined peers. The "cycle management philosophy" the report repeatedly stresses is exactly this playbook of "grabbing a bigger slice of an existing pie."

The only thing with a touch of "semi-new market" color is mortgage credit-risk transfer (CRT) and its self-styled positioning as the "only globally diversified insurer of mortgage credit risk"; this is a relatively new, capital-markets-based risk-sharing structure. But the report also concedes that the mortgage segment's 2025 net premiums were only about $1.3 billion, or 6% of the total, and that growth is "challenging because of constrained new-home supply and high mortgage rates"; the full-year mortgage underwriting profit of about $1.0 billion at a 14.6% combined ratio, though good, has a limited scale ceiling and depends heavily on the U.S. housing cycle, so it can't support a "whole new market" narrative.

In a word: the ceiling is a "wide but low-growth" existing market, and Arch's value lies in taking share from it more disciplinedly than peers and driving up book value per share, rather than riding a brand-new exponential demand curve. By Baillie Gifford's "create a new market" standard, this dimension isn't a standout.

Jun 10, 2026
Can its revenue at least double over the next five years? Is growth driven mainly by volume, price or new business?3/10

Bottom line first: revenue "doubling" over the next five years is almost unrealistic; it would require roughly 15% compound annual premium growth, whereas Arch is currently in a phase of softening underwriting cycle with premiums falling rather than rising year over year. Its growth engine is "price (rates) + share across the cycle" rather than "rapid volume expansion," and the near-term direction is negative. This dimension clearly falls short of Baillie Gifford's "revenue doubling in five years" bar.

First the hard data. In 2025 gross premiums written were about $22.9 billion and net premiums written about $15.4 billion. And Q1 2026 already shows contraction signals: per the company's Q1 2026 results, insurance-segment net premiums written fell about 1.4% year over year and reinsurance-segment net premiums written fell about 6% year over year; the reinsurance decline was driven mainly by a proactive reduction in property-catastrophe business at the January 1 renewals, plus reinstatement premiums in the prior-year period tied to the California wildfires. In other words, in a soft market the company is proactively cutting volume and holding discipline, rather than forcing scale to "grow revenue." This is exactly the discipline the report praises, but it directly conflicts with the goal of "doubling revenue in five years."

To double in five years, net premiums would need to rise from about $15.4 billion to about $30.8 billion, implying roughly 15% compound annual growth. In a mature industry with low-single-digit organic growth, that could appear only briefly amid a sustained hard market plus large-scale M&A; yet the report explicitly notes the company's own statement that "competition had increased entering 2026," and Chubb has publicly acknowledged that parts of property and financial lines are already "soft or softening." Chasing double-digit premium growth in a softening cycle usually means loosening pricing discipline, which is precisely the "loss of underwriting discipline" the report lists as a sell signal.

Is growth driven by volume, price or new business? All three are limited: volume: organic volume growth in a mature market is low single digits, and the company is currently cutting volume proactively; price: rising rates in the hard market were the main driver of the 2023-2025 high growth, but rates have now peaked and are rolling over; new business: mortgage and CRT have structural appeal, but 2025 mortgage net premiums were only about $1.3 billion, and with high mortgage rates and constrained new-home supply making "growth challenging", they can't support an overall doubling.

Worth noting: what Arch has truly "doubled" beautifully is book value per share (from $26.42 in 2019 to $65.11 in 2025, about 16%/year over six years), achieved through retained underwriting profit + buyback-driven share shrinkage, rather than doubling revenue. This is its essence as a "compounding machine"; but if we ask strictly by Baillie Gifford's "revenue doubling in five years" standard, the answer is: it can't, and its current revenue direction is downward.

Jun 10, 2026
Five years out, what will take over as the next growth engine? Does this "second curve" exist today?4/10

Bottom line first: Arch has no "second curve to take over" in the typical sense; its growth structure is "three mature platforms rotating and hedging one another," shifting capital to whichever segment is in a favorable cycle, rather than relying on a new growth pole that is visible today and will erupt in the future. By Baillie Gifford's standard of "does the second curve exist today," the answer is "there's no clear second curve, only capital reallocation across platforms."

The report lays out Arch's structure clearly: specialty insurance, global reinsurance and mortgage coexist as three platforms, "hedging catastrophe volatility with mortgage profits in some years and capturing high-payoff opportunities in a hard reinsurance market in others." This is Arch's real distinction and the key to crossing cycles; but note, this is three already-mature curves propping each other up, not "a brand-new curve taking over after the main business peaks." The "second curve" Baillie Gifford asks about points to a disruptive new growth engine (a new category, a new technology platform), whereas the rotation of three insurance platforms is essentially the same business of "taking on risk and investing float" shuffled among different lines.

Which segment most resembles a "potential second curve"? Mortgage and mortgage credit-risk transfer (CRT). The report cites the company's 2022 annual report calling itself the "only globally diversified insurer of mortgage credit risk," and this business's capital-markets-based sharing structure does have an innovative tint. But it can't shoulder the takeover role: in 2025 the mortgage segment's underwriting profit was about $1.0 billion, net premiums only about $1.3 billion, and about 6% of total premiums, and the company explicitly says growth in this business is pressured by "constrained new-home supply and high mortgage rates," making it more of a high-margin but low-growth "cash-cow hedge block" than a fast-scaling new engine. The report also flags that mortgage profit depends heavily on low loss ratios, and "the current low loss ratio is not a permanent constant."

So what takes over in five years? The most realistic picture is: there is no "successor"; the same compounding machine keeps running: underwriting profit + float investment income (higher rates in 2025 lifted investment income) + opportunistic buyback share shrinkage, steadily driving up book value per share. This is exactly the source of the expected annualized return the report gives (neutral scenario about 9%-10%, from BVPS compounding at about 10%/year + maintaining 1.4x PB), rather than the scaling of some new business line.

Frankly: taking the "second curve" as a bonus factor, Arch isn't a standout on this dimension. Its sense of safety comes from the diversification and cycle resistance of three mature platforms, rather than from a new growth pole planted today that erupts tomorrow. For investors chasing "sustained rapid growth" that's a minus; for those seeking "high-quality compounding + low permanent-loss risk" it isn't necessarily a flaw; but it does not fit the kind of "second-curve imagination" Baillie Gifford looks for.

Jun 10, 2026
What is its core competitive advantage? Will this moat widen or narrow over the next three to five years?6/10

Bottom line first: Arch's core competitive advantage is the set of moats that are real but not extremely deep in insurance: "high-rated capital + a disciplined cycle-management underwriting culture + three-platform diversification"; over the next three to five years it will most likely be "broadly stable, stretching and compressing locally with the cycle," rather than steadily widening. The report scores the moat 4/5, and I agree with this above-average characterization.

The moat's three-layer structure is clear. The first layer is ratings and regulatory qualification: the report notes that the main insurance subsidiaries carry high ratings such as S&P AA-, Moody's A1, Fitch AA-, and A.M. Best A+. These are the "entry ticket" to underwriting large, complex risks, retaining brokers and corporate clients, and participating in reinsurance and capital-market transactions, and they "can be lost overnight but are very hard to rebuild quickly." They are backed by thick capital: at the end of Q1 2026 the company had about $26.9 billion of capital, total shareholders' equity of about $23.4 billion, and a light interest-bearing-debt burden.

The second layer is underwriting culture and cycle discipline, the hardest to replicate and the most valuable layer. The report stresses in several places Arch's "cycle management philosophy": disciplined exits in soft markets, scaling up underwriting in hard markets, and dynamically allocating capital across businesses and geographies. This layer of moat "shows up not in patents but in fewer mistakes, disciplined pricing, and decisive exits." A real-world proof point: in Q1 2026 the company proactively reduced its property-catastrophe reinsurance exposure, with net premiums written down about 6% year over year, preferring to cut volume rather than sell underwriting capacity cheaply in a soft market; this is exactly the expression of discipline, and it also compressed the reinsurance segment's Q1 combined ratio to 75.9%.

The third layer is three-platform diversification: insurance, reinsurance and mortgage coexist, giving it an earnings profile and capital-allocation flexibility that other pure-P&C peers lack.

Will it widen or narrow over the next three to five years? My judgment is broadly stable, "appearing" to stretch and compress with the cycle. The report puts it precisely: in a hard market the underwriting culture and capital strength look wider, while in a soft market price competition intensifies and the moat "looks narrower." Right now happens to be a softening phase (the report cites Chubb publicly acknowledging that parts of property and financial lines are "soft or softening" and the company's own statement that competition increased in 2026), so over the next three to five years the moat is more likely to be "pressed down by the cycle, appearing to narrow" than to widen on its own.

Its pricing power is limited and peculiar: the report notes this pricing ability "comes not from brand but from the discipline to exit business that isn't profitable enough." Short-tail lines and newly written business can be repriced fairly quickly, while long-tail liability lines are more dangerous.

Against Baillie Gifford's standard: this is a real moat built on capital and years of underwriting/claims/channel accumulation, one that competitors "can't easily replicate in two or three years even by throwing money at it," but it isn't the "naturally widening, stronger-with-use" deep moat of a consumer brand or a platform network effect. The characterization that fits best is "a stable, above-average moat, where not narrowing over the next three to five years already counts as good"; this dimension is one of Arch's relatively solid strengths, but it's far from extremely deep, let alone steadily widening.

Jun 10, 2026
If its core business were disrupted, does it have the genes to reinvent itself? How does it treat mistakes and bad news?5/10

Bottom line first: Arch's "genes for self-reinvention" lie in a plainer and scarcer capability, rather than in disruptive transformation: disciplined acknowledgment of mistakes, decisive exits, and booking bad news as quickly as possible; its attitude toward mistakes and bad news is an important basis for the report's 4/5 management score. On the Baillie-Gifford-style hypothesis of "the core business being disrupted," the form of "disruption" for the insurance business differs from that for tech stocks, and Arch's response genes are broadly credible but still need continued verification.

First, the implicit premise of "the core business being disrupted." Insurance is unlikely to be disrupted overnight by a single technology; the "disruption" it faces is more like a line of business running at a chronic loss (such as runaway inflation in long-tail liability), catastrophe models failing, or a business line structurally deteriorating. Arch's genes for responding to such shocks are embodied precisely in its cycle-management philosophy: the report stresses that it will "hold discipline in soft markets, scale up underwriting in hard markets, and dynamically allocate capital across businesses and geographies," and "exit business that isn't profitable enough." In other words, its "reinvention" isn't tearing everything down and starting over, but continuously pulling capital out of deteriorating areas and moving it to where risk-adjusted returns are highest. A real-time piece of evidence: in Q1 2026 it proactively shrank property-catastrophe reinsurance, with net premiums written down about 6% year over year, preferring to cut volume rather than cling to a fight in a soft market.

Its attitude toward mistakes and bad news is the most crucial part of this question and Arch's most impressive trait. The report explicitly cites the company's 2023 shareholder letter, which says that on reserves it would rather "recognize bad news quickly and good news slowly." This is a conservative, honest reserving culture; for an insurer, the biggest "invisible landmine" is reserve under-provisioning that only surfaces years later, rather than fraud, and "recognizing bad news quickly" is precisely the core discipline for hedging such landmines. On this basis the report also judges that there is "no direct sign of obvious financial fraud or aggressive accounting," and credits the company for discussing "the cycle, competition, claims volatility and liquidity constraints fairly candidly" in its shareholder letters.

But to be honest about one blemish: the report notes that during the 2024 leadership transition the company granted one-time outperformance equity awards, which pushed the 2025 say-on-pay support rate down to 84.7%. This shows governance is "good but not perfect," and the discipline during new CEO Nicolas Papadopoulo's tenure still needs watching.

Against Baillie Gifford's standard: insurance's "self-reinvention" looks more like "continuous error-correction and capital reallocation" than disruptive transformation, and on this point Arch has real, shareholder-letter-corroborated discipline genes (especially honesty about bad news), but it isn't the tech-style reinvention of "jumping to a whole new track after the core is disrupted." For a mature insurer this dimension is already excellent and highly credible, but the new management still needs several cycles to keep proving it.

Jun 10, 2026
Does management (especially the founder) have a long-term horizon, with interests deeply tied to the company? Are they willing to sacrifice current profit for five to ten years out?5/10

Bottom line first: Arch's management has a clear long-term vision, with incentives fairly well tied to shareholder interests (executives hold about 3.3% in aggregate, and pay is heavily linked to three-year TBVPS growth), and it is willing to sacrifice short-term scale for long-term per-share value; but it is not a founder-controlled company, and the depth of alignment sits between an "excellent professional-manager sample" and a "Berkshire-style owner." The report scores management and capital allocation 4/5, and I agree.

First the hard data on alignment. The report notes that as of March 9, 2026: CEO Nicolas Papadopoulo held about 1.6599 million shares, CFO François Morin about 738,200 shares, and Maamoun Rajeh about 894,700 shares; all directors and officers together held about 11.6187 million shares, roughly 3.3% of common stock. At the current share price of about $90.41, the CEO's personal stake is worth about $150 million; the report's assessment is fair: "This isn't a Berkshire-style controlling stake, but it's already enough to make executives genuinely feel the pain." It should be noted that Arch has no deeply aligned sitting founder; it is a classic high-quality professional-management structure, which falls short of the "founder at the long-term helm" that Baillie Gifford prefers.

Next, whether the pay design truly orients toward the long term. The report cites the 2026 proxy: 77% of the CEO's target pay is performance-linked and 68% is long-term incentives; long-term performance shares are anchored to three-year TBVPS (tangible book value per share) growth, with a TSR modifier; all NEOs meet the stock-ownership requirements, and executives must retain 50% of net shares acquired until they hit the threshold. Anchoring the core metric to "three-year book-value-per-share growth," rather than short-term EPS or premium scale, is a good institutional arrangement for "sacrificing the short term for the long term."

"Are they willing to sacrifice current profit for five to ten years out?" The behavioral evidence is affirmative. The strongest point is proactively cutting volume in a soft market: the cycle-management philosophy the report stresses requires "holding discipline in soft markets and exiting business that isn't profitable enough," and in Q1 2026 the company's net premiums written fell in both insurance (about -1.4%) and reinsurance (about -6%), preferring to sacrifice current-period premium income rather than loosen pricing discipline. This is exactly the proof of "sacrificing short-term scale for long-term underwriting profit."

Capital allocation likewise reflects a long-term orientation, but one point needs continued watching. The positives: long-run opportunistic buybacks; the report notes that since the program began, cumulative buybacks are about 455 million shares for roughly $7.8 billion, with about $1.9 billion repurchased in 2025 alone, and about $783 million more in Q1 2026, bringing the weighted diluted share count down to about 359.7 million, a genuine boost to per-share intrinsic value. The caution to keep: the report notes the one-time incentive in the 2024 leadership transition dragged say-on-pay down to 84.7%, and that "big buybacks in good years aren't automatically smart"; whether the new CEO "still buys back heavily when valuations are elevated" still needs verification (the report even lists this as a sell signal).

Against Baillie Gifford's standard: management quality, long-term orientation and incentive alignment are all real strengths for Arch (4/5), but it lacks the "deep founder alignment" that Baillie Gifford values most; it is an excellent professional-manager sample, rather than owner-operated, family-style long-termism. The alignment is enough to "hurt," and the system orients toward the long term, but in depth it doesn't match the founder-led companies Baillie Gifford most cherishes.

Jun 10, 2026
If it disappeared tomorrow, how much would customers miss it? Is its growth model sustainable, without relying on harming society and regulation?5/10

Bottom line first: if Arch disappeared tomorrow, its corporate clients, brokers and lenders would "moderately miss it"; insurance is a necessity but substitutable, and Arch is not an indispensable sole supplier. But its growth model is highly sustainable and does not rely on harming society or regulation (insurance is by nature positive infrastructure for society's risk-sharing). On this question, "indispensability" is moderate while "social and regulatory sustainability" is fairly strong.

First, one implicit premise: indispensability. Insurance and reinsurance are highly competitive markets, a point the report stresses repeatedly: clients face many comparably rated underwriters, and if a single company exits, the risk is picked up by peers. So from the angle of "can clients not do without you," Arch's answer is "they'd miss it but can substitute"; its value lies in being one of the few qualified underwriters with AA-/A1/A+ high ratings that can reliably take on large, complex risks, rather than in monopolistic irreplaceability (the report notes S&P AA-, Moody's A1, Fitch AA-, A.M. Best A+). For large enterprises and cedants that need highly rated counterparties, qualified suppliers are already few, which gives it a degree of "being needed," but it is far from "clients being stranded if it disappeared."

The mortgage segment's indispensability is a bit higher: the report cites the company's self-description as the "only globally diversified insurer of mortgage credit risk," and the 2025 U.S. primary portfolio persistency rate reached 81.8% (the report also carries this figure); a high persistency rate means client stickiness isn't low; risk-in-force was about $87.9 billion with a default rate of only about 2.2%. But this segment likewise faces peer competition from MGIC, Essent, Radian and others, and isn't exclusive.

Second implicit premise: whether the growth model is sustainable and whether it harms society and regulation. On this, Arch is quite clean, even leaning positive. Insurance's social function is to transfer and smooth the tail risks of companies and individuals; it is positive infrastructure for the functioning of the economy, with no issue of "harming society in exchange for growth." The report describes its growth as coming from "disciplined underwriting + high-quality float investing + opportunistic buybacks + multi-platform diversification," with no model relying on regulatory arbitrage or harming clients. On the contrary, it depends deeply on the recognition of the regulatory and rating systems: high ratings and regulatory capital adequacy are themselves its moat and its vulnerability, and the report lists "a rating downgrade or clear pressure on regulatory capital" as a sell signal; this means its business model is "symbiotic" with regulation rather than "adversarial," and its sustainability is strong.

The only sustainability risk worth honestly flagging lies in the cycle and reserves, rather than at the "social/regulatory" level: the report warns of long-tail liability reserve under-provisioning, big-catastrophe years, and rising mortgage loss ratios when housing/employment deteriorate; these are risks to "profit sustainability," but not risks of "an immoral/non-compliant growth model."

Against Baillie Gifford's standard: Arch's growth model is sustainable, symbiotic with regulation and socially positive (this side is fairly strong), but "how much customers would miss it / indispensability" is only moderate; it is a high-quality, qualified but substitutable supplier, rather than a critical node whose "disappearance would collapse the whole ecosystem." On balance it is above-average, with its strength in "sustainable, not harming society or regulation" and its weakness in "substitutability."

Jun 10, 2026
What are the unit economics of this business (gross margin, incremental returns)? Does it get better or worse at scale? Where does the money it earns go?6/10

Bottom line first: insurance has no "gross margin" in the manufacturing sense; what to look at is the combined ratio, operating ROE and float quality. Arch's unit economics are excellent within the industry (all three platforms have combined ratios below 100%, and operating ROE is double-digit); it is asset-light and broadly improves at scale; the money it earns goes mainly into buyback-driven share shrinkage, a genuine boost to per-share value. This is one of Arch's truly solid strengths.

First correct the measure. The report puts it plainly: "For an insurer, what really matters is not gross margin but the combined ratio, reserve adequacy, investment yield and return on capital." A combined ratio below 100% means underwriting itself makes money (equivalent to "positive underwriting gross profit"), and layering on float investment income gives insurance its two-engine profitability.

The core metrics of unit economics:

Underwriting profit (combined ratio) is positive across all three platforms. In 2025 the insurance-segment combined ratio was 95.2% and the reinsurance segment 80.8%, with the mortgage segment at just 14.6% (mortgage insurance has structurally high margins); in Q1 2026 the insurance segment was 96.5% and the reinsurance segment 75.9%, a marked improvement after the high base of the California wildfires. All three beat the 100% breakeven line, which is quite excellent underwriting quality for the industry.

Return on capital (this is insurance's true "incremental return") is excellent. In 2025 the company's annualized net ROE was 20.1%, operating ROE 17.1%, and full-year after-tax operating income about $3.7 billion (a record high); in Q1 2026 annualized net ROE was 17.8% and operating ROE 15.4%. The report's neutral assumption is a normalized operating ROE holding at 12%-14%, meaning that even if the cycle retreats, the return on unit capital remains healthy.

Asset-light, with margins improving as scale grows. The report's historical table shows extremely light fixed-asset capex (only about $40-50 million a year), a natural advantage of a financial company: growing premium scale requires almost no proportional heavy-asset investment, and the diversified platform can also spread fixed costs and smooth volatility. Growing larger is broadly "better" for Arch (stronger diversification, greater rating leverage, thicker capital to seize hard markets), rather than "worse."

But to honestly flag one "hidden risk at scale": insurance's unit economics are strongly cyclical. The report repeatedly warns whether the high 2023-2025 ROE "embeds a strong cyclical dividend," and that underwriting-profit elasticity will drop when a soft market arrives; and the larger the scale, the greater the cumulative-error risk in long-tail liability reserves, which is the core interrogation of "whether the cash profit is real." The report therefore treats owner earnings conservatively: after deducting from operating profit the capital that must be retained for rating capital and reinsurance volatility, conservative owner earnings are about $2.2-2.6 billion (materially below reported CFO), corresponding to about 13-15x, showing that an insurer's "real cash distributable to shareholders" is far smaller than reported operating cash flow.

Where does the earned money go? The report gives a clear answer: mainly opportunistic buybacks. Since the buyback program began, cumulative repurchases are about $7.8 billion and about 455 million shares, with about $1.9 billion repurchased in 2025 alone, and about $783 million more in Q1 2026, bringing the weighted diluted share count down to about 359.7 million. Buying back at around 1.37x PB is a mild boost to book value per share; the rest of the capital is kept as "ammunition" for solvency and hard-market underwriting.

Against Baillie Gifford's standard: unit economics (underwriting profit + high ROE + asset-light) are Arch's most solid strength, broadly improving at scale with rational capital allocation; the only reservation is that these unit economics carry strong cyclicality, and the truly distributable cash is far below the reported figures. On this dimension Arch is excellent, and it is the core support for the "good business" judgment.

Jun 10, 2026
For it to 5x in ten years, what conditions must hold simultaneously? Are they realistic? What expectations does today's share price imply?3/10

Bottom line first: for Arch to 5x in ten years (about 17.5%/year), three conditions must hold simultaneously: "a sustained high long-term operating ROE + no big reserve hole + significant multiple expansion," which conflicts with its profile of strong cyclicality, a moderate moat and an already-reasonable current valuation. By Baillie Gifford's "5x in ten years" standard, this is almost unrealistic. Today's roughly $90 share price implies "reasonable-but-slightly-cheap, steady compounding," not a "5x-in-ten-years breakout expectation." This question is Arch's biggest weakness as a growth stock.

First break "5x in ten years" into conditions that must all hold:

Condition one: per-share value must compound fast enough. The report's most optimistic scenario assumes BVPS compounds at about 12% over ten years and the exit PB rises to 1.6x, corresponding to an annualized return of only about 13%, about 3.4x over ten years, short of 5x. For 5x, BVPS would need to compound at 14%-16% for years on end (close to the roughly 16% historical boom of 2019-2025), and that growth rate benefited precisely from the triple tailwind of a hard reinsurance market, high-rate investment income and low mortgage loss ratios. The report repeatedly warns whether this "embeds a strong cyclical dividend" that is hard to sustain when a soft market arrives.

Condition two: no permanent capital loss. The report defines the biggest risks as long-tail liability reserve under-provisioning and big-catastrophe years, which "come back to bite years later." Just one big adverse reserve development or big-catastrophe year in ten years would break the compounding curve.

Condition three: significant multiple expansion. Currently BVPS is $66.19, the share price about $90, and PB about 1.37x. For valuation to contribute extra upside, PB would have to expand from 1.37x to above 2x (close to WRB's 2.62x quality-premium level in the report). But insurance-stock PB is suppressed by the cycle, and the market is more likely to compress it from 1.37x back to 1.1x-1.2x in a soft market (the report's bear logic), rather than voluntarily award it a consumer-stock-style high premium.

Multiplying the three conditions together (sustained high ROE + zero major reserve incident + big multiple expansion), the probability that they all hold is very low, and they even contradict one another (a sustained boom usually means a cycle top, where the valuation is harder to expand further). So a Baillie-Gifford-sense "5x in ten years" is unrealistic for Arch.

So what expectations does today's roughly $90 price imply? It implies precisely "reasonable-but-slightly-cheap steady compounding," not a breakout. Cross-checking with the report's own three methods: owner-earnings discounting gives conservative about $81, neutral about $119 and optimistic about $172 per share, and the current price of about $90 lands "above the conservative value and below the neutral value," which the report characterizes as "reasonable-but-slightly-cheap, with an insufficient margin of safety." On multiples, PB is about 1.37x and TTM PE only about 7x (the report's snapshot computes 7.4x on $95.87; at the current price it is lower); the PE is astonishingly low, but the report precisely points out this is "closer to the illusion of peak-cycle profit" than truly cheap. In other words, the market has already priced in "a high-quality insurer with double-digit ROE," but it hasn't, and shouldn't, award it a "5x-in-ten-years" growth premium.

Against Baillie Gifford's standard: the conditions for 5x in ten years are both unrealistic and self-contradictory, and today's price implies a "steady-compounding" expectation (neutral annualized about 9%-10%) rather than rapid growth. As a value-compounding target it qualifies; as a Baillie-Gifford-style "5x-in-ten-years growth stock" it does not, and this dimension is clearly on the weak side.

Jun 10, 2026
Why hasn't the market realized all this yet? Is it that it can't understand, looks down on it, or can't see far enough? What would become the "narrative inflection point"?3/10
A
Bottom line first: the market has actually "understood" Arch and hasn't badly mispriced it; the low PE is not a perception gap but the market's reasonable pricing of "strong cyclicality + profit possibly at a peak." The real disagreement in perception is "how long the current double-digit ROE can hold," on which bulls and bears bet differently. The so-called "narrative inflection point" is not the market suddenly discovering it, but whether, after a complete soft cycle, it can prove that its underwriting discipline and per-share value growth cross the cycle. The honest answer to this question is: Arch is not a "buried growth stock" but a "reasonably priced cyclical high-quality stock."

First dispel the illusion that "the market doesn't understand." The report's TTM PE of a bit over 7x is highly misleading, looking at first glance like deep undervaluation. But the report repeatedly makes clear this is "the illusion of peak-cycle profit." The market hasn't failed to see this low PE; rather it deliberately declines to award a high valuation, because the high 2023-2025 profits benefited simultaneously from a hard reinsurance market, high-rate investment income, low mortgage loss ratios and buyback amplification, and the market worries these tailwinds will reverse. So PB is the better lens: current BVPS is $66.19, the share price about $90, and PB about 1.37x, which the report judges is actually "fair-to-slightly-cheap" rather than "a fire sale."

So is the market "not understanding, looking down, or not seeing far"? None of the three is the main reason; the truth is "it sees clearly, but discounts":
- Not a failure to understand: institutional investors are very familiar with the logic of the insurance cycle, reserves and float, and the report's peer comparison (RNR, WRB, CB) shows the market has mature differentiated pricing for insurance stocks: awarding the stable-quality WRB as much as 2.62x PB, the more cyclical RNR only 1.21x PB, and Arch's 1.37x sits right in the middle, so the pricing is rational.
- A slight element of "looking down": the market naturally discounts companies that are strongly cyclical and whose profit contains a cyclical dividend, and is unwilling to award a high multiple at a cycle top; this is reasonable risk pricing rather than prejudice.
- Not a case of "not seeing far" and being mispriced: the report's own three valuation methods all place the current price "above the conservative value and below the neutral value," with the current price of about $90 corresponding to a conservative intrinsic value of about $81 and neutral about $119, a limited margin of safety. This shows the market's pricing closely matches the fundamental analysis, with no significant, arbitrageable perception gap.
The real disagreement (and the short thesis) is: how long the current ROE can hold. Bulls believe the diversified platform + cycle discipline can keep normalized operating ROE in double digits for the long run (the report's neutral assumption is 12%-14%); bears believe that once a soft market arrives, the premium decline already visible in Q1 2026 (insurance about -1.4%, reinsurance about -6%) will evolve into a "double whammy" of ROE retreating + PB compressing from 1.37x back to 1.1x-1.2x. The report considers this short thesis "not absurd, the one most deserving of serious attention today."

What would become the "narrative inflection point"? (implicit premise) The inflection isn't "the market suddenly discovering this good company," but the moment the cycle gives an answer:
- Positive inflection: after a complete soft market, if ex-catastrophe underwriting profit stays solid, reserves keep developing favorably, and BVPS still compounds at double digits, this would prove its high ROE is structural rather than a cyclical dividend, and the market may be willing to award a higher PB.
- Negative inflection (listed by the report as sell signals): clear adverse reserve development for two or more consecutive years, the ex-catastrophe combined ratio approaching or exceeding 100%, a downgrade of the high rating, or the new CEO still buying back heavily at elevated valuations; if any appears, the "high-quality compounding" narrative would be replaced by the "peak-cycle illusion" narrative.
Against Baillie Gifford's "why hasn't the market realized" standard: Arch doesn't fit the profile of a "mispriced, undervalued growth stock"; the market has priced it fully and rationally, the low PE is a cyclical discount rather than a perception gap, and the direction is even neutral rather than positively undervalued. This dimension does not constitute a perception edge for buying; its return is more likely to come from the realization of fundamental compounding, rather than from a market "light-bulb" revaluation.
Jun 10, 2026

Ask about this report

Members can ask about this report; once answered it appears under "Reader Q&A" on this page. You can also highlight a passage in the text to ask about it directly.