They’ve compounded book value at 15% for 23 years, just bought back 5.6% of the company in one year, and the market still prices it at 1.5x

Arch Capital (ACGL). Specialty P&C insurer, plus reinsurance, plus mortgage insurance. Roughly $34B market cap. I've been digging into it for the past few weeks and figured I'd write up where I land.

the setup

Stock is at $97. Tangible book is $61.71 per share. So you're paying 1.56x tangible book today.

Their five-year average ROE is 20.1%. The current quarter is 19.5%. A rough rule for insurers is that justified P/TB is ROE divided by your cost of equity. At 10% cost of equity that's 2.01x. So the gap between what the math says and what the market is paying is real, not enormous, but real.

Plug it in: $61.71 x 2.01 = $124. Versus $97. About 22% upside before any growth in book value.

the thing I keep coming back to

Float of $24B. Investment portfolio of $47B. Five-year average combined ratio of 88%. Latest quarter 86%.

Translate that. They are being paid roughly 12 cents on every dollar to hold $24B of other people's money. They invest that money and generated $1.62B of investment income last year, on top of the underwriting profit. That is the Berkshire model in miniature, and it has compounded book value per share at over 15% for 23 years.

why I think it's mispriced

Two things spooked the market in 2024 and 2025. The hurricane season was rough. And property catastrophe renewals in Q4 2025 came in soft, rates down 10 to 20%.

Most insurers respond to soft pricing by writing more volume to keep the top line growing. Arch did the opposite. They shrank net premiums written by 4% year-over-year. Management has been pretty open that they would rather underwrite less business at the right price than chase volume at the wrong one.

You can argue with that strategy, but it's the opposite of the behavior that destroys insurers.

what management is actually doing

In the trailing twelve months they bought back $1.89B of stock. That's 5.6% of the float at current prices. Then on April 19 of this year the board authorized another $3B on top of that. So they have roughly $3.1B loaded in the chamber, against a market cap of $34B.

This is a 19.5% ROE business buying its own stock at roughly 8.4x earnings. I don't know what better signal you want from a management team.

what I'm worried about

Regulatory capital is the one I take seriously. If global regulators decide insurers need to hold more capital against the same book of premium, ROE comes down mechanically. There is no clever way around it. This is the risk I think actually deserves the multiple discount.

The other one is catastrophe risk. Their own modeled 1-in-250-year peak zone loss is $1.9B, or about 8.2% of tangible equity. That is survivable, but it's a real number, and 1-in-250 events have been showing up faster than once every 250 years lately.

I'm less worried about insurtech eating their lunch. Arch writes specialty casualty, complex reinsurance treaties, and mortgage insurance. You don't bind a multimillion dollar corporate liability tower through a smartphone app.

where I come out

You're paying 1.56x tangible book for a 20% ROE business with a disciplined management team that is currently buying back its own stock with both hands. The intrinsic value math says low-120s, the current price says 97, and book value should keep compounding around 15% per year while you wait.

I could be wrong about how much regulatory pressure shows up over the next few years, and a really bad cat year could absolutely chew up a year of earnings. But I don't see a scenario where the underlying business is structurally broken.

I own a position. Not advice, do your own work.