Actionable insights — What I learned from the Berkshire Meeting

1. Value an insurer on whether its float is free, not on its investment returns

1. Establish the size of the float — premiums collected but not yet paid out — and its growth over a decade.
2. Check the combined ratio. Below 100% the underwriting itself makes money, and the float costs less than nothing; above 100% it is borrowing with an interest rate.
3. Only then look at the investment returns earned on the float, and treat those returns as leveraged returns on shareholders' capital.
4. Test the two together: excellent investing on expensive float is a worse business than mediocre investing on free float.

• Reserve releases flattering the combined ratio. A ratio below 100% that depends on releasing prior-year reserves is not underwriting skill.
• Float that shrinks. Growing float is the growth engine; a run-off book turns the compounding off.

2. Treat free float as leverage, and stress it in both directions

1. Recognise what the structure is: an investment portfolio funded partly by capital you do not pay for.
2. Compute the ratio of investments to shareholders' equity — that is the multiplier on whatever the portfolio earns.
3. Apply the same multiplier to a bad year, and to a bad underwriting year happening in the same year.
4. Check the asset mix against the liability duration: a claims book that can be called on quickly should not be funded by illiquid holdings.

• Catastrophe exposure concentrated geographically. The float is only free until it isn't, and it stops being free all at once.
• "By definition" arguments generally. The definition holds for the return on equity; it says nothing about the price you paid for that equity.

3. Use your benchmark's own concentration as the case for leaving it

1. Measure the top-ten weight and the largest sector weight in the index you would otherwise buy.
2. Ask what single driver those weights share — not what sectors they are labelled with.
3. Look for a substitute that gives broad exposure without that driver, and check what it holds instead.
4. Price the substitute against the index rather than in isolation, since the decision is relative.

• Swapping one concentration for another. The substitute here is roughly a third cash and heavily insurance — a very different risk, not a smaller one.
• A single-company substitute for an index. You have added key-person, regulatory and underwriting risk that the index does not carry.

4. Price a large cash balance as optionality, not as a drag

1. Establish the cash as a share of total assets and how long it has been held.
2. Ask whether it is deliberate (no attractive prices) or structural (no ability to deploy at scale).
3. Check the deployment record: has this management historically spent large cash balances well, and at what points in the cycle?
4. Value it as the ability to buy when others are forced to sell, and discount that value if the holder has never actually done so.

• Cash held for so long that the opportunity cost exceeds the option value. A third of a portfolio in cash through a rising market is expensive.
• Scale constraints. At Berkshire's size very few purchases are large enough to matter, which is the law-of-large-numbers concession the post itself makes.

5. Test a holding company's decentralisation claim against its structure, not its rhetoric

1. Ask who is accountable for each subsidiary's results, and whether that person's compensation is tied to them.
2. Check management tenure across subsidiaries — long retention is the evidence that autonomy is real.
3. Confirm the legal separation allows partial disposals without loss of control, which is what converts autonomy into financial flexibility.
4. Look for the failure mode: decentralisation without capital discipline at the centre is just a conglomerate.

• Autonomy that is real for operations but not for capital. The centre must still allocate, or the structure compounds mediocrity.
• Selling "small parts" repeatedly. Serial partial disposals can be a funding tell rather than a flexibility feature.

6. In insurance, look for the segment where the insurer sets its own price

1. Distinguish rate-regulated lines from ones where the insurer prices freely — the second is where underwriting skill can be paid for.
2. Within those, prefer specialist niches where the barrier is expertise and data rather than capital.
3. Verify with the combined ratio relative to peers over a full cycle, not one year.
4. Check founder or management ownership, because a specialist underwriter's discipline is a culture question.

• Soft markets. E&S pricing is cyclical, and discipline is only visible when competitors are cutting rates.
• Growth outrunning the underwriting. Rapid premium growth in a specialist book is the classic precursor to reserve trouble.

7. Treat a historical analogue as a hypothesis, and go looking for its disconfirmation

1. When a current drawdown is matched to a historical one, list what is the same and what is different before accepting the conclusion.
2. Ask how many comparable episodes exist. One prior instance is an anecdote, not a base rate.
3. Separate the mechanism from the pattern — is there a reason the reversal happened, or only the fact that it did?
4. Note what the analogy would look like if it failed, and watch for that instead.

• The succession explanation being dismissed too quickly. "This has nothing to do with Greg Abel" is asserted, not argued.
• Unsourced comparative figures. The 40% number carries no date range or index reference.