Actionable insights — Buying The Next Berkshire Hathaway

1. Buy only from a published shortlist, and let the lag between listing and purchase be visible

1. Maintain a small, ranked shortlist of names you would own, separate from what you already own.
2. Refresh it on a fixed cadence, and require any new purchase to have been on it first.
3. When you buy, say when the name was listed and at what rank. The lag is the evidence that the decision was made in advance rather than reactively.
4. Review the names that were listed and never bought — that residue is where the discipline actually shows.

• A shortlist long enough that something on it is always eligible — the discipline comes from the list being short.
• The remaining top-three name: the hint implied two purchases from three, so one more is outstanding and its absence is information too.

2. State the length of record at which you will treat a track record as skill

1. Decide, in advance and in general, how long a record has to be before you will underwrite it as skill rather than luck.
2. Apply it symmetrically — to managers you like and to those you do not.
3. Check that the record was produced by the same person, the same strategy and the same capital base you are buying into now.
4. Separate the record from the forecast: a long record supports the claim that the process works, not the claim that a specific growth rate will be achieved.

• Key-person risk that grows with exactly the evidence you are relying on: a forty-year record means a founder forty years older.
• Base sizes: a strategy that compounded at 19.5% on a small book faces a different problem at CA$47bn — the "smaller Berkshire" framing implicitly acknowledges this and does not resolve it.

3. When underwriting by analogy, name the specific mechanism that transfers

1. Identify the structural mechanism the comparison rests on — here, insurance float invested in securities at low cost.
2. Verify it operates the same way in the candidate: is the underwriting profitable, so the float is genuinely free?
3. Name the differences explicitly — size, geography, capital structure, regulatory regime, succession.
4. Reject the analogy if what transfers is only the biography. "Run by a great investor" is not a mechanism.
5. State what the analogy predicts, so it can be checked later.

• The combined ratio, which is the test of whether the float is free. It was raised in the July write-up and is not revisited here.
• Analogies that survive because they are never made specific enough to fail — "the next Berkshire" has been claimed for many companies.

4. Read a limit price as a statement of intent, and set yours deliberately

1. Decide before placing an order which you are optimising for: certainty of execution, or price.
2. A limit above the market buys certainty and admits you do not think the price matters at this level. A limit below it demands a discount and accepts non-execution.
3. Check liquidity before deciding — a small order in a liquid name makes the question moot, and pretending otherwise is theatre.
4. Publish or record which you chose and why, so the decision can be reviewed against the fill.

• Published limits that never fill and are quietly forgotten — the opposite failure, and one this archive has already produced once.
• Alerts issued ahead of the manager's own execution: subscribers buying first is the stated design, and it moves thin stocks.

5. Hold new purchases to the same standard of proof as existing holdings

1. Write down the evidence you require before opening a position — for this archive that is a fair value, a forward multiple against the company's own history, and a reverse DCF.
2. When a purchase is announced with less than that, note the gap rather than filling it in yourself.
3. Ask what is doing the work instead. Here: a track record, an analogy and management's own forecast — all three of which are weaker evidence than a valuation.
4. Decide whether the substitution is justified by the type of business (holding companies genuinely resist earnings multiples) or whether the standard simply slipped.
5. Record the omission so the position can be reviewed against it later.

• Management forecasts imported as inputs rather than tested as claims — a 15% growth target from the company itself is a marketing number until it is checked.
• A separately-linked "full investment case" doing the valuation work: it may well be there, but a transaction alert that stands alone should still carry the number that justifies the price paid.