Actionable insights — The 30-Year Buying Opportunity

1. Convert a factor-level dislocation into named positions, or leave it alone

1. Establish the dislocation at the style level and against its own long history — how far quality is lagging, and when it was last this wide.
2. Find the precedent and state what happened next, with a number, so the claim is checkable rather than atmospheric.
3. Then immediately narrow: which specific holdings does the dislocation make cheapest, ranked, with an expected return each.
4. Act only where you already own the business. A factor view is not a reason to buy something you have not underwritten.

• A single historical precedent doing the work of a base rate — n = 1 is an anecdote, and 1999 was a different market structure.

2. Measure "cheaper" as the price move against the earnings move, over the same window

1. Take a fixed window — a year works — and record two numbers: the share price change and the EPS change.
2. Combine them into a single statement about the multiple: price −15% with EPS +15% is a stock roughly 30% cheaper.
3. Do the same at portfolio level, over a longer window, to describe the whole opportunity set in one line.
4. Use it as the buy trigger. This is a fact about the price paid, not a forecast, so it does not depend on your growth assumption being right.

• EPS growth that came from buybacks or a one-off — the multiple genuinely fell, but the earnings base may not be repeatable.

3. Underwrite a new name by re-reading the original of the type

1. When a business resembles a famous historical winner, go back to the primary source on that winner rather than to the folklore.
2. Extract the two or three structural features that actually produced the outcome, and test the new name against exactly those.
3. Keep the list short and mechanical — segment growing faster than the market; company gaining share within it — so the analogy can fail.
4. Then add the company-specific checks: underwriting quality, runway, moat source, capital allocator.

• Analogy as permission rather than as a test — the GEICO comparison only means something if you would have dropped the name when a feature failed.

4. Use the founder's own capital-allocation words as the valuation opinion

1. Read the AGM transcript and the latest results call for statements about buybacks — the cleanest signal management can give about its own view of the price.
2. Prefer explanations of why not: a founder saying he would buy back stock but cannot afford to is more informative than one who simply does.
3. Check the alignment behind it — here, over 46% insider ownership — so the statement carries personal cost.
4. Note the price change since the statement was made; a cheaper stock strengthens the same testimony.

• The financing behind the constraint — this company raised debt because opportunities exceeded capital, and leverage against pledged founder stock is exactly the exposure that surfaces as a governance problem in April.

5. Recruit the most inconvenient available authority against the bear case

1. Identify who would be expected to benefit from the disruption narrative you are arguing against.
2. Find whether they have said anything that contradicts it, and quote the mechanism rather than the verdict.
3. Restate the mechanism in your own words so the reader can check it — here, that AI is being built to operate existing software, making software the toolbox rather than the target.
4. Pair the outside authority with a structural argument of your own, so the case does not rest on borrowed conviction.

• The witness's own incentive — the chip vendor benefits from AI being additive to software, so his interest and your thesis are aligned rather than independent.

6. Choose a strategy you can hold, not the one with the best backtest

1. Write your strategy in three lines. If it does not fit, it is not simple enough to hold through a drawdown.
2. Test it for temperament fit rather than optimality — the return you actually get is the strategy's return minus your abandonment of it.
3. Accept that a great investor's record is evidence that a coherent strategy works, not evidence that his strategy is yours to run.

• Simplicity used to avoid scrutiny — a three-line strategy still needs a written rule for when a holding leaves the portfolio.

7. Check what an "undervaluation" percentage is measured against

1. Whenever a discount is quoted, work out the denominator: a discount from fair value can never exceed 100%; an uplift to fair value can.
2. Convert it into a price multiple you can sanity-check — 118.7% "undervalued" means fair value is about 2.2x the current price.
3. Compare it against the expected return from the same model. Three names with near-identical expected returns (14.4%, 15.3%, 14.6%) but wildly different stated undervaluations (118.7%, 94.9%, 69.0%) tells you the exit-multiple and time assumptions are doing most of the work.

• Headline discounts quoted without the horizon over which the gap is expected to close — the same gap over three years or ten years is a completely different investment.