17:53 1. The crash-driven checklist — learn only from wrecks (the FAA model)
The repeatable method
- Borrow the FAA's discipline: only change your process after a "crash." In investing, a crash is a position that lost money; a zero is a crash with no survivors.
- Study the great investors' losers, not their winners. For each, ask: was the loss visible before takeoff? Turn each visible cause into a checklist question.
- Run the checklist on every new idea before buying — "a pilot does not take off before running the pre-flight checklist." Where you can't answer a question, you haven't done the work yet; go find the answer.
- Keep it pragmatic — don't invent hypothetical failure modes that never actually caused a crash (the "peanut girl" who forced a plane back to the gate added a check that wasn't required).
Here: Buffett's Dexter Shoes (US shoemaker wiped out by cheap foreign labor) became the question "can this business be killed by cheap foreign competition?" — which flags a US bicycle maker but not a US bank. His checklist has grown from ~70 to 213 questions over ~17 years.
Watch for
- Recurring causes of permanent loss across many investors' mistakes; any new idea you can't fully answer the checklist on (that gap is the risk).
25:25 2. The retail three-item checklist — leverage, moat, owners
The repeatable method
- No leverage. The #1 reason investors lose money. Prefer companies with zero/low debt (the best businesses don't need it), and never borrow to buy stock yourself.
- Durable moat. Judge how hard it is for competitors to take the business — and be honest about whether the moat is deep or shallow.
- Owner/manager quality (governance). Do they care about shareholders, employees and customers — do they love the business, or just the money? It's fine to like money, not to love it.
Here: IKEA's founder ran the company 70+ years with zero debt (deep-moat, owner-obsessed) as the leverage/quality benchmark; 090430.KS Amorepacific is the shallow-moat counter-example; the memory names (000660.KS, 005930.KS, MU) are the deep-moat case.
Watch for
- Debt on the balance sheet or in your own account; a moat that's really just a good product; managers who love money more than the business.
32:15 3. The 10-year cash-flow-vs-market-cap test — only buy the obvious
The repeatable method
- For any single stock, ask: what cash flow will this business generate 5, 10, 15, 20 years out, and how does that compare to today's market cap?
- If you can't answer with a high degree of certainty, don't buy it — buy the index instead.
- Only act when the answer is so obvious it's a no-brainer: e.g. a business you can buy for ~3× earnings that pays out ~100%, returning all your capital in ~3 years while you still own it.
Here: a hypothetical regulated Korean power company at 3× earnings paying 100% dividends — money back in 3 years, then decades of dividends "risk-free." Contrast: asking the same question about 000660.KS SK Hynix ($500B cap, ~$50B/yr today) is far less certain.
Watch for
- Boring, regulated, cash-gushing businesses at low single-digit earnings multiples where future cash flow is near-certain.
33:51 4. Hated-and-unloved — hunt anomalies, screen the cheapest market
The repeatable method
- Do nothing most days — just watch. Once every 6-24 months something turns up that "makes no sense"; step in then.
- Fish where it's boring and despised: risk-free bargains cluster in "hated and unloved" sectors (boring utilities at 3-5× earnings) because everyone's chasing excitement elsewhere.
- Screen at the market level for the cheapest country in the world, then go look at the actual companies on the ground before deciding.
Here: Turkey screened as the world's cheapest market, so he visited; RYSAS.IS Reysas was a ~$16M cap vs ~$800M liquidation value ("a $1M apartment for $20,000") and he ended up owning 40%. Korean power companies at 3× earnings are the domestic analog.
Watch for
- Whole markets/sectors trading far below asset or earnings value out of neglect; a specific name priced at a fraction of its liquidation value.
38:47 5. Heads-I-win/tails-I-don't-lose — then stress-test the one real risk
The repeatable method
- Before buying, answer "how can I lose money here?" — and the answer should be "it basically can't."
- List the concrete ways the thesis could break, then chase each to ground with evidence rather than assuming it away.
- When the downside is genuinely capped and the upside is open, buy — and then just hold; "anytime we have to sell, we made a mistake."
Here: Reysas — no debt, 100%-leased prime warehouses, tenants can't leave quickly. The one real risk was an earthquake, so he verified earthquake insurance and 8.0-rated construction; subsequent Istanbul quakes did <$10 of damage.
Watch for
- A single identifiable catastrophe risk on an otherwise bulletproof asset — verify it directly instead of hand-waving it.
41:45 6. Shiny-object rotation — treat "deeply loved" as a sell signal
The repeatable method
- Track what the crowd currently loves — that's what to avoid. Money rotates from one shiny object to the next, and each rotation funds the buy by dumping the last darling.
- Read the rotation forward: identify the current darling and the next one queuing up, and stand aside from both.
- Do the opposite of the flow — buy hated and unloved. If you can't resist chasing the loved thing, just buy the index instead.
Here: Bitcoin was deeply loved until AI arrived; the crowd dumped Bitcoin (100k → 70k) for AI; next they'll sell AI to buy the SpaceX mega-IPO. His tell that a theme is topping: it has become "deeply loved."
Watch for
- An asset everyone adores and a bigger, newer story lining up behind it — the hand-off is the exit, not the entry.