Actionable insights — Stocks Will Fall -70%

8:00 1. Audit a "borrowed-credibility" metric before you trust it

1. When a bear leans on a named indicator (e.g. "the Buffett indicator"), notice that the famous name is doing persuasive work — separate the brand from the math.
2. Check whether the originator still uses it: if the person it's named after abandoned it decades ago (and has argued the opposite since), that's a red flag.
3. Test whether the metric's assumptions still hold today — has the world changed in a way that breaks it?

• A metric carried by a famous name rather than current logic; the originator no longer citing it; structural changes (globalization, margins, buybacks) the ratio ignores.

11:43 2. Reject the unfalsifiable forecast

1. Ask the forecaster for a timeframe and an invalidation condition. A prediction with no way to be scored is entertainment, not analysis.
2. If the timing window is so wide it can't be wrong ("2 weeks to 2 years"), treat the call as content-free regardless of how dramatic the number is.
3. Weight only forecasts that put a date and a level on the table and can later be graded.

• A scary magnitude with a vague/elastic timeframe; "it's coming" without a level or date; excuses pre-built for when it's wrong.

12:36 3. Score a forecaster on the full record, not the cherry-picked hit

1. When someone cites a famous correct call (2000, 2007), pull the entire timeline — including how early and how long they were wrong around it.
2. Compute the cost of having followed them: if you'd have been out of the market for the years before and after the call, the "win" may still have lost you money vs staying invested.
3. Demand symmetry: count the missed bull years against the called crashes.

• A record built on one or two calls; no accounting for years of being early/wrong; "I said overvalued" walk-backs of what were clearly crash calls.

5:53 4. Model the realistic investor, not the worst-case single-top buyer

1. When a bear says "you'd have waited X years just to break even," check the hidden assumption: it usually presumes you bought one exact top and never invested again.
2. Re-run it for how people actually invest — dollar-cost averaging in over time — and the "lost decade" framing usually dissolves.
3. Remember normal price behavior: stocks sit 10–20% below highs and then spike; the flat stretches are the holding cost of the eventual gains.

• "Flat for a decade" claims that compare bubble-peak to bubble-peak; an implicit lump-sum-at-the-top assumption; ignoring dividends/averaging.

18:14 5. Don't accept "the multiple is high" as a thesis — decompose why

1. A high PE versus history is an observation, not a conclusion. Before calling overvaluation, list the structural reasons a multiple could legitimately be higher now.
2. Check each: more global/diversified revenue (less risk), higher margins (more profitable businesses), wider moats, faster growth — all push fair multiples up.
3. Only if none of those explain the re-rating is "expensive" a real warning.

• "It's expensive vs history" with no mechanism; ignoring margin/moat/growth/globalization changes; using an elevated multiple as both the evidence and the conclusion.

28:05 6. Separate "I dislike how it's run" from the economics — buy the governance discount

1. When the market discounts a quality business for a governance reason (founder voting control, unpopular spending), isolate whether the objection actually impairs growth, margins or the moat.
2. If the founder's bets plausibly widen the moat, the discount is an opportunity, not a warning — weigh the valuation against the growth.
3. Keep the disliked-action and the sell-decision separate, exactly as the "judgment calls" rule does.

• A cheap multiple on a fast grower explained mainly by a governance/founder objection; capex that builds owned capability vs rented; the moat strengthening despite the discount.