Actionable insights — Bill Ackman Just Bought 6 New Stocks

15:08 1. Size the AI-disruption fear against the revenue line it actually threatens

1. When a stock falls on an AI-disruption headline, name the specific product the new tool competes with — not the company, the product line.
2. Pull that line's share of revenue and of profit from the filings. A feature-level threat to a single-digit-percentage line is not a company-level threat.
3. Run the zero test: assume the threatened line goes to zero and recompute the forward multiple on the remaining earnings. If the doomsday multiple is still reasonable, the market has priced a scenario worse than annihilation.
4. Then check where the growth comes from — if the compounding engine is a different segment entirely, the disruption story can't drive the thesis either way.

• A viral demo rather than a customer loss; the threatened product being a desktop/UI layer while the moat is the underlying data; management calls describing "business as usual" while the multiple compresses to a five-year low.

9:34 2. Decompose a "deteriorating" metric into quality and mix before believing it

1. Take the single metric the bears are quoting (here, watch time per subscriber) and ask two questions: are all units of it the same quality, and is the denominator changing?
2. Quality: identify the units that drive the outcome you care about (sign-ups, retention, pricing) even though they're a small share of the total.
3. Mix: check whether the average is falling because new customers are structurally lighter users — growth into a lower-intensity cohort mechanically lowers a per-user average while the total rises.
4. Only after both adjustments is a declining average evidence of decline.

• A per-user average falling while totals rise; a bear case that rests on one ratio; management explaining mix and being ignored; the same metric being used to compare fundamentally different products (a series versus a short-form feed).

17:09 3. Ask where the disruptor is actually incumbent — new rails often grow the pie

1. For any "new technology kills the incumbent network" claim, map the use cases where the new technology has a real advantage.
2. Check whether the incumbent already serves those use cases. If the new rail wins mainly where the incumbent was never the incumbent, the two grow in parallel rather than substituting.
3. Separately, ask whether the technology changes transaction volume. Anything that removes friction from buying tends to create more transactions — good for a per-transaction toll-taker.
4. Only treat it as displacement when the new rail attacks the incumbent's core, high-margin corridor.

• A disruption narrative that never names the corridor being taken; adoption metrics concentrated in markets the incumbent doesn't serve; friction-removal technologies being scored as substitutes rather than volume multipliers.

1:20 4. Audit overlap with a famous investor by sequence, not by holding

1. When your book overlaps a well-known manager's, list each shared name with the date you established it versus the date their stake was disclosed.
2. Names you owned first are independent confirmation; names you bought after are the ones to audit for borrowed conviction — can you state that thesis without citing them?
3. Be explicit about the follow-ins rather than hiding them; that's what keeps the distinction honest over time.
4. Treat a large overlap as a signal that a whole category is being mispriced, not as validation of any one pick.

• A thesis you can only articulate by quoting the famous holder; a growing share of the book that post-dates someone else's filing; conviction that softens when they trim.

0:44 5. Mine the letter for implied growth rates, not the position list

1. When a manager publishes a full letter, skip past "what he bought" to the section giving valuation and implied multi-year growth per holding.
2. Line each implied growth rate up against sell-side consensus for the same name.
3. Rank by the gap. The biggest divergences are where the manager is making a real, testable claim — and where you can decide whether you agree.
4. Then read the qualitative section for that name to see what the gap rests on.

• Large gaps between a manager's implied growth and consensus with no stated mechanism; a letter that lists positions but never puts numbers on the forward case.

22:15 6. Turn an index-breadth statistic into a shopping list

1. Decompose the index's year-to-date return by sector: what share of companies, what share of market cap, and what share of the gain.
2. If a tiny slice of the index produced nearly all the return, the rest has been sold to fund it — that's a flow phenomenon, not a verdict on those businesses.
3. Invert the screen: hunt in the bucket that contributed nothing, and require an identifiable narrative (not a fundamental break) as the reason each name is down.
4. Size it as cash deployment into dispersion, not a market-timing call — the index level is irrelevant to the exercise.

• Return concentration in a handful of sectors; quality names at five-year-low multiples with intact fundamentals; the narrative for each drawdown being a headline rather than a number.

9:19 7. Treat re-entering a former loser as discipline, not inconsistency

1. Separate the past outcome of a position from the current setup. A prior loss in a name is a sunk fact, not information about today's price.
2. Ask what has materially changed since you sold — in the business, and in what the market now believes.
3. If your original objection has been answered and the price is lower in real terms, the ego cost of buying back is the only thing stopping you.
4. Apply the mirror rule too: exiting fast when the thesis breaks is what earns the right to re-enter later.

• Refusing to revisit a name purely because you lost money in it; a former objection that the company has since demonstrably addressed; the stock back at or below your old exit while the business is materially better.