Actionable insights — Ferg's Finds (Sep 1, 2026)

1. Convert a high multiple into a failure probability — the overdetermined-failure screen

1. Stop treating a rich valuation as a risk to be sized around and start treating it as a forecast. The question is not "could this fall?" but "how many independent ways does this now have to disappoint?"
2. Enumerate the distinct failure paths for the story the multiple is pricing — a demand air-pocket, a competitor, a capacity build, a financing rate, a regulatory move, a customer concentration, an accounting normalisation. Each individually looks unlikely; write down the count.
3. Assign each a low, honest probability (Kedrosky's working number: "maybe 5% at most") and combine them. Twenty roughly-independent 5% paths do not stay a 5% problem — they compound into a majority-odds outcome.
4. State the result as a probability rather than a warning: an "overdetermined system" at these levels implies "greater than a 60% chance of failure," which flips the asset from unpredictable to highly predictable.
5. Apply the test symmetrically. A low multiple has the mirror property — the disappointments are already priced, so the count of paths that hurt you is small. That is what makes a cheap side of a valuation spread a structurally different bet, not just a cheaper one.
6. Sanity-check the mechanism, which is behavioural: at a high multiple "the market increasingly looks at it and says, 'How can this thing continue to outperform expectations?' And the answer is, whoops, it can't" — the multiple itself is what makes every subsequent print a potential disappointment.

• Any position where the bull case requires a sequence of things to keep going right. Count the required conditions; if there are more than a handful, the multiple is doing the forecasting for you.
• The tell that the regime has turned: the market "starts pointing around" — attributing a miss to one cause, then another. That is the enumeration happening in public.

2. Date the supply, not the demand — the capex-to-price-collapse clock

1. In any boom-bust industrial sector, ignore the demand forecasts (everyone has them) and track the cash going into capacity — inflows to manufacturers, announced fabs/mines/plants, equipment orders.
2. Add the build lead time to get the arrival date of that capacity. Capacity funded now is product later, and the date is a fact, not a projection.
3. Ask what happens when it lands into a business with high fixed costs: the operator cannot hold it back, because "I've got to cover my fixed costs. So this stuff is all going out the door." Output is sold at whatever it fetches.
4. Conclude in price terms with a date attached, not a vague warning — the point of the exercise is that the supply side is knowable in advance while the demand side is not.
5. Position accordingly: be wary of names whose current multiple capitalises today's pricing power into perpetuity when the supply clock says pricing power expires on a specific horizon.

• Capex-to-revenue ratios at the manufacturers rather than at the buyers; government-subsidised capacity, which is even less price-disciplined.
• The 2027–2028 window specifically for semiconductors, per the quoted call — and, generally, the moment order books stop lengthening.

3. The "KOL delisting" set-up — buy a forced liquidation with a published terminal price

1. Monitor fund closure and delisting announcements — ETFs, closed-end funds and trusts being wound up by their sponsors. The announcement publishes the schedule: last trading day, suspension date, liquidation-distribution date.
2. Recognise what that schedule gives you: a known terminal value on a known date. The wind-up pays out net asset value, so the exit price is not a forecast — it is arithmetic, subject only to what the underlying holdings fetch.
3. Look for the forced sellers that make the pre-liquidation price diverge: index and mandate-constrained holders that can no longer hold a delisting security, plus holders unwilling to wait for cash. Thin final-session liquidity does the rest.
4. Estimate the discount to NAV in the final sessions. The wider the discount and the shorter the wait to the distribution, the higher the annualised return on a position whose payoff does not depend on the market direction.
5. Weight the risk correctly: the risk is not the market — it is NAV slippage while the manager liquidates illiquid underlying holdings (frontier-market equities being the extreme case), plus any withheld reserve.
6. Backtest before you size it. Ferg's note is that he ran "a quick backtest" — the set-up only counts as a screen once the historical distribution of discounts and outcomes has been checked.

• Sponsor announcements of fund closures — particularly funds holding illiquid, unloved or hard-to-access markets, where the forced-seller discount is widest.
• The gap between last-trade price and last-published NAV in the final week; the length of the wait to the distribution date.

4. Swap the weighting scheme instead of taking the other side of it

1. When you conclude that a dominant approach is fragile, separate two different trades: betting against it and expressing the same exposure differently. Ferg rules the first out explicitly — "this doesn't mean I want to take the other side of market cap weighting; I want to go long in a smarter, passive way."
2. Diagnose the fragility in structural terms rather than in market-direction terms. Market-cap weighting is "incredibly efficient and is hard to beat" — the analogy is the just-in-time supply chain: "can't beat them until something breaks; then resilience/redundancy is everything." The objection is to the missing redundancy, not to the efficiency.
3. Audit the concentration with one number: the weight in the top three (or top ten) holdings, not the holdings count. A count of holdings is a marketing statistic; the top-weight share is the actual bet.
4. Choose a replacement weighting rule anchored to something a company cannot fake. Shareholder yield — dividends plus buybacks plus debt paydown — measures cash that has already left the building, which is why it is harder to manufacture than reported earnings.
5. Check the price of the swap before making it: if the alternative weighting also lands you on a materially lower multiple, the trade costs nothing in expected return terms and buys the redundancy for free — "little opportunity cost and a fat margin of safety."
6. Make it the benchmark, not just an allocation. Ferg commits to what he "plan[s] to benchmark my portfolio against moving forward" — a discipline that forces every discretionary position to justify itself against the cheap mechanical alternative he could hold instead.

• Any index fund you own where the top three weights exceed ~20–25% — run the same audit on it.
• The condition that would validate the thesis early: a drawdown concentrated in the mega-weights, where the "diversified" cap-weighted fund and its three largest holdings fall together.

5. Triage the reading queue by independent recommendation count

1. Do not curate a reading list by your own impulse or by publisher marketing — both are noisy and both are biased toward what is new.
2. Keep a running tally of independent recommendations for each title, from sources that are not talking to each other.
3. Set a hard threshold and obey it mechanically: "I have a rule: after a book is recommended three times, I read it!"
4. Why it works: the third independent recommendation is evidence of durability rather than of promotion. A book that keeps surfacing across unconnected people has survived a filter that no single review can supply — and the rule costs nothing, because it simply orders a queue you were going to work through anyway.
5. The same threshold generalises past books: papers, newsletters, tools, and the analysts worth reading.

• Titles crossing the third mention. Also note the companion habit visible in the post: pair a current research piece with an historical one on the same mechanism, and read them close together — the pattern-matching is the point.

6. Pre-commit to the social cost of a contrarian holding

1. Before entering an out-of-consensus position, write down the three phases of criticism you have agreed to absorb: "when you find a winner, people will say you are wrong. When you hold a winner, people will say you are stupid. When you get rich from a winner, people will say you got lucky" (Ian Cassel).
2. Use the sequence as a confirmation that you are in the right kind of position rather than as a reason to review it — the criticism arrives on schedule and carries no information about the thesis.
3. Separate, explicitly, the two things that look identical from outside: the thesis breaking (a reason to sell) and the position being unpopular (not a reason to sell). Only the first belongs in the sell discipline.

• Your own reaction to the second phase ("stupid") — the drawdown-plus-mockery window is where most contrarian positions are actually abandoned, well before the thesis is tested.