Trader Ferg — Ferg's Finds (Sep 1, 2026)

The format — a weekly curation email

• "This is a short weekly email that covers things I've found interesting during the week." Five fixed sections — Article/Research, Podcast/Video, Quote, Tweet, Charts — plus a closing "Something I'm Pondering," which is where the actual portfolio thinking lives.
• Worth noting for reading the archive: the ideas in the first four sections are other people's; the last section is Ferg's own, and it is the one that produces a decision.

Article/Research — Myrmikan on AI debt and the next round of bailouts

• The pick: Myrmikan Research, "AI Debt Failure Will Prompt Another Wave of Fed Bailouts."
• Ferg's whole comment is a statement of method rather than of view: "nothing like a good dose of history to help understand the present."
• The sequence being pointed at — debt-financed build-out → failure → central-bank rescue — is the frame the rest of the post's valuation material sits inside.

The book rule — three recommendations and it gets read

• "I just started Tomorrow's Gold (I have a rule: after a book is recommended three times, I read it!)."
• A filtering heuristic rather than a reading list: independent recommendation frequency does the triage, so the queue is ordered by other people's revealed conviction instead of by the reader's own impulse.
• The payoff is noted as a coincidence of timing — a passage in the book struck him as "humorous, having only recently read the above" (the Myrmikan piece), i.e. the historical pattern and the current one rhyming.

Podcast/Video — Kedrosky on overdetermined failure

• The pick: "The AI Boom Is Creating a Financial System We Haven't Seen Before (Paul Kedrosky Explains)" — flagged as "the best articulation of the danger of high multiples/valuations I've heard."
• The mechanism, quoted at length: "at high valuations, failure is overdetermined in a statistical sense, meaning that there are so many ways to fail, all of which are low likelihood—maybe 5% at most—that when you combine them all and turn it around and say, given 20 different ways this could fail because we're at this precipice… an overdetermined system tells you that… there's greater than a 60% chance of failure."
• The inversion that makes it usable: "what looks not particularly risky or unpredictable is actually highly predictable because… when your PE is 70… failure is overdetermined."
• Why the multiple itself is the cause, not a symptom: "the market increasingly looks at it and says, 'How can this thing continue to outperform expectations?' And the answer is, whoops, it can't. And it starts pointing around. And the same thing is true in this AI moment."

The dated supply call — a chip tsunami in early 2028

• "I've been pointing lately to the incredible and unprecedented cash inflows into Taiwanese and Chinese chip manufacturers, with what looks like a tsunami of supply in early 2028."
• The industry structure that turns the capex into a price collapse: "and you know this is a boom-bust industry. So once you lock in supply, my friend, prices are going to zero."
• The reason clearing prices go to marginal cost rather than to zero volume: "Why? Because I've got to cover my fixed costs. So this stuff is all going out the door." Capacity, once built, is sold at whatever it fetches.

Quote — Ian Cassel on the social cost of holding a winner

• "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.
• Ferg adds only that Cassel's "new book is now on my to-read list" (the title is not given in the post).
• The relevance to the rest of the email: a 9x-multiple, out-of-favour allocation held against a 20–22.5x index is precisely the position that draws each of those three reactions in sequence.

Charts — the FM liquidation, and the set-up he missed

• "I'm annoyed I missed this! It was a perfect 'KOL delisting' set-up."
• The event, quoted with its dates: "FM: iShares Frontier and Select EM ETF, was delisted after its final trading session on January 6, 2025. Trading was suspended before the market opened on January 7, and shareholders received the liquidation distribution of $27.23 per share on January 9, 2025."
• "A quick backtest…." — i.e. the set-up was tested, not just observed. The generalisable feature: a fund winding up pays out net asset value on a scheduled date, so the terminal price is known in advance while the market price in the final sessions need not match it.

Something I'm Pondering — is market-cap weighting just-in-time investing?

• "I'm pondering how market cap weighting became the optimal way to allocate capital passively. Yes, it is incredibly efficient and is hard to beat."
• The analogy that carries the argument: "yet it reminds me of the same arguments made for 'just-in-time supply chains.' Can't beat them until something breaks; then resilience/redundancy is everything."
• Note the shape of the claim — it is not that cap-weighting underperforms, but that its optimality is conditional on nothing breaking, and that the cost of the missing redundancy is paid all at once.

The exhibit — EEM's 1,226 holdings and its three real ones

• "Take EEM, where it's comical that it has 1,226 holdings yet allocates 27.61% to three companies…"
• The diagnostic worth reusing: measure a fund's diversification by the weight in its top few positions, not by its holdings count.

The alternative — Meb Faber's shareholder yield, adopted as the benchmark

• The negative case is ruled out first: "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."
• "My favourite approach is Meb Faber's shareholder yield. As I went over in my recent piece, I see little opportunity cost and a fat margin of safety in shareholder yield, as it has outperformed and is what I plan to benchmark my portfolio against moving forward."
• The commitment embedded in that sentence is the notable part: he is not merely allocating to it, he is agreeing to be measured against it — which raises the bar for every discretionary position he takes.

The price of the alternative — 9x against 20x and 22.5x

• "What's even more impressive is that EYLD sits on a forward PE of 9x, while SPY is 20x and QQQ is 22.5x."
• "My view is that the above performance gap will continue to widen for years to come."
• Read with the Kedrosky section, this is the same idea expressed twice: the number of ways to disappoint scales with the multiple, so the cheap side of a 9x-vs-22.5x spread carries structurally fewer failure paths.

P.S. — the prior work this continues

• "I really enjoyed putting together this piece, which is really a continuation of what I started with Hugo's Portfolio: 60/40 for the next Decade."
• Useful archive context: the shareholder-yield material here is the current instalment of a longer, ongoing project on how to build the passive core of a portfolio for the coming decade — not a one-off observation.