Episode summary
We Asked a Hedge Fund Manager With 40%+ Annual Returns Why Too Much Compute Could Break the AI Trade
Summary
David Orr argues that investing works more like poker than like a classroom model of finance: the edge comes from acting on incomplete information, moving quickly, and sizing bets small enough that being wrong does not matter. He says more research can actually hurt by adding bias and by delaying decisions until the market has already moved. His process is fundamentally driven, but deliberately simple, with a focus on a small number of clear long and short theses rather than deep dives.
He describes Militia Capital as a high-turnover, highly diversified book with about 500 positions and roughly 400% gross leverage, while paying close attention to correlation buckets and crowded trades. He is especially skeptical of crowded factor trades and says many classic factor strategies have been overowned. He also explains why he thinks Japan remains attractive: corporate reforms, buybacks, and cheap valuations are creating real opportunities, especially when the thesis is visible in day-to-day filings and business behavior.
On AI, Orr is bullish on the tools but cautious on the trade. He expects AI to replace a lot of research work and possibly become part of portfolio management, but he thinks the best setup for years will be a hybrid of human judgment and AI. He worries compute could become oversupplied as models get more efficient, which he says could eventually hit AI stocks hard.
Key takeaways
- Orr thinks the best investing decisions come from acting on 10% to 20% of the information, not waiting for a perfect thesis.
- He says Militia runs about 500 positions and roughly 400% gross leverage, with risk controlled mainly through correlation management.
- He believes Japan remains one of the best hunting grounds because reforms, buybacks and cheap valuations are already showing up in the data.
- He sees AI as a major productivity boost for research and risk management, but not yet a full replacement for human judgment.
- He expects compute to become oversupplied as model efficiency improves, which could pressure AI infrastructure and related stocks.
- For most individual investors, he recommends broad index funds like the S&P 500 or VT rather than active stock picking.
Notable quotes
“what you want to be able to do is operate off of 10 or 20% of the information”
“Stop analyzing so much. I just make decisions, don't get emotional, don't think about it too much and just put the put the bet on small”
“I've got 500 positions on”
“Computes a commodity. That's pretty obvious”
“the the big one is just stop analyzing so much, just stop trying to be so high IQ about everything”
Topics
- ai compute oversupply
- incomplete information investing
- poker probabilities and uncertainty
- long short portfolio risk
- japan equity opportunities
Companies and tickers mentioned
- Militia Capital
- AQR Capital Management
Watch the episode
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from Excess Returns
-
My Biggest Bet Isn't the Stock Market | Inside Meb Faber's Personal Portfolio -
Stocks Are Falling. Put Buying Hit a Record. So Why Does an Options Trader See a Thanksgiving Rally? -
We Asked Fidelity’s Director of Global Macro Who Wins If AI Spending Slows -
An AI Bubble Could Hit Stocks. Inflation Could Hit Bonds. How Do You Diversify and Keep Both? -
Tech Is Priced for a Boom. 86% of the Economy Grew 0.7%. What If AI Slows? -
Tech Is the Cheapest Since 2022. Walmart Is 46 Times Earnings. Where Is the Bubble?
Excess Returns, this episode and its artwork belong to Excess Returns. Octagon is not affiliated with the show; this page is an independent summary.