
Nov 17, 2026 · 1h 2m
Poker discipline reshapes hedge-fund investing in the AI era
From Poker Pro to Elite Hedge Fund Manager: David Orr on Investing, Risk and AI
David Orr connects probabilistic decision-making, portfolio construction, and AI-assisted research to the harder question of where human judgment still creates an edge.
- 1Poker’s lessons—small bets, emotional discipline, and probabilistic thinking—translate into investing under uncertainty.
- 2Orr favors diversified portfolios, simple theses, careful correlation management, and adaptability over deep research or fixed factor styles.
- 3AI can make research and portfolio management more capable, but human judgment remains essential for challenging consensus and managing change.
Don't miss
Orr delivers the episode’s sharpest practical conclusion: most individual investors should use broad index funds and reserve active investing for a small hobby allocation.
The brief
David Orr traces his move from professional poker to hedge-fund investing, arguing that both fields reward probabilistic thinking, emotional discipline, and acting before information is complete.
His process favors small positions, simple long and short theses, broad diversification, and correlation control—while rejecting deep research when it mainly reinforces an existing view.
Japan provides a practical case study: corporate-governance reforms and improving capital returns can create opportunities, but stubborn family control without a catalyst remains a warning sign.
Orr uses AI for translation, earnings triage, research, and risk management, yet expects the strongest systems initially to pair customized tools with humans who can challenge consensus.
His advice to individuals is notably restrained: use broad index funds, treat active investing as a small hobby allocation, and test honestly whether skill—not social reinforcement—is driving results.
What was said on this episode
56 statements · 24 positive · 27 negative · 5 neutral
Investors should act using limited information rather than waiting for complete certainty.
“what you want to be able to do is operate off of 10 or 20% of the information”
Listen at 1:03
Waiting for a perfect investment thesis causes the price advantage to disappear.
“by then the price is already moved”
Listen at 3:29
Stocks involve imperfect information and probabilistic decision-making like poker.
“stocks i think are the exact same thing”
Listen at 4:25
Persisting with a flawed poker strategy loses money over time.
“if you stick with that leak, it's just going to burn money over time.”
Listen at 4:53
Financial markets are substantially inefficient and can offer large investment edges.
“this is not efficient at all, actually”
Listen at 5:57
Investing offers a larger edge than poker because investing is positive-sum while poker is negative-sum.
“the edge is way bigger in investing because poker is a negative sum game and investing is a positive sum”
Listen at 6:22
Beta has little importance for his investment returns and is mostly noise.
“The beta doesn't really matter. It's kind of noisy.”
Listen at 8:19
Leveraged investors were forced into severe losses during the market sell-off.
“all the like levered investors are blowing up”
Listen at 9:11
Stocks may reprice lower following systemic market stress.
“maybe stocks will reprice you know a bit cheaper”
Listen at 10:35
His portfolio should generally gain or remain unaffected during market crashes.
“normally in a crash like that, I would expect to probably win or not not be affected.”
Listen at 11:18
Some Japanese businesses appear inexpensive at roughly nine times earnings.
“why is this only like? nine times earnings”
Listen at 11:52
Deep investment research can reduce performance by introducing bias.
“I think it gives you a negative edge because you get biased.”
Listen at 12:25
Extended investment research has a poor return on time.
“your return on time is low”
Listen at 12:29
His investment process focuses on businesses' five- to ten-year prospects.
“I'm completely focused on, you know, where the business is going to be in five or 10 years or whatever.”
Listen at 13:12
Markets are not efficient.
“markets are, in fact, not efficient at all”
Listen at 13:27
Militia Capital holds approximately 500 positions.
“I've got 500 positions on.”
Listen at 13:56
Simple value factors no longer reliably produce excess returns.
“that simple of a factor is not really going to cut it”
Listen at 15:10
Traditional factor investing does not work reliably.
“I don't think that kind of factor investing works.”
Listen at 15:57
AI will make investment tools cheaper and more capable.
“It's going to make all the tools really cheap compared to before and better probably.”
Listen at 18:06
AI could eventually manage investment portfolios directly.
“the AI itself would become the portfolio manager”
Listen at 18:21
Human portfolio managers could become obsolete because AI can process more information.
“I think humans could easily go obsolete.”
Listen at 18:43
Militia Capital's AI-assisted risk system is substantially better than its previous system.
“we built this risk system that's like way better”
Listen at 19:21
AI investment tools will become broadly available at very low cost or free.
“everyone's going to have it for a very low price or for free”
Listen at 20:26
Orr now conducts his investment research entirely through AI.
“I don't really do any other kind of research anymore. It's all done through AI now.”
Listen at 20:40
Current AI investment analysis largely reproduces consensus views.
“AI basically doesn't. It's just consensus views.”
Listen at 22:19
Hybrid human-AI investment teams will be optimal for at least five to ten years.
“for at least five years, maybe 10, it'll be the you combine a really good custom AI system with a really good human portfolio manager. That will be the best thing.”
Listen at 23:36
Human portfolio managers may become irrelevant in roughly ten years.
“in 10 years, I'm worried that human becomes irrelevant”
Listen at 23:48
AI competition could eventually eliminate most investment edges.
“the edges could go to basically zero”
Listen at 24:12
AI will be broadly beneficial for society.
“I think it's really good for society.”
Listen at 27:58
AI will displace many workers.
“it'll put a lot of people out of work”
Listen at 28:10
AI could phase out accounting and corporate-law jobs over ten to twenty years.
“if this can get phased out over 10 20 years”
Listen at 28:46
Identifying the winning AI company or business category is difficult.
“picking the winner is hard”
Listen at 30:55
No one can reliably know which AI businesses will win long term.
“I don't know who's going to win. I don't think anybody can know.”
Listen at 31:49
Compute is a commodity and will follow a boom-bust cycle.
“Compute's a commodity.”
Listen at 32:13
Building frontier AI models appears likely to be a poor long-term business.
“the people actually making these fancy models actually seems like a pretty terrible business to me long-term.”
Listen at 32:20
Compute capacity may already be close to short-term oversupply.
“we're close to oversupplied short-term now”
Listen at 33:28
AI stocks could fall approximately 80% if compute is already oversupplied.
“the ai stocks are going to go down like like 80% or something.”
Listen at 34:20
AI research response times may fall to roughly five seconds within eighteen months.
“in another year and a half, is that going to be like five seconds?”
Listen at 36:06
Long positions should generally have growing earnings per share.
“on longs, I just want things that are going to grow earnings per share”
Listen at 37:57
Some businesses could become ten times larger over a ten-year horizon.
“if you're looking out 10 years, you can sort of say, oh, I think this business could be easily 10 times as big”
Listen at 38:26
Photocopiers are no longer needed.
“We don't need photocopiers anymore. It's just nonsense.”
Listen at 39:26
Xerox's market capitalization was approximately $100 million.
“I think the market cap's like a hundred million bucks now or something.”
Listen at 39:38
Investors should keep individual company positions small to control risk.
“You bet small on each company.”
Listen at 40:13
Crowded short positions should be avoided because they are highly correlated.
“one thing i like to avoid are these kind of crowded shorts because they all move together”
Listen at 40:44
Shorting correlated thematic baskets can lose despite correct long-term fundamentals.
“If you're shorting it, you're going to lose anyway.”
Listen at 42:58
Core portfolio managers have approximately 0.15 cross-correlation.
“our cross correlation between the core portfolio managers is about 0.15.”
Listen at 43:27
Hikari Sushin is substantially undervalued.
“This stock is so cheap.”
Listen at 45:19
Hikari Sushin trades at approximately ten times earnings.
“it's 10 times earnings”
Listen at 45:48
Japanese companies are already improving balance sheets and returning cash to shareholders.
“they're already way, way improving their balance sheets and returning the cash.”
Listen at 46:27
Investors should avoid stubborn family-controlled Japanese companies lacking credible catalysts.
“I would avoid the very hard family control situations where they've already shown they're being stubborn.”
Listen at 48:14
Militia Capital has operated at approximately 400% gross leverage since inception.
“I've run the fund 400% gross leverage since inception.”
Listen at 48:36
Good portfolio managers adapt their strategies to changing markets and opportunities.
“any good PM is kind of adapting to the market and the opportunity set.”
Listen at 49:42
Long historical factor records may contain only about eight independent market periods.
“we're talking about like eight different correlation periods. You have a sample size of eight there.”
Listen at 52:05
The S&P 500 or VT are suitable core investments for most people.
“I think that that or VT is a good one.”
Listen at 52:31
Most investors should place roughly 95% of their assets in the S&P 500.
“why don't you just put like 95 of it into the S&P 500”
Listen at 53:43
Investors should make small bets so mistakes have limited consequences.
“just put the put the bed on small so if you're wrong it doesn't matter”
Listen at 56:52
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.