We Study Billionaires
We Study Billionaires

RWH015: Betting Better in Markets & Life w/ Annie Duke

IN THIS EPISODE, YOU’LL LEARN: 12:49 - How Annie Duke got sick, quit academia, & became a professional poker player. 18:24 - How poker taught her to “embrace uncertainty” & recognize the limits of what we know. 27:26 - How a legendary poker champion taught her to think more rationally about

Featured Speakers

Stig Brodersen HostAnnie Duke Guest

Topics Discussed

Episode Summary

Executive Summary: Annie Duke explains how poker taught her to think probabilistically, respect uncertainty, and build rules for quitting before emotions take over. She links these lessons to investing, showing why people hold losers, ignore new evidence, and let identity and bias distort decisions. The conversation centers on kill criteria, precommitment, and using outside voices to make better choices in markets and life.

Main Topics: From academic psychology to poker to decision science (Priority: 5/5): Duke recounts her planned academic career, illness-induced exit from graduate school, and transition to poker, framing all later work as part of one thread: understanding learning, decisions, and feedback under uncertainty. Lila Gleitman’s influence and language bootstrapping (Priority: 5/5): She honors mentor Lila Gleitman, explaining how psycholinguistics and syntactic bootstrapping shaped her scientific mindset and her later approach to decision-making. Poker as a laboratory for uncertainty and probabilistic thinking (Priority: 5/5): Poker taught Duke to accept incomplete information, base rates, luck, and variance, and to distinguish outcome quality from decision quality. Applying poker lessons to investing and forecasting (Priority: 5/5): She argues that investing, especially options and venture capital, is structurally similar to poker and can be improved by explicit probabilistic forecasts and feedback loops. Quit, kill criteria, and precommitment contracts (Priority: 5/5): Duke’s new book focuses on how to decide when to stop. She recommends writing kill criteria in advance so decisions aren’t hijacked by emotion, sunk cost, or fear of realizing losses. Bias, loss aversion, and identity traps (Priority: 5/5): The conversation explores sunk cost, sure-loss aversion, status quo bias, omission bias, and endowment/identity effects, showing why people cling to bad positions even when evidence changes. The value of outside voices and ‘quitting coaches’ (Priority: 4/5): Duke emphasizes seeking candid, trusted people who can challenge your beliefs and tell you what you don’t want to hear, because self-deception and social politeness block rational stopping.

Key Arguments: Every decision is probabilistic, even when it doesn’t feel like it; the future is always a forecast under uncertainty. People are much worse at processing losses and negative signals after they’ve already committed, so quitting decisions should be preplanned. Kill criteria work because they shift the decision from emotion-laden improvisation to an advance agreement about what evidence would justify stopping. Outcome quality and decision quality are different; a good call can lose, and a bad decision can win, so one should judge the process, not just the result. Investment mistakes often persist because of sure-loss aversion, sunk cost, status quo bias, and identity attachment to the original thesis or tribe. Out-of-consensus beliefs are harder to abandon because they become part of identity; people double down when challenged rather than update. The best way to improve judgment is to create explicit forecasts, measure them, and close feedback loops over time. Having a ‘quitting coach’ or trusted dissenter can counteract cheerleading, politeness, and self-justification. Poker and investing transfer well because both rely on base rates, variance, and statistical reasoning rather than analogies or intuition alone. The most important question after new information arrives is not whether the original thesis sounded reasonable, but whether you would buy the position today given current information.

Data Points: Poker winnings: More than $4 million - Duke won this amount over two decades as a professional poker player. Age at career pivot: 26 - She became ill and left her planned academic path around age 26. World Series of Poker bracelets (Eric Seidel): 9 - Mentioned as evidence of Seidel’s elite poker career. Estimated earnings (Eric Seidel): About $40 million - Used to illustrate Seidel’s long-term success as a player. Call-win probability: 81.5% - Duke describes a hand where her pocket jacks were an 81.5% favorite against two nines. Average player preflop play rate: 25% - Example of using base rates in poker forecasting. Typical best poker session length: 6 to 8 hours - Duke says she learned she played best only within this window before fatigue reduced decision quality. Forecasting improvement: Novices can become really good quickly - Referenced in her dissertation work with Phil Tetlock and Barb Mellers on novice forecasters. S&P/venture style example: Series A probability forecast - At First Round Capital, partners were asked to explicitly forecast the probability a seed company would raise a Series A. Sears retail share: 1% of U.S. GNP by 1950 - Used to show how a large, diversified company still failed due to identity and strategic rigidity. Alibaba loss example: Down 59% - William Green cites his own losing position as a live example of loss aversion and other biases. Cognitive state example: 12 miles into a marathon - Used to explain being cognitively ‘in the losses’ even when not literally underwater on a balance sheet. Marathon distance: 26.2 miles - Used to explain why people keep going even after injury because of the finish-line goal structure. Warning threshold example: One and a half points of inflation - Illustrative threshold Duke proposes for bitcoin if its inflation-hedge thesis stops working. Research sample size: 6,000 forecasts - Cited in a study showing analysts update differently when an out-of-consensus forecast is challenged. Vanta customers: More than 10,000 - From sponsor read; not central to the discussion but present in transcript. NetSuite users: Over 42,000 businesses - From sponsor read; not central to the discussion but present in transcript.

Pivotal Quotes: "You are thinking probabilistically." — Annie Duke: Her core rebuttal to the idea that only explicit probability work counts as probabilistic thinking. "Is there a question here?" — Eric Seidel: Seidel’s response to Duke’s complaint about a bad beat, emphasizing process over emotional venting. "I don't want to hear about it if there's not a question." — Eric Seidel: Seidel pushes Duke to focus on what she can learn rather than on the unfairness of variance.

Implications: Listeners should predefine stop rules, measure decisions by process, and seek candid dissent. In markets, this can reduce attachment to losing positions; in life, it can prevent identity and emotion from trapping people in bad paths.

🔓 Sign Up for Unlimited Episode Search

About We Study Billionaires

We interview and study famous financial billionaires, including Warren Buffett, Ray Dalio, and Howard Marks, and teach you what we learn and how you can apply their investment strategies in the stock market. We Study Billionaires is the largest stock investing podcast show in the world with 180,000,000+ downloads and is hosted by Stig Brodersen, Preston Pysh, William Green, Clay Finck, and Kyle Grieve. This podcast also includes the Richer Wiser Happier series hosted by best-selling author Wi...

View all episodes from We Study Billionaires