What former poker champion turned investing coach Annie Duke says most of us get wrong about risk
Annie Duke spent nearly two decades making high-stakes decisions with incomplete information at the poker table. Today she coaches investors and executives on decision strategy. Across both domains, her message is consistent: most people misunderstand risk. We obsess over outcomes, equate risk with being wrong, and cling to failing bets because quitting feels like losing. Duke argues that better results come from reframing risk as a property of decisions under uncertainty, not as a moral scorecard on outcomes you can’t fully control.
Here are the core errors she sees—and how to correct them.
We judge risk by outcomes instead of by decision quality
Duke popularized the term “resulting”: the habit of deciding whether a choice was good or bad based on how it turned out. Win a hand with a shaky call? People think it was smart. Lose with the right read? People think it was foolish. In markets, a trade that made money can still have been reckless; a loss can reflect sound process in a bad-luck run.
Correction: Separate the quality of the decision process from the outcome. Ask what you knew at the time, which alternatives you considered, how you sized the position, and whether your thesis and risk plan were explicit. Keep a decision journal so you can audit process independent of P&L.
We confuse probability with payoff
Most of us treat risk as the chance of being wrong. Duke says that’s only half the story. What matters is the combination of likelihood and magnitude—the expected value. A 30% probability of a big gain can be a great bet; a 90% probability of a tiny gain paired with catastrophic downside can be terrible.
Correction: Frame every decision as a bet. Map scenarios, attach rough probabilities, and quantify payoffs. Favor positive expected value and positive asymmetry (limited loss, large potential gain). Avoid negative asymmetry even when the win rate looks high.
We ignore base rates and over-trust stories
Poker players who believe they’re “running bad” and investors who fall in love with a narrative both neglect base rates—the average frequencies of events in the relevant reference class. That inside-view bias pushes us to overestimate how special our situation is.
Correction: Start with the outside view. Before diving into specifics, ask: In the reference class for this decision—startups of this type, drug trials at this phase, turnarounds in this industry—what usually happens? Use that base rate as your prior and adjust, not the other way around.
We mistake uncertainty for risk and overpay for certainty
We crave certainty that doesn’t exist. Markets, like poker, are games of incomplete information. Treating uncertainty as a problem to eliminate leads to analysis paralysis, overfitting, and expensive “certainty theater.”
Correction: Aim for calibrated confidence, not certainty. Express beliefs as ranges. Update in small increments as new information arrives. Make small, reversible bets when information is scarce, and scale only as conviction and evidence improve.
We use the wrong yardstick: volatility instead of ruin and goals
Volatility feels risky because it hurts in the short term. But for many investors, the true risk is failing to meet long-term goals or suffering irreversible loss. Volatility can be tolerable if drawdowns don’t force you to liquidate at the worst time.
Correction: Define risk in relation to your bankroll, time horizon, liquidity needs, and risk of ruin. Size positions so adverse runs don’t push you out of the game. Judge risk by the chance of not meeting objectives, not just by price squiggles.
We forget that inaction is a bet
Doing nothing can feel safe. Duke points out that the status quo is itself a bet—with opportunity costs that compound. Folding every hand in poker keeps you from losing chips in a pot, but blinds and antes will still drain your stack.
Correction: Always compare choices to the best available alternative, including redeploying capital. Ask, “If I didn’t own this today, would I buy it at this price?” If not, inaction is likely the higher-risk move.
We hang on too long because quitting looks like failing
Humans escalate commitment to bad paths due to sunk costs, identity, and fear of regret. In both poker and investing, this shows up as doubling down on a weak hand or “marrying” a thesis. Duke argues that quitting is a crucial skill: it preserves option value and reclaims resources for better bets.
Correction: Precommit to kill criteria before you start—a set of objective conditions that will trigger an exit. Make them thesis-based (lost a key customer, failed a clinical endpoint, broke a covenant), not just price-based. Treat quitting as a success if it moves resources to higher expected value opportunities.
We sample too little and learn the wrong lessons
Short runs are noisy. A good process can lose repeatedly; a bad process can look brilliant for a while. Drawing broad conclusions from a handful of outcomes cements overconfidence or unwarranted pessimism.
Correction: Expand your sample size by making more small bets and by aggregating decisions across comparable situations. Review outcomes on a cadence (quarterly, annually) that matches the variance of the process you’re evaluating.
We underweight correlation and bankroll management
A set of individually attractive bets can still create portfolio-level fragility if they’re correlated. In poker this is bankroll management; in investing it’s sizing and diversification.
Correction: Size positions to survive. Consider correlations and tail dependencies, not just standalone expected value. A slightly lower-return portfolio you can stick with beats a fragile one that forces capitulation.
We skip decision hygiene
Under time pressure, groupthink, and incentives, even smart people drift into narrative and hindsight. Duke advocates for lightweight process guardrails.
Correction: Build simple habits:
– Write a one-page pre-commitment before entering a position: thesis, key assumptions, base rates, scenarios, probabilities, expected value, alternative uses of capital, and kill criteria.
– Run a premortem: “It’s 18 months from now and this investment failed. What most likely happened?”
– Appoint a red team to challenge the thesis.
– Track forecasts and score them (e.g., Brier scores) to calibrate over time.
– Conduct post-mortems that grade process, not just outcome.
How this looks in practice
Suppose you’re considering a biotech stock ahead of a Phase 3 readout.
– Base rate: Historical approval probability for similar trials is around 50% post–Phase 3.
– Scenarios and payoffs: If successful, you estimate a 200–300% upside; if it fails, a 60–80% downside.
– Expected value: A rough calculation suggests positive asymmetry, but the variance is huge.
– Bankroll and correlation: You have other healthcare bets; correlation is high. You cap the position size accordingly.
– Kill criteria: Exit if the readout is delayed twice, if the company raises dilutive capital beyond X%, or if a competitor releases superior data.
– Updates: As interim data and FDA communications arrive, you update probabilities incrementally rather than swinging from “bullish” to “bearish.”
– Process review: Regardless of outcome, you grade the decision against your pre-committed plan.
A compact toolkit for thinking about risk the Annie Duke way
– Treat every decision as a bet with odds and payoffs.
– Start with base rates; then adjust with inside information.
– Write it down: thesis, assumptions, scenarios, probabilities, expected value, alternatives.
– Set kill criteria before you commit.
– Size for survival; consider correlation and liquidity.
– Make small, reversible bets while you learn; scale only as evidence strengthens.
– Judge decisions by process quality, not single outcomes.
– Run premortems and red-team challenges.
– Track and score your forecasts to calibrate.
– Remember that not changing course is also a bet.
The deeper point
Risk isn’t the enemy; it’s the raw material of progress. The enemy is self-deception—about how sure we are, how much we can lose, and how sticky our identities make our bad bets. Annie Duke’s contribution is to offer a language and a set of habits for staying honest in the fog of uncertainty. Think in bets, prize flexibility, and measure yourself by the quality of your decisions. Do that consistently, and the outcomes will take care of themselves over a long enough run.
