Every investment decision is made before the outcome. The grading usually happens after, once the result has made the past look obvious.
Annie Duke spent two decades as a professional poker player after training as a cognitive psychologist, and Thinking in Bets (Portfolio, 2018) is her case against that habit. Poker is her laboratory: hidden information, shifting probabilities, real money, and immediate feedback that still doesn’t tell you whether you played well.
I read it as a portfolio manager who starts every position at the exit — what can I lose, what can I make, how do I get out. The book doesn’t touch any of that. What it adds is the discipline that comes before: say what you believe, how confident you are, and what would change your mind, in writing, before the market grades you.
The Book in Brief
Duke’s core claim is that outcomes reflect two things — decision quality and luck — and we routinely collapse them into one. She calls the error “resulting.”
Her opening example is Pete Carroll’s goal-line pass call at the end of Super Bowl XLIX. Malcolm Butler intercepted it, Seattle lost, and the call was declared the worst in Super Bowl history within minutes. Duke walks through the clock, the downs, and the historically tiny interception rate on that throw, and shows the decision was defensible even though the outcome was a disaster. Chess doesn’t work this way — it’s a game of complete information where the better player wins. Poker does, and so do markets.
From there the book covers beliefs (we accept information first and scrutinize it later, if ever; asking “wanna bet?” puts stakes on a claim and forces the scrutiny forward), learning from results (she calls it “fielding” — sorting outcomes into skill and luck, which we do badly because we claim our wins and disown our losses), and the group remedy: a small truth-seeking pod that rewards accuracy over agreement and runs on Robert Merton’s CUDOS norms — share the information, judge the claim rather than the source, disclose the conflicts, organize the skepticism.
The final chapter is the most practical. Temporal discounting — Seinfeld’s Night Guy leaving problems for Morning Guy — gets countered with mental time travel: the 10-10-10 question, the premortem, backcasting, and the Ulysses contract, named for Odysseus tying himself to the mast because he knew his future willpower wouldn’t hold.
Where It Earns Its Shelf Space
Two ideas justify the book for anyone managing money.
First, resulting is the dominant error in how investors evaluate themselves and everyone else. A strategy that takes many small, intentional losses while holding exposure to a large move will look broken for long stretches — and gets abandoned right before it pays. A strategy that collects small gains while carrying hidden tail risk looks brilliant for years — until one loss returns them. If you grade each outcome as it lands, you’ll quit the first and add to the second. Hit rate isn’t decision quality.
Second, precommitment works because it moves the decision to a moment when you can still think. Every rule I run — the exit set before entry, the maximum position size, the limit on total portfolio risk — is a Ulysses contract. The rule isn’t there for the calm days. It’s there for the day the position is down, the tape is ugly, and the version of me at that moment can’t be trusted to decide from scratch.
What the Book Leaves Out
Duke’s framework stops at probability, and probability is only half a bet. The other half is payoff.
A position that wins 70% of the time and pays half what it risks loses money. A position that wins 40% of the time and pays three times its risk makes money. Expected value — probability times payoff, summed across the outcomes — decides, and the book barely mentions it. Neither does it deal with price (a correct forecast the market already expects has no edge), position size (a positive-expectation bet sized too large can still ruin you), correlation (ten positions driven by one liquidity regime are one bet), or the difference between a drawdown you planned for and one that ends the game.
None of that is a criticism of the book she wrote. It’s a warning about the book she didn’t. Calibrated beliefs are an input to investing, not a system for it.
How I Apply It
The one practice from this book I’d made mandatory long before reading it is the ex ante record — the written version of the decision before the outcome exists.
Before I enter a position, the record already contains the exit price and how it was set, the amount at risk between entry and exit, the position size that keeps that risk to a set fraction of the portfolio, and what evidence would invalidate the idea. When the position closes, I review the record, not my memory of it — because memory edits the past to flatter the present. The question at review isn’t “did it make money.” It’s whether the plan executed, and whether the same plan, run across many similar decisions, produces a positive mathematical expectation.
That’s where poker and portfolio management genuinely meet. One hand proves nothing. One trade proves nothing. The evidence is the distribution — many comparable decisions, graded against what was written down before each one, across different market conditions. That’s what I built ASYMMETRY® to produce: not a prediction of any single outcome, but a repeatable process whose losses are defined before they happen and whose gains are free to run past them.
Verdict
Thinking in Bets teaches the mental hygiene of deciding under uncertainty, and teaches it well. What it can’t teach is the mechanics that turn a calibrated belief into a good investment: price, payoff, size, timing, and staying solvent along the path. Read it for the first. Don’t mistake it for the second.