The Kelly Criterion in Blackjack, Sports Betting, and the Stock Market
Read the paperopens doi.org in a new tab
What they found
Thorp, who used Kelly in blackjack and then ran one of the first quant hedge funds, wrote this as a practical guide. He walks through Kelly for simple bets, for blackjack with variable edge, for sports betting with many simultaneous bets, and for continuous investment in securities, including the formula for the optimal fraction when returns are approximately normal (edge divided by variance). He shows the drawdown properties of full Kelly (a full-Kelly bettor has a 50% chance of losing half their capital at some point), and argues for fractional Kelly in practice because true edges are uncertain.
See it drawn
Original diagrams for the ideas on this page. Illustrative, not real market data.
What you can use
- For a strategy with excess return m and variance s-squared, the Kelly fraction is roughly m divided by s-squared; this is the formula that translates Kelly to trading.
- Full Kelly implies a 1-in-2 chance of halving your account at some point; half Kelly reduces that to about 1-in-8.
- Overestimating your edge is the normal case, so treating half Kelly as the ceiling is a standard practitioner rule.
Caveats
Book chapter rather than a peer-reviewed study. The continuous-time approximation assumes normal returns, which understates fat-tailed risk. A free PDF circulates on academic archives and the author's site.
Tags: risk, position-sizing, kelly, practitioner
Summaries are our own reading of the paper, not the authors' words. Educational only, not advice. Discuss it in Book Club.