The Kelly criterion is a formula for sizing a bet as a fraction of your bankroll, based on your edge and the odds, to maximise long term growth.
Last reviewed 13 October 2025 · Educational, not advice
The Kelly criterion is a rule for deciding how much of your bankroll to stake on a bet when you believe you have an edge. It was described by John Kelly at Bell Labs in 1956 and chooses the stake that maximises the long term growth rate of your bankroll, rather than the expected value of any single bet. In words, it tells you to bet more when your edge is larger and the odds are more generous, and less when they are not.
The common form of the formula is f equals b times p minus q, all divided by b, where f is the fraction of your bankroll to stake, b is the profit received per unit staked at the odds on offer, p is your estimated probability of winning, and q is one minus p. If the formula gives a number at or below zero, it is telling you the bet has no edge and you should not stake anything.
The strength of Kelly is that it balances growth against the risk of ruin. Betting more than the Kelly fraction raises your expected growth only up to a point and then increases the chance of a damaging drawdown, while betting far above it can destroy a bankroll even when each bet has an edge. Kelly sits at the level that grows wealth fastest over the long run without staking so much that variance wipes you out.
In practice most people use a fraction of the full Kelly stake, often a quarter or a half. The full formula assumes your probability estimate is exactly right, which it rarely is. Since prediction market edges are estimated and uncertain, betting a fraction of Kelly reduces the damage from an overconfident estimate and smooths the ride, at the cost of slightly slower growth.
Kelly is only as good as the numbers you feed it. If your probability estimate is wrong, the formula will size a bad bet confidently, so the honest input is a realistic edge, not a hopeful one. It also assumes you can reuse your bankroll across many independent bets, which is why it suits repeated trading more than a single wager. None of this removes the risk of loss.
Suppose you estimate a 55 percent chance of winning a bet that pays one unit of profit per unit staked, so b is 1, p is 0.55, and q is 0.45. Kelly suggests f equals 1 times 0.55 minus 0.45, divided by 1, which is 0.10, or ten percent of your bankroll. A half Kelly user would stake five percent.
Illustrative only. Numbers are examples, not a quote or a prediction.
Kelly assumes your probability estimate is correct, and a wrong estimate sizes a bad bet with confidence. It reduces the risk of ruin but never removes the risk of loss, which is why many use a fraction of the full stake. Stake only what you can afford to lose, never to chase a loss, and never on borrowed money. If it stops feeling like a free choice, step back. In the United States you can call or text the helpline on 1-800-GAMBLER or visit ncpgambling.org.
A formula that sets your stake as a fraction of your bankroll, based on your edge and the odds, to maximise the long term growth of your bankroll.
A common form is f equals b times p minus q, all divided by b, where f is the fraction to stake, b is the profit per unit at the odds, p is your win probability, and q is one minus p.
Because the formula assumes your probability estimate is exactly right. Betting a quarter or half of the Kelly stake cushions against estimate errors and reduces swings.
The rules change fast. Get the changes that affect you, plain and current, not tips.
Independent. Every claim dated and sourced. No platform pays for its place.