Know your edge?
Now size the bet.
The Kelly criterion turns an edge into a stake. Enter the odds you're getting and your true win probability — this returns the bankroll fraction that grows money fastest without ruining you.
01What Kelly actually solves
The Kelly criterion answers one question: given an edge, how much should you bet to grow your bankroll fastest over the long run? Bet too little and you leave growth on the table; bet too much and variance wipes you out before the edge plays out. Kelly is the mathematically optimal middle. The formula is f = (bp − q) / b, where b is the decimal profit per unit, p is your win probability, and q is 1 − p.
02The catch: it's only as good as your p
Kelly assumes you know your true win probability. You don't — you estimate it, and estimates are noisy. Feed Kelly an inflated probability and it tells you to bet huge, right into a loss. This is why almost no serious bettor runs full Kelly. The honest move is to be conservative with your probability input and then scale the result down.
03Why fractional Kelly is the real answer
Half-Kelly captures about three-quarters of the growth rate of full Kelly with roughly half the volatility — a dramatically better ride for a small cost in expected growth. Quarter-Kelly is gentler still and standard for anyone whose probability estimates are uncertain (which is everyone). Treat full Kelly as the ceiling, not the recommendation. The highlighted row below is the half-Kelly stake; size down to quarter if your read is shaky.
Kelly criterion FAQ
The Kelly criterion is a formula for sizing bets to grow a bankroll fastest over the long run. Given the odds and your true win probability, it returns the fraction of bankroll to wager. The formula is f = (bp − q) / b, where b is the decimal profit per unit, p is your win probability, and q is 1 − p.
Almost no serious bettor uses full Kelly. It assumes you know your win probability exactly, which you never do — you estimate it. Half-Kelly captures about three-quarters of the growth with roughly half the swings, a far smoother ride. Quarter-Kelly is gentler still and standard when your estimate is uncertain.
Treat full Kelly as the ceiling, not the recommendation.
Your honest estimate of how often the bet wins — not the book's number. The hard part is getting that estimate right. A good starting point is the fair, vig-free probability from the no-vig calculator, adjusted by whatever edge your own read gives you. Garbage in, garbage out: an inflated probability makes Kelly recommend a dangerously large bet.
If your win probability doesn't beat the price's implied probability, there's no positive expected value, and the Kelly fraction goes negative. The correct stake is then zero — pass. This calculator flags that case instead of showing a nonsensical negative bet, which is exactly the discipline that protects a bankroll.
No. Kelly only maximizes long-run growth if your probability estimates are accurate and you have a genuine edge. It's a sizing tool, not an edge-finding tool. If your reads are wrong, Kelly sizes losing bets efficiently. It works alongside good handicapping, not as a substitute for it.