Kelly criterion calculator
Enter your estimate of the chance to win, the odds on offer and your bankroll. The Kelly criterion gives the share of the bankroll that maximises long-run growth; most people bet a fraction of it, such as half Kelly, to reduce swings.
Runs in your browser: nothing you enter is sent anywhere.
How to use it
- Enter your estimated probability of winning, in percent.
- Enter the odds, decimal or American.
- Enter your bankroll and the fraction of Kelly you want to bet (100% is full Kelly, 50% half Kelly).
How it works
With decimal odds d, the profit per 1 staked is b = d − 1. With win probability p and q = 1 − p:
- Full Kelly f* = p − q ÷ b.
- Stake = bankroll × f* × fraction, or 0 when f* is 0 or less.
- Edge = p × d − 1: the expected profit for each 1 staked.
American odds are turned into decimal first: +A → 1 + A ÷ 100, −A → 1 + 100 ÷ A.
Worked example
You think a team wins 60% of the time and the decimal odds are 2.00, so b = 1. Full Kelly = 0.60 − 0.40 ÷ 1 = 20% of the bankroll. At half Kelly with a bankroll of 1,000 you stake 100. The edge is 0.60 × 2 − 1 = 20% per 1 staked.
At 55% and −110 (decimal 1.9091, b = 0.9091), full Kelly is 0.55 − 0.45 ÷ 0.9091 = 5.5%, and the edge is 5%.
Tips
- Small errors in p change the stake a lot. Be conservative with your estimate.
- Recalculate the stake from your current bankroll before each bet.
- Kelly is a growth rule, not a guarantee: losing streaks still happen. Not financial advice.
Frequently asked questions
What is the Kelly formula?
f* = p − q ÷ b, where p is the chance to win, q = 1 − p the chance to lose and b the profit per 1 staked (decimal odds − 1). f* is the share of the bankroll to bet.
Why bet half Kelly instead of full Kelly?
Full Kelly assumes your win probability is exactly right and leads to deep drawdowns along the way. Betting a fraction gives up some growth for much smaller swings and protects you when your estimate is too optimistic.
What does a negative Kelly result mean?
The odds pay too little for your chance to win, so the bet has a negative expected value. The calculator then shows a stake of 0.
Where does the win probability come from?
From you: a model, statistics or your judgement. The calculator cannot know it. The implied probability of the odds is not your edge; Kelly only says to bet when your probability is higher than that.
Is this for trading too?
The same idea is used for position sizing, but trades rarely have fixed odds and known probabilities. Treat any Kelly number there as an upper limit, not a target.