About the Kelly Criterion Calculator
The Kelly criterion is a formula for the share of your bankroll to risk on a repeated bet when you know your edge. Enter the chance of winning, how much you gain per unit risked when you win and how much you lose when you lose. The calculator returns the full Kelly fraction, a reduced version such as half Kelly, the edge per unit and the expected growth per bet. If the edge is zero or negative it tells you plainly that the answer is not to bet. The numbers are only as good as your estimates, so the tool is for learning and is not advice.
Updated October 2026
How to use the Kelly Criterion Calculator
- 1Type your estimated chance of winning as a percentage.
- 2Enter the gain when you win, as a percentage of what you risk (100 for an even-money bet).
- 3Enter the loss when you lose, usually 100.
- 4Pick full, half or quarter Kelly.
- 5Optionally enter your bankroll to see an amount.
Why a fraction and not everything
If you have an edge, betting your whole bankroll every time gives the biggest average profit but wipes you out on the first loss. Betting too little grows slowly. Kelly finds the balance in between by maximising the expected logarithm of your wealth, which is the same as maximising how fast the money compounds over many bets.
The growth curve is flat near the top and falls steeply on the right. That is why people choose half Kelly: you keep around three quarters of the growth with much smaller swings, and being off on the probability by a few points costs far less.
Examples with real numbers
- A 60 percent coin with even money: full Kelly is 20 percent of your bankroll, half Kelly is 10 percent.
- A 40 percent chance at even money has a negative edge, so Kelly says 0.
- A 30 percent chance paying 3 to 1 (gain 300, loss 100) gives a Kelly fraction of 0.3 minus 0.7 divided by 3, which is about 6.7 percent.
- A trade that wins 55 percent of the time, makes 2 and loses 1 per unit risked suggests a stake of about 32 percent, a sign you should halve or quarter it.
Related tools
Turn the stake into a quantity with the Crypto Position Size Calculator, or study the odds first with the Odds Converter. The Expected Value Calculator shows whether a bet is worth taking at all, and the Sharpe Ratio Calculator measures how a series of past results balanced reward and risk. Remember that Kelly is a formula about numbers you type in, not a promise about the future.
Features
- Full, half and quarter Kelly
- Works for bets and for trades with different gain and loss sizes
- Edge and expected growth shown
- Amount to risk from your bankroll
- Warns when there is no edge
- Free and private
Frequently asked questions
What is the Kelly criterion?
It is a formula that picks the stake which maximises the long-run growth rate of your money for a repeated bet with a known edge. It was published by John Kelly in 1956.
What is the formula?
The stake fraction is p divided by a, minus q divided by b, where p is the chance of winning, q is 1 minus p, b is the gain per unit risked and a is the loss per unit risked. For an even-money bet it reduces to 2p minus 1.
Why use half Kelly?
Full Kelly assumes your estimates are exactly right and produces large swings. Half Kelly gives up some growth but cuts the drawdowns, and protects you when you have overestimated your edge.
What if the result is negative?
A negative result means you have no edge at those numbers, and the calculator shows 0 percent. Betting more than Kelly can even turn a good bet into a losing strategy.
Can I use it for stocks or crypto?
You can, by using the average win and average loss of your trades, but outcomes are not fixed like a coin flip, so the result is only a rough guide. Many traders keep far below the full figure.
What are b and a for odds of 3 to 1?
A bet that pays 3 to 1 gains 300 percent of the stake when it wins, so type 300 as the gain. The loss is 100 percent of the stake.
Is the result financial advice?
No. It is a mathematical tool based on numbers you provide. Real bets rarely have known probabilities, so use it to learn, not to decide.