What the Kelly criterion is
For win probability p and payoff ratio b, the bet size that maximizes long-run compound growth is f* = p − (1 − p) ÷ b. With a 50% win rate and an average win twice the average loss (b = 2), f* = 0.5 − 0.5 ÷ 2 = 25%.
Reading the result
- Full Kelly is the theoretical maximum. Because win rate and payoff are estimates, traders commonly use half or quarter Kelly.
- If Kelly is zero or negative, there is no edge. The mathematical answer is not to bet less — it is not to bet.
- As the table shows, betting double Kelly can push long-run growth to zero or below even with a good edge.
Caveats
Kelly assumes independent trades and fixed probabilities. Real win rates and payoffs drift, and drawdown control matters more. This tool promises nothing; check the same inputs in the losing-streak simulator to see the risk side.
FAQ
Does 25% Kelly mean putting 25% of my account in one trade?
In theory it is the fraction to risk. For trades with a stop, people translate it to the share of equity lost if the stop is hit — and usually use far less.
How do I get the payoff ratio?
Average profit of winners ÷ average loss of losers from your records. The trading journal computes it for you.
Why half Kelly?
With estimation error, full Kelly over-bets and drawdowns balloon. Half Kelly gives up little growth while cutting volatility a lot.
Kelly is negative.
That win rate and payoff have no edge. Re-check inputs or the strategy.