Kelly Criterion
Also known as: kelly formula, kelly bet sizing, optimal f
What is it?
The Kelly criterion is a formula for deciding how much of your capital to risk on a trade so your account grows as fast as possible over the long run without going broke. It takes your edge - your win rate and your reward-to-risk - and returns the fraction of capital that mathematically maximizes long-term growth.
For a strategy that wins 50% of the time at a 2:1 reward-to-risk, full Kelly suggests risking 25% of the account per trade. That is far more aggressive than most traders can stomach, because the same math that maximizes growth also produces brutal drawdowns along the way.
In practice, traders use a fraction of the figure - 'half Kelly' or less - to keep most of the growth while cutting the swings. Kelly also punishes overestimating your edge: feed it an inflated win rate and it tells you to bet too big, which is how over-betting turns a winning system into a losing one.
Why it matters: Kelly defines the mathematically optimal bet size - and shows over-betting it, not under-betting, is what blows up accounts, so most pros trade a fraction.
Kelly % = W - (1 - W) / R (W = win rate, R = reward-to-risk)
Position size compounds every trade, so betting above the Kelly fraction is a leading cause of account blow-ups even with a winning strategy.
Real-world example
A 55% win rate at 1:1 reward-to-risk gives a 10% full-Kelly bet; a trader using half-Kelly risks 5% per trade, accepting slower growth for far shallower drawdowns.
How SignalBots handles it
SignalBots exposes each feed's historical win rate and reward-to-risk - the two inputs Kelly needs - so you can size positions deliberately instead of guessing; past results never guarantee future ones. See /risk-warning.
Pro tip
Use half-Kelly or less and base the inputs on a long, out-of-sample track record - a short winning streak inflates your apparent edge and Kelly with it.
Common pitfalls
Plugging an optimistic win rate into full Kelly and over-betting, so a normal losing streak causes a far deeper drawdown than expected.
Frequently asked questions
Should I use full Kelly?
Most traders should not - full Kelly maximizes growth but produces drawdowns few can tolerate. Half Kelly or less keeps much of the benefit with far smaller swings. Every position size still carries risk of loss.
What does the Kelly formula need?
Two inputs: your win rate and your reward-to-risk ratio. Both should come from a long, realistic track record, because small or cherry-picked samples overstate your edge.
Why does over-betting blow up accounts?
Past the Kelly fraction, extra size adds more risk than growth, so a normal losing streak compounds into a deep, sometimes unrecoverable drawdown. Under-betting only slows growth; over-betting threatens survival.
Can Kelly tell me to risk nothing?
Yes - if the formula returns zero or a negative number, your inputs imply no edge, meaning the strategy is not worth trading at that win rate and reward-to-risk.
Is Kelly the same as a fixed-percent risk rule?
No. A fixed-percent rule risks the same share every trade; Kelly derives the share from your edge. Many traders combine them, capping Kelly's output with a fixed maximum such as 1-2%.
Trading involves substantial risk of loss. Historical and backtested results do not guarantee future performance. Read the full risk warning.