The Kelly criterion and why nobody should use it fully
Posted: Fri Oct 30, 2026 1:37 pm
Kelly is a formula that gives the fraction of capital to risk on each trade to maximise long-run growth, using win probability and payoff ratio.
Applied to a strategy with a 55 percent win rate and 1:1 payoff, Kelly suggests risking 10 percent per trade. The maths is correct. The practice is risky.
The formula assumes the inputs are known exactly. In reality, win rate and payoff are estimates from limited data. Small errors in the estimate lead to oversizing, and oversizing under Kelly produces brutal drawdowns, often 50 percent or more.
Professionals commonly use a fraction of Kelly, such as one-quarter or one-half, trading some growth for much smaller drawdowns.
For retail scalpers, a simpler fixed fraction of 0.5 to 1 percent usually sits well below any Kelly estimate and leaves room for estimation errors.
The principle is useful as a ceiling, not a target.
How does the risk per trade here compare with a conservative Kelly estimate for the strategy?
Applied to a strategy with a 55 percent win rate and 1:1 payoff, Kelly suggests risking 10 percent per trade. The maths is correct. The practice is risky.
The formula assumes the inputs are known exactly. In reality, win rate and payoff are estimates from limited data. Small errors in the estimate lead to oversizing, and oversizing under Kelly produces brutal drawdowns, often 50 percent or more.
Professionals commonly use a fraction of Kelly, such as one-quarter or one-half, trading some growth for much smaller drawdowns.
For retail scalpers, a simpler fixed fraction of 0.5 to 1 percent usually sits well below any Kelly estimate and leaves room for estimation errors.
The principle is useful as a ceiling, not a target.
How does the risk per trade here compare with a conservative Kelly estimate for the strategy?