Most traders obsess over what to trade and almost never ask the question that actually compounds an account over time: how much to put on each trade. The Kelly Criterion is one of the few position-sizing formulas with a real mathematical claim behind it — it identifies the bet size that maximizes the long-run growth rate of your capital. Here is what it is, how to compute it from numbers you already have, and why almost every professional who uses it deliberately bets less than it says.
What the Kelly Criterion actually is
The Kelly Criterion, published by Bell Labs researcher John Kelly in 1956, answers a single question: given an edge, what fraction of your bankroll should you risk to grow it fastest without going broke? Bet too little and you leave growth on the table. Bet too much and volatility — and eventually ruin — catches up with you. Kelly finds the balance point.
The formula
For a trade with a fixed reward-to-risk ratio, the Kelly fraction is:
f* = W − (1 − W) / R
where:
- W = your win rate (probability a trade wins), expressed as a decimal
- R = your payoff ratio (average win size ÷ average loss size)
A worked example
Say your trading journal shows you win 45% of the time (W = 0.45), and your average winner is 1.8× the size of your average loser (R = 1.8):
f* = 0.45 − (0.55 / 1.8) = 0.45 − 0.306 = 0.144
Kelly says risk about 14% of your account on that trade. To most experienced traders, that number should look terrifyingly large — which brings us to the most important part of this article.
Why you should almost never bet "full Kelly"
Full Kelly maximizes growth in theory, assuming your win rate and payoff ratio are known exactly and never change. In real trading, both are estimates pulled from a limited sample, and markets shift regimes. If your true edge is even slightly worse than you think, full Kelly overbets and the drawdowns become brutal — a full-Kelly strategy can routinely see 50%+ peak-to-trough declines.
That is why practitioners use fractional Kelly — typically one-quarter to one-half of the number the formula spits out. Half-Kelly captures roughly three-quarters of the growth rate with about half the volatility. In the example above, a half-Kelly trader would risk ~7% and a quarter-Kelly trader ~3.6% — much closer to the 1–2% risk rules most risk managers recommend.
Where traders get it wrong
- Garbage in, garbage out. Kelly is only as good as your win rate and payoff inputs. Estimate them from a real, statistically meaningful sample of trades — not a hopeful guess.
- A negative result means no edge. If f* comes out zero or negative, the math is telling you this setup has no positive expectancy. The correct size is zero.
- Edges decay. Recompute as your data grows and your strategy evolves; last quarter's win rate may not be this quarter's.
How to use it at TEG
Rather than run the arithmetic by hand, plug your numbers into TEG's free Kelly quick check, or use the full risk tools to compare full and fractional Kelly side by side. Pair it with your trading journal so the win rate and payoff ratio you feed in come from your actual results, not from memory.
The bottom line
The Kelly Criterion is a discipline tool, not a magic number. Its real value is forcing you to translate a vague sense of "I'm pretty good at this" into concrete probabilities — and then reminding you, mathematically, to bet a fraction of what your confidence suggests. Used with fractional sizing and honest inputs, it keeps you in the game long enough for a genuine edge to compound.
Educational content only. Nothing here is financial advice or a recommendation to make any particular trade. Position sizing does not eliminate risk — trading involves substantial risk of loss, and you can lose more than you expect. Past performance and historical win rates do not guarantee future results.