Kelly criterion

The Kelly criterion tells you the fraction of your capital to risk on a bet so that your account compounds at the fastest possible long-run rate. It is a growth-optimal rule, not a comfort-optimal one: it assumes you want to maximize the geometric return over many trades and that you know your edge exactly. Bet more than Kelly and your growth rate falls while your risk climbs; bet less and you trade slower growth for a smoother ride. In practice you almost never bet full Kelly, but the number is the anchor every serious sizing decision leans on.

How it's calculated

For a two-outcome trade — you either win or lose — the Kelly fraction is:

Say you win 45% of the time and your average winner is 2x your average loser. Then W = 0.45, R = 2, and:

f* = 0.45 − (0.55 / 2) = 0.45 − 0.275 = 0.175

Kelly says risk 17.5% of your capital on each trade. That is a huge number — and that is exactly the problem, which is why almost nobody trades it raw.

The standard fix is half-Kelly: bet half the computed fraction, so 8.75% here. Half-Kelly gives up only about 25% of the theoretical growth rate but roughly halves the volatility of your equity curve and dramatically shrinks your drawdowns. Many traders go further and run quarter-Kelly. The math rewards restraint: growth rate is flat near the peak, so shading down costs little, while overbetting past the peak destroys capital fast.

Why it matters

Full Kelly is optimal only if your inputs are exact. They never are. Your win rate and payoff ratio are estimates pulled from a finite sample, and even a modest overestimate of your edge pushes the "optimal" fraction well past the true peak — into the region where growth turns negative. Kelly is unforgiving on the high side: the penalty for betting too much is far worse than the penalty for betting too little.

Full Kelly also produces drawdowns most humans cannot sit through. A full-Kelly bettor should expect to see their account cut in half from time to time purely as normal variance. Combine fat drawdowns with the fact that a losing streak on a 45%-win strategy is routine, and you get a sizing scheme that is mathematically elegant and psychologically ruinous.

That is why the practical rule is: compute Kelly, then bet a fraction of it. Fractional Kelly buys you a margin of safety against estimation error and keeps your equity curve in territory you can actually hold through. If your fraction still leaves a realistic path to zero, you are effectively overbetting — study your risk of ruin before you size up.

Related concepts

Kelly is one answer to the broader question of position sizing: how much to put on each trade. It depends directly on your win rate and payoff ratio, the same inputs that drive your trading expectancy — a positive expectancy is the precondition for Kelly to return anything above zero. To turn a Kelly fraction into an actual order, run the numbers through the position size calculator, then track how your realized win rate and payoff ratio drift over time on your Tradeways dashboards so your sizing stays anchored to your real edge, not the one you assumed.