What the tool is for. Estimate fractional Kelly position size and compare it with a fixed-risk approach. Both inputs are estimates you supply, so the output inherits whatever optimism went into your win rate and payoff ratio.
Kelly maximises growth, not comfort.
Kelly math is useful because it is uncomfortable. It often tells traders that their confident position size is larger than the edge can justify. What it does not tell you, and what the single output below can easily obscure, is what it was optimising for in the first place.
The formula maximises the long-run growth rate of capital. It does not minimise drawdown, and it has no opinion about whether you can psychologically survive the path. Betting the full fraction is, by construction, the most aggressive size that does not eventually ruin you across infinite repetitions — and "does not eventually ruin you" leaves a great deal of room for losing most of the account on the way. That is the honest reason experienced traders use a fraction of it. The growth curve is flat near its peak and drops away steeply past it, so cutting the fraction in half costs surprisingly little expected growth while cutting the swings substantially. Overshooting the optimum, by contrast, is punished hard.
Then there is the input problem, which in crypto is severe. Kelly assumes a binary bet with a known probability and a known payoff. A real trade has a distribution: stops slip, targets fill partially, a gap can produce a loss larger than the one you defined. The payoff ratio you type in is already an average that hides its own tail, and the win rate comes from a sample that was measured in a market regime that may no longer exist. Both inputs are estimates, both are usually optimistic, and the formula amplifies optimism rather than correcting it.
Lower your win rate by five percentage points and leave everything else alone. If the recommended fraction collapses, your edge was never large enough to bet at full Kelly — because a five-point estimation error is entirely ordinary when the sample is a few dozen trades.
Use referral code BN16188 for up to 20%* off trading fees. Binance availability and product access vary by jurisdiction. CoinView may receive affiliate compensation, and this page is not an official Binance property.
where p = win rate and q = 1 − p
The fraction is of your equity now, not of your original stake
Kelly is a proportional rule, so every percentage here applies to the account as it stands today: the correct size shrinks after a loss and grows after a win. That makes the dollar figure valid only until the next trade settles. Working out the fraction once, converting it to a fixed amount and then betting that amount trade after trade is a different strategy wearing the same name — and the fixed version has lost the property that made the fraction defensible, because it no longer scales down while the account is falling.
Cross-check the inputs before trusting the fraction
Both inputs deserve to be re-derived from filled trades rather than from the plan: the payoff ratio should use the prices you actually got after fees and slippage, and the win rate should come from a sample whose size you can state out loud. If you cannot say how many trades produced it, the output is a guess wearing the clothes of arithmetic. Then hold the recommended fraction up against a drawdown you have genuinely lived through rather than one you have imagined, because the formula optimises the growth rate and carries no term for the path it takes. And when the estimates come from a regime that has since changed, the fraction is neither aggressive nor conservative — it is describing a market that no longer exists.
Risk note. Crypto assets are volatile and not suitable for every investor. This page is editorial analysis, not financial advice.