The Kelly criterion answers “how much?”, not “what?”. Give it your honest odds and it returns the fraction of your money that compounds fastest without blowing up. Two versions below: a single position held against cash, and a two-fund mix that seesaws. This is arithmetic, not a recommendation — it will size you into trouble just as confidently if your inputs are wrong.
You put a fraction of your money into one position; the rest sits in cash. Each round it either gains or loses.
Growth is the long-run compounding rate per round, in log terms — it is what builds over many rounds. Above the Kelly size the growth rate falls and the swings grow, which is why “more confidence” never means “more than full Kelly”.
Funds A and B move in opposite directions: when A wins, B loses, and the other way round. Put w in A and the rest in B. If A wins with probability p it gains x while B loses y; if B wins it is the reverse.
If the Kelly weight comes out above 100%, the model wants leverage; without it the answer is simply “all in A”. If it comes out negative, Kelly says hold no A at all. See the Kelly guide for where these come from.
The output is a fraction of your investable money — not a fund pick and not a return forecast. “Half Kelly” (or a quarter) is what most people actually use: it keeps most of the compounding benefit and leaves room for the day your probability estimate is too high.
Kelly assumes you know the true probabilities and that the same bet repeats independently. Neither is true in markets: probabilities are estimates, and losses cluster in crashes rather than arriving as tidy ±20% flips. The calculator is a way to feel the sizing trade-off — that under-betting wastes an edge and over-betting destroys growth — not a dial to set your portfolio from.
The fraction of your money that maximises long-run compounding, given your win chance and the size of a win versus a loss. It also shows half and quarter Kelly, and the growth rate at the size you choose.
Because the edge is large relative to the swings. Above 100%, Kelly is proposing leverage; without leverage the practical answer is simply to hold the full position.
Rarely. The full number assumes your probabilities are exactly right and that the bet repeats cleanly. Most practitioners scale it down — half or a quarter — to allow for estimation error and bigger-than-modelled losses.
No. It is plain arithmetic meant to make an idea click. It does not know your circumstances and it does not predict any fund.
More: Guide: what the Kelly criterion is · Rebalancing calculator · Drawdown recovery calculator · Annualized vs cumulative · All tools