Work Out a Kelly Stake and See What It Risks
Type in an edge and Kelly hands you a stake instantly. It never shows what that number costs if your edge is wrong.
This tool turns your edge into a stake, then tests what happens if you are wrong about it
Enter a bankroll, the odds and the probability you think a bet will win, and the calculator below returns the full Kelly stake alongside what Half, Quarter and Tenth Kelly would risk instead. A Monte Carlo simulation then runs your inputs hundreds of times to chart a spread of outcomes, not one lucky run, and reports the odds of your bankroll dropping under half its starting value along the way. The probability field is the weak point in all of this; see below for what that number cannot know.
How this Kelly criterion calculator reads your bet
Fill in the fields below with a real bet you are weighing, or leave the demo numbers loaded, and every readout updates as you type. If limits are the reason you ended up here rather than Kelly itself, why bookies limit you in the first place covers that wider question; this page assumes you already know it and just want a stake sized properly for whatever room you still have.
Read the edge and the risk-of-ruin readout before you look at the stake itself. A low reading is the one that earns the stake Kelly just proposed; a high reading is the one that says to take less than it proposed, whatever the number on screen looks like.
Half Kelly grows slower. That is the whole point.
Full Kelly maximizes long-run growth in theory, and it does that by accepting swings most bettors would find unbearable in practice. Move to Half Kelly and the readouts above keep roughly three quarters of that growth rate while cutting the deepest drawdowns hard. Quarter Kelly gives up more of the upside again, closer to half the growth rate of full Kelly, in exchange for a chance of falling below half your starting bankroll that is often close to zero over a normal run of bets.
The reason experienced bettors settle on a quarter of the full Kelly stake instead of the whole thing is not caution for its own sake, but a structural response to a number that amplifies whatever error sits inside your probability estimate the moment you use more of it, which is a mechanism the simulation above makes visible rather than assumed. Kelly does not correct a wrong estimate. It has no idea.
A second figure keeps turning up in the same practitioner writing: a cap of around 5% of bankroll on any single stake, whatever the formula above returns, because normal variance can still drain an account fast even on a bet with a genuine edge. the win rate math that decides whether betting pays at all is worth reading first if the break-even side of the question worries you more than the staking side.
A few signs your probability estimate might be more confident than it should be:
- It came from your own model, with no outside line to check it against.
- The market barely moved once it opened.
- You are a few bets into a run that has gone your way.
- Nobody else appears to price this outcome the way you do.
None of that means the bet is wrong. It does mean a lower fraction costs you less if it turns out to be. This is a staking tool for adults aged 18 or over, and the honest use of it is to bet less when the risk-of-ruin number looks uncomfortable, never more.
What this number cannot know about you
Kelly is only ever as good as the two numbers you hand it. It has no idea whether the edge you typed in is real, and it has no idea whether this bet's result depends on the last one you placed, or the next one you are about to place on a related market. Back three different markets on the same match and Kelly, run separately on each, will happily size all three as if they were unconnected coin flips; a bad afternoon in one can still drag the other two down with it. testing a price against a sharp market is one way to build a probability estimate worth trusting before this calculator ever sees it.
If you are weighing whether to treat betting as a serious, repeatable activity rather than a hobby, what to prove to yourself first covers ground this page does not, including the separate question of whether a bookmaker will even let you keep betting your edge. Bookmakers that do let winners keep betting are not universal, either; which ones actually accept winning players is the place to check that once the sizing question above is settled. None of it changes what the calculator returns. It only changes how much you should trust the number you fed it.
Frequently asked questions
What is the Kelly criterion formula in sports betting?
Written out, full Kelly is f* = (bp - q) / b, where b is the decimal odds minus 1, p is the probability you think the bet wins, and q is 1 minus p. The calculator above runs that same formula every time you change a field, so there is little reason to do the arithmetic by hand unless you want to check it.
Why is full Kelly risky even when my edge is real?
A genuine edge does not make the maths gentler. Full Kelly stakes exactly the fraction that maximizes long-run growth, which also sits close to the largest fraction that still avoids ruin, so a short losing run, entirely normal even with a real edge, produces a drawdown most bettors would not sit through. Switch the fraction above to Full Kelly and watch the 10th percentile line and the risk-of-ruin readout both move.
What is fractional Kelly, and why do serious bettors use half or quarter Kelly instead of full Kelly?
Fractional Kelly means staking a fixed share, commonly a half, a quarter or a tenth, of whatever the full Kelly formula returns. Why serious bettors favor the smaller fractions is covered above; switch the toggle to see what each one costs in growth and saves in risk.
What happens to my results if my estimated probability is wrong?
The gap compounds instead of staying fixed. Fill in a lower number in the real probability field than in your estimated one and rerun the simulation: the stake itself does not change, since Kelly still sizes the bet on your estimate, but the trajectory chart and the risk-of-ruin readout shift to show what that optimism actually costs.
What does Kelly not know about my bet?
Less than it lets on: see what this number cannot know about you above for the two blind spots behind every stake this calculator returns.