Kelly stake
calculator.
Enter your bankroll, your estimated win probability, and the offered odds. Get the mathematically optimal stake size: full, half, or quarter Kelly. Free, no signup, no tracking.
Kelly
in one line.
f* = (b·p − q) / b
f* is the fraction of your bankroll to stake. b is the net decimal odds (decimal − 1). p is your estimated win probability. q is 1 − p.
Kelly maximizes the expected logarithm of your bankroll, which is equivalent to maximizing long-run geometric growth. It is the optimal sizing rule when your probability estimate is accurate. It is brutally punishing when it isn't, which is why every serious operator uses a fractional Kelly. VAR's prediction market integration sizes at half Kelly by default for exactly that reason.
Questions,
answered.
What is the Kelly criterion?
The Kelly criterion is a mathematically optimal bet-sizing rule for any wager with a known edge. Given your estimated win probability and the offered odds, it tells you what fraction of your bankroll to stake to maximize long-term geometric growth. Bet less than Kelly and you grow more slowly; bet more, and you grow more slowly while taking on dramatically higher ruin risk.
Why do most professionals use half or quarter Kelly?
Full Kelly is optimal only if your estimated probability is exactly correct. In practice, every probability estimate carries error, and Kelly is severely punitive when you overestimate edge. Halving (or quartering) the stake gives up a small amount of expected growth in exchange for much lower drawdown variance and ruin risk. VAR uses half Kelly internally for that reason.
What does it mean when Kelly recommends no bet?
If your estimated probability is below what the offered odds imply, your edge is negative; the bet is unprofitable in expectation. Kelly returns a negative fraction in that case, which is interpreted as 'do not bet.' This calculator returns a $0 stake when that happens.
How do I convert American odds to win probability?
First convert to decimal odds. For positive American odds, decimal = (american / 100) + 1. For negative, decimal = (100 / |american|) + 1. The implied probability is 1 / decimal. American -110 converts to 1.909 decimal, which implies a 52.4% win rate, the standard ATS break-even threshold after the bookmaker's vig.
Should I use my model's probability or my own gut estimate?
Kelly is only as good as the probability fed into it. A calibrated model, one whose stated probabilities match observed frequencies out-of-sample, is the right input. Gut estimates almost always overstate edge, which is why naïvely applied Kelly destroys bankrolls. If you don't know whether your estimate is calibrated, use a smaller fraction (quarter Kelly or less).
Want the
model behind the math?
VAR builds calibrated probability models for franchises, media companies, and prediction-market operators. The Kelly calculator is the easy part. The hard part is the probability you feed it.
See our methodology