Kelly Criterion Calculator

Enter your edge and odds to find the mathematically optimal bet size.

Bankroll ($)

Decimal odds

Your estimated win probability (%)

Kelly fraction (1 = full, 0.5 = half)

Full Kelly bet (%) —

Fractional Kelly bet (%) —

Bet amount —

Your edge —

Expected value per bet —

Kelly formula: f* = (bp - q) / b, where b = decimal odds - 1, p = win probability, q = 1 - p. Half-Kelly is commonly used to reduce variance.

🛒 Shop related: Strategy books · Poker chips
Affiliate links — we may earn a commission at no cost to you. Learn more

Get your results + weekly betting tips

One email per week. Tips, tool deals, new calculators. Unsubscribe anytime.

Free tools, useful tips. We recommend products we'd actually use.

The Kelly Criterion Explained

The Kelly criterion is a formula for sizing bets to maximize long-term bankroll growth. It balances the tradeoff between betting too little (leaving edge on the table) and too much (risk of ruin).

Full Kelly can produce large swings. Most professional bettors use fractional Kelly (typically 25-50% of the full Kelly amount) to reduce variance while still capturing most of the expected growth.

If the Kelly result is negative, you have no edge — the bet has negative expected value and should not be placed.

Kelly Criterion determines the optimal bet size to maximize long-term bankroll growth: f = (bp − q) / b, where b = decimal odds minus 1 (net odds), p = your estimated win probability, q = 1 − p. Example: a bet at 2.0 decimal odds (even money) where you estimate 55% win probability: b = 1.0, p = 0.55, q = 0.45. f = (1.0 × 0.55 − 0.45) / 1.0 = 0.10 (bet 10% of bankroll).

Half-Kelly convention: full Kelly bets are optimal only if your probability estimates are perfectly accurate. Since estimates are imperfect, most professional bettors use 1/4 to 1/2 Kelly to reduce variance. Half-Kelly in the example above = 5% of bankroll. This sacrifices some growth rate but dramatically reduces drawdown risk from estimation errors.

Kelly below zero: when (bp − q) is negative, Kelly recommends not betting (expected value is negative). This is useful as a filter — any bet with negative Kelly by your probability estimate is a losing bet over time. If the implied probability exceeds your estimate, skip the bet.

Limitations: Kelly assumes accurate probability estimates, fixed odds, and repeated independent bets. In practice, edge estimation is difficult, bankrolls are finite, and many bets correlate. The formula is a theoretical optimal — treat it as an upper bound, not a precise prescription. Overconfident probability estimates combined with full Kelly sizing cause rapid ruin.

Educational purposes only. Problem gambling: NCPG 1-800-522-4700.