Quick answer: The Kelly Criterion is a bankroll management formula that calculates the optimal bet size based on your edge over the line and the offered odds. The classic formula: Kelly % = (Probability × Decimal Odds – 1) / (Decimal Odds – 1). The result tells you what percentage of your bankroll to wager. Most professional bettors use ‘half Kelly’ or ‘quarter Kelly’ to reduce variance, since full Kelly is mathematically optimal but creates wild bankroll swings.
The Formula in Practice
Say your model gives a bet 55% probability of winning at +110 (decimal 2.10). Kelly calculation: (0.55 × 2.10 – 1) / (2.10 – 1) = (1.155 – 1) / 1.10 = 14.1%. So full Kelly says bet 14.1% of bankroll. Half Kelly says 7.05%. Quarter Kelly says 3.5%. Most pros bet quarter Kelly because it captures most of the long-run growth while cutting the variance dramatically. The math says full Kelly is optimal in theory, but full Kelly produces drawdowns of 50%+ that few bettors can stomach.
Why Bet-Size Discipline Matters More Than Edge Discovery
An unsizable +EV bet is barely worth placing. A 0.5% Kelly bet at +1% EV doesn’t move the needle. Conversely, betting too large on uncertain edge ruins bankrolls. The dirty secret of pro betting is that bet sizing matters more than picking winners. PropsBot publishes daily picks with edge calculations, but we always remind users: Kelly-style sizing on those picks is what compounds returns. Our High ROI Signal hits 31.7% verified ROI on 101,881 MLB props at moderate edge percentages, which means realistic Kelly fractions per bet are 1-3% of bankroll.
When NOT to Use Kelly
Kelly assumes you know your true probability accurately. If your model is overconfident, full Kelly will bankrupt you. The remedy is fractional Kelly, which is mathematically equivalent to assuming your probability estimate is somewhat off. Most sharp bettors use quarter Kelly or even one-eighth Kelly when betting markets they don’t fully trust their model on. The other case to avoid Kelly: parlays. The math gets tangled because of correlation between legs. Stick to single bets when applying Kelly cleanly.
A Worked Bankroll Example
Starting bankroll: $5,000. PropsBot model says bet has 58% probability at -120 (decimal 1.83). Edge over no-vig probability: ~6%. Full Kelly: (0.58 × 1.83 – 1) / (1.83 – 1) = 0.062 / 0.83 = 7.5%. Quarter Kelly: 1.9%. So bet $95 of the $5,000 bankroll on this single play. Over 200 bets at this edge level, expected bankroll growth is approximately 32%, with realistic swings of ±15% along the way. The PropsBot Kelly Calculator (propsbot.ai/tools/kelly-criterion-calculator) handles the math for any bet.
Frequently Asked Questions
What is the Kelly Criterion formula?
Kelly % = (Probability × Decimal Odds – 1) / (Decimal Odds – 1). The result is the percentage of your bankroll to wager.
Why do most bettors use fractional Kelly?
Full Kelly produces bankroll swings that most bettors can’t tolerate emotionally. Half or quarter Kelly captures most of the long-run growth while cutting variance roughly in half.
Does Kelly work on parlays?
Not cleanly. Parlay legs are correlated, which violates Kelly’s independence assumption. Most sharp bettors stick to Kelly on single bets and skip parlays altogether.
What happens if I overestimate my probability?
Full Kelly amplifies the error and can ruin a bankroll quickly. Fractional Kelly (quarter or eighth) provides a cushion against probability misestimation.
Can I use Kelly for daily fantasy or DFS contests?
DFS optimization is more complex than single-event Kelly. The math involves entry fees, payout curves, and contest structure. Kelly is best applied to traditional fixed-odds wagers.
Part of the PropsBot.AI Sports Betting Glossary. Updated 2026-05-04.
