Kelly Criterion Calculator – Optimal Bet Sizing Guide

The Kelly Criterion answers a question that plagues every serious bettor: how much should I stake? Developed by mathematician John Kelly in 1956, the formula calculates the optimal stake size that maximises long-term bankroll growth while managing risk of ruin. Too little stake means leaving profit on the table. Too much stake means risking bankruptcy from normal variance. Kelly finds the precise balance.

The criterion requires two inputs: your edge over the bookmaker and the odds available. Given these figures, Kelly produces a percentage of your bankroll to stake on each bet. This percentage adjusts automatically as your bankroll grows or shrinks, maintaining optimal risk exposure throughout your betting career.

British horse racing contributes £4.1 billion annually to the UK economy and supports approximately 85,000 jobs according to the House of Commons Library. Within this significant industry, serious punters seek mathematical frameworks for bankroll management. The Kelly Criterion provides exactly such a framework, borrowed from financial mathematics and applied to betting contexts.

Kelly Criterion Formula

The Kelly formula expresses optimal stake as a fraction of bankroll: f = (bp – q) / b. In this formula, f represents the fraction of bankroll to stake, b represents the decimal odds minus 1 (the net profit per unit staked), p represents the probability of winning, and q represents the probability of losing (which equals 1 – p).

Breaking this down: the numerator (bp – q) measures your expected profit. If bp exceeds q, you have positive expected value and should bet. If q exceeds bp, you have negative expected value and should not bet at all. The denominator (b) scales the stake appropriately for the odds involved.

Consider a horse you estimate has a 40% chance of winning at 3/1 odds. Here p = 0.40, q = 0.60, and b = 3. Kelly stake = (3 × 0.40 – 0.60) / 3 = (1.20 – 0.60) / 3 = 0.60 / 3 = 0.20. You should stake 20% of your bankroll on this bet.

The formula automatically returns zero or negative values for -EV bets. If your 40% probability horse was available at only 1/1 odds: (1 × 0.40 – 0.60) / 1 = -0.20. The negative result tells you not to bet. Kelly never recommends staking on losing propositions.

Note that Kelly assumes you can accurately assess probabilities. If your probability estimates are wrong, Kelly’s recommendations are wrong. The formula optimises stake size given accurate inputs, but garbage in produces garbage out.

Calculating Your Edge

Your edge represents the difference between your estimated probability and the bookmaker’s implied probability. Without an edge, Kelly recommends zero stake. With a large edge, Kelly recommends larger stakes. Accurately quantifying your edge is essential for Kelly to work.

Start by converting bookmaker odds to implied probability. At 4/1 (decimal 5.0), implied probability is 1/5.0 = 20%. If you estimate the true probability at 25%, your edge is 5 percentage points. This edge drives Kelly’s stake recommendation.

Edge can be expressed in different forms. Absolute edge (25% – 20% = 5%) shows the probability gap. Relative edge (5% / 20% = 25% better than the bookmaker thinks) shows proportional advantage. Either form can inform your betting decisions, though Kelly uses absolute probabilities in its formula.

Honest edge assessment proves difficult. Most punters overestimate their edge because winning bets feel like skill while losing bets feel like bad luck. Track your results over hundreds of bets to discover your true edge. If you consistently overestimate, reduce your probability estimates accordingly before applying Kelly.

The bookmaker’s margin affects edge calculations. A horse with true 25% probability might be priced at 3/1 (25% implied) at one bookmaker and 11/4 (26.7% implied) at another. Your edge changes based on where you bet. Always calculate edge against the specific odds you are taking.

Full Kelly vs Fractional Kelly

Full Kelly maximises theoretical bankroll growth but creates significant volatility. Staking 20% of your bankroll on a single bet, even a +EV one, exposes you to substantial swings. A losing run can decimate your bankroll before the long-term mathematics rescue you.

Fractional Kelly addresses this volatility by staking a fraction of the full Kelly recommendation. Half Kelly means staking 10% when full Kelly suggests 20%. Quarter Kelly means staking 5%. These reduced stakes sacrifice some expected growth in exchange for dramatically reduced variance.

The mathematics favour fractional approaches for most bettors. Half Kelly produces 75% of full Kelly’s expected growth rate while reducing variance significantly. Quarter Kelly produces about 50% of expected growth but with variance reduced by approximately 75%. The smoother journey often proves more sustainable psychologically.

Probability estimation errors further justify fractional Kelly. If you overestimate your edge, full Kelly overstakes, accelerating losses. Fractional Kelly provides a buffer against estimation mistakes. Many professional bettors use quarter Kelly or less, acknowledging uncertainty in their probability assessments.

Your risk tolerance should guide fraction choice. Aggressive bettors comfortable with larger swings might use half Kelly. Conservative bettors prioritising bankroll preservation might prefer quarter or even eighth Kelly. There is no objectively correct fraction; the right choice depends on your circumstances and psychology.

Example Calculations

Your bankroll is £1,000. You estimate a horse has 30% probability of winning at 4/1 odds.

Full Kelly: f = (4 × 0.30 – 0.70) / 4 = (1.20 – 0.70) / 4 = 0.50 / 4 = 0.125. Stake 12.5% of bankroll = £125.

Half Kelly: 0.125 / 2 = 0.0625. Stake 6.25% = £62.50.

Quarter Kelly: 0.125 / 4 = 0.03125. Stake 3.125% = £31.25.

Now consider a lower edge scenario. Same horse, but odds are only 5/2 instead of 4/1.

Full Kelly: f = (2.5 × 0.30 – 0.70) / 2.5 = (0.75 – 0.70) / 2.5 = 0.05 / 2.5 = 0.02. Stake just 2% = £20.

The dramatic reduction from £125 to £20 illustrates how Kelly responds to edge changes. Smaller edge means smaller stake because the mathematics no longer justify aggressive betting.

Finally, consider what happens when your probability estimate drops to 20% at 4/1 odds.

Full Kelly: f = (4 × 0.20 – 0.80) / 4 = (0.80 – 0.80) / 4 = 0. Kelly recommends no stake at all because the bet is exactly break-even. Any lower probability would produce negative values, actively warning against betting.

Practical Application

Implementing Kelly requires discipline and honest self-assessment. Begin with conservative fractional Kelly until you have established a track record demonstrating genuine edge. Premature full Kelly staking based on imagined edge destroys bankrolls quickly.

Update your bankroll figure regularly. Kelly’s percentage recommendations only work correctly when applied to current bankroll. After wins, your stakes should increase proportionally. After losses, stakes should decrease. This automatic adjustment protects you during downswings and compounds gains during winning runs.

Multiple simultaneous bets complicate Kelly application. The formula assumes each bet is independent and that you are reinvesting your entire bankroll between bets. When betting multiple races on the same day, purists calculate Kelly separately for each, summing the percentages, while practitioners often simplify by dividing available bankroll across opportunities.

Our calculator handles the mathematics automatically. Enter your probability estimate, the odds available, and your current bankroll. The calculator displays full Kelly stake, half Kelly, and quarter Kelly recommendations. It also shows expected value to confirm the bet offers positive expectation before staking.

Remember that Kelly optimises long-term growth, not individual bet outcomes. Any single bet might lose regardless of its Kelly-recommended stake. Trust the process across hundreds of bets rather than judging Kelly by short-term results.