Kelly Criterion Calculator

Size the bet the Kelly way, then apply a fraction that survives variance.
Author
Arthur VBF
27 August 2026

0%

0%

$0

Kelly Criterion Calculator

You found an edge. Now size it. Enter the decimal odds, your win probability and your bankroll, and the calculator returns the exact Kelly stake. It also shows the fractional version most sharp bettors actually use. Read the number before you place anything.

7-day free trial, no credit card required.

3 inputs

Odds, probability, bankroll

Full + fractional

Half and quarter Kelly output

Any odds format

Decimal, fractional, American

Negative = skip

Flags no-edge bets instantly

A value bettor spots odds of 2.10 on a market they price at 52 percent. The edge is real, but the next question is harder: how much of the bankroll goes on it? Bet too big and one bad run erases months of work. Bet too small and the edge barely compounds. The Kelly criterion answers that with a formula, and the calculator above runs it for you.

What this Kelly calculator settles

Stake sizing, solved

One formula between you and overexposure

Two bettors can find the same +EV bet and end the season far apart, purely on stake sizing. Staking by feel is where most edges quietly die.

The Kelly criterion, published by John Kelly at Bell Labs in 1956, maximises the long-run growth rate of a bankroll. It stakes more on a bigger edge and less on a thinner one. With no edge, it tells you to bet nothing at all.

Use the result as a ceiling, not a command. The sections below show why almost no serious bettor stakes the full figure.

Your probability 52% Implied by odds 47.6% EDGE +4.4 points

Edge drives the stake. No edge, no bet.

Convert any price to decimal first with our odds converter tool so the number you paste in is clean. The probability input is the one that matters most, and our guide to positive EV betting covers how to derive a fair probability from Pinnacle or Betfair prices.

How to use the calculator

Four inputs, one output

1

Enter the decimal odds

The price at your bookmaker, in decimal form. Odds of 2.10 return 2.10 units on a winning one-unit stake.

2

Enter your win probability

Your honest estimate, as a percentage. It should come from a model or devigged sharp odds, never from how confident you feel.

3

Enter your bankroll

The total pot set aside for betting, and nothing else. Kelly returns a percentage of this number, so keep it honest.

4

Read the stake, then apply a fraction

The calculator shows the full Kelly stake. Multiply it by a half or a quarter before you bet, for the reasons shown below.

A worked example, step by step

Odds 2.10, probability 52%, bankroll 1,000

Same bet, full Kelly versus quarter Kelly

Decimal odds

2.10

Win probability

52%

Bankroll

1,000

Full Kelly

8.36%

A stake of about 83.64 units. Mathematically optimal, but volatile.

Quarter Kelly

2.09%

A stake of about 20.91 units. Same edge, a quarter of the swings.

Change one number. If the price drifts to 1.90 while your estimate stays at 50 percent, the calculator returns roughly minus 5.56 percent. A negative result is not a small bet, it is a clear instruction to skip. Passing on that bet is the profitable move.

Full Kelly versus fractional Kelly

Growth rate by staking fraction

Why almost nobody stakes full Kelly

Growth peaks at full Kelly but only if your probability is perfect. It never is. Betting a fraction sacrifices little growth and cuts the swings hard. Half Kelly keeps about three quarters of the growth. Bet double and growth collapses to zero.

growth fraction of full Kelly 0.25x 0.5x Full Kelly 2x = 0 0 0.5 1.0 1.5 2.0
Long-run growth rate
Full Kelly, the peak
Double Kelly, growth gone

There is no universally correct fraction. For a bettor with a well-tested model and a large sample behind their edge, half Kelly is defensible. For someone newer to +EV betting, or working from noisy estimates, quarter Kelly protects against the very real chance the true edge is smaller than it looks. When your inputs are uncertain, staking less is the honest choice. The theory sits in our explainer on the Kelly criterion, worth reading once before you trust any staking output.

Mistakes that wreck a Kelly stake

Most damaging

Overstating your probability

Kelly is highly sensitive to this input. Nudging 52 up to 55 percent because a bet feels good inflates the stake and turns a small edge into overexposure.

Volatility trap

Full Kelly on an unproven edge

Betting the full fraction on a 2 percent edge you cannot yet prove over a large sample is a fast route to a painful drawdown. Fractional Kelly exists for this.

Drift

Not updating the bankroll

Kelly is proportional. Keep entering an old bankroll figure and every stake drifts from optimal. Update the number as the bankroll actually moves.

Reality check

Ignoring the market limit

A 200-unit stake means nothing if the soft book caps you at 30, or if staking it flags your account. Size against real liquidity, not just the theoretical figure.

Two habits close the loop. Cap your staking fraction in advance, a quarter to a half, and never override it mid-session. And track closing line value on every bet, since positive CLV across a few hundred bets confirms your probabilities are sound long before the profit and loss does. Our bankroll management guide ties the staking maths to the discipline that makes it work.

Sizing the bet is half the job. Finding the edge is the other half.

The +EV scanner surfaces mispriced soft-book odds against Pinnacle and Betfair in real time, straight to Telegram.

See the +EV scanner

Common questions about the Kelly calculator

What if the calculator returns a negative number?

A negative Kelly fraction means the odds offer no value at your estimated probability. The implied probability of the price is higher than your own estimate, so the expected value is negative. The correct action is to place no bet at all. Treat a negative output as a filter that just saved you a losing stake, not as a figure to bet small.

Should I use full Kelly or a fraction?

Almost everyone should use a fraction. Full Kelly is optimal only with perfect probability estimates, which do not exist in sports betting. Half Kelly keeps most of the growth with far less variance, and quarter Kelly is calmer again. If your edge is unproven over a large sample, quarter Kelly is the safer starting point.

Where do I get the win probability to enter?

From a repeatable method, not a feeling. Many value bettors devig the odds from a sharp book such as Pinnacle or Betfair Exchange to estimate a fair probability, then compare it to the soft book price. A model or a documented system also works. The one thing to avoid is entering a number because the bet looks appealing.

Does a Kelly stake stop my account getting restricted?

No. Kelly sizes a bet for bankroll growth, not for account safety. Soft bookmakers restrict winning accounts regardless of staking method, often within 50 to 500 bets. Staking to the market limit or placing conspicuously large bets can speed that up. Sizing sensibly and spreading action across books helps, but no calculator removes restriction risk.

Can I use the Kelly criterion for anything besides betting?

Yes. Kelly originated in information theory and is used in investing to size positions under uncertainty. The maths is identical: an edge, a payoff and a probability. The caution is the same too, since full Kelly is volatile wherever the probability estimate is imperfect, which in real markets it always is.

Is a bigger edge always worth a bigger stake?

In theory yes, and the calculator reflects that by raising the fraction as the edge grows. In practice, a large fraction on a single bet concentrates risk, and your edge estimate on that bet may be wrong. This is why a staking cap and fractional Kelly matter more as the suggested stake climbs.

Turn a real edge into a stake that survives variance

A Kelly calculator is only as good as the probability you feed it and the discipline you wrap around it. Size the bet, apply a fraction you decided in advance, and log every result so your closing line value can confirm the edge is real. Then test it on your own bookmakers, on real markets, for seven days.

Start your 7-day free trial

No credit card required.

Betting involves risk. Past performance does not guarantee future results. Bet responsibly. If you or someone you know has a gambling problem, visit begambleaware.org, GamCare, or the NCPG in the United States.