In tennis betting, picking the right side is only part of the job. The second part is deciding how much to stake. Bet too little, and a genuine edge barely matters. Bet too much, and a normal losing streak can damage your bankroll even if your long-term model is profitable.
The Kelly Criterion is a staking formula that converts your estimated edge into a recommended bankroll percentage. It is designed to maximize expected logarithmic bankroll growth when your probability estimate is accurate. That condition matters: if your true probability input is wrong, Kelly can recommend stakes that are too aggressive.
This Tennis Kelly Criterion & Variance Calculator estimates full Kelly, fractional Kelly, expected value, fair odds, recommended stake, losing-streak probability, and possible drawdown after a bad run.
Important: Kelly does not remove risk. It can still produce large drawdowns, especially in tennis where underdogs win often, retirements affect markets, and probability estimates can be noisy. Many bettors use half, quarter, or eighth Kelly to reduce volatility and protect against model error.
Tennis Kelly Criterion Calculator
Calculate full Kelly, fractional Kelly, expected value, fair odds, and losing-streak risk.
How to Use the Calculator
1. Enter Your Bankroll
Enter the bankroll allocated only for betting. Do not include money needed for rent, bills, savings, or daily expenses.
2. Enter Bookmaker Odds and Your Estimated Probability
Enter the decimal odds offered by the bookmaker and your estimated probability of the player winning. This probability is the most important input. The calculator does not estimate it for you.
For example, if the bookmaker offers 3.00 and you estimate the player has a 40% chance, the bet has positive expected value because the fair odds for 40% are 2.50.
3. Choose a Kelly Fraction
Full Kelly is mathematically aggressive. Fractional Kelly reduces stake size and volatility.
- Eighth Kelly: very conservative, useful when your edge estimate is uncertain.
- Quarter Kelly: common conservative setting for model-based betting.
- Half Kelly: more aggressive, higher growth but larger drawdowns.
- Full Kelly: maximum log-growth target if the probability estimate is correct, but high volatility.
Kelly Criterion Formula
For decimal odds, the full Kelly formula is:
Kelly % = ((Decimal Odds – 1) × Probability – (1 – Probability)) ÷ (Decimal Odds – 1)
Equivalent simplified form:
Kelly % = (Probability × Decimal Odds – 1) ÷ (Decimal Odds – 1)
| Input | Meaning |
|---|---|
| Decimal odds | The price offered by the bookmaker. |
| Your probability | Your estimate of the real chance of the player winning. |
| Expected value | The theoretical edge per unit staked. |
| Full Kelly | The bankroll percentage suggested by the full Kelly formula. |
| Fractional Kelly | The adjusted stake after applying your selected risk fraction. |
Worked Example: High-Value Underdog
Suppose you estimate an underdog has a 40% chance to win, and the bookmaker offers odds of 3.00.
- Bookmaker odds: 3.00
- Bookmaker implied probability: 33.33%
- Your estimated probability: 40.00%
- Expected value: +20% per unit staked
The full Kelly stake is 10% of bankroll. If you use Quarter Kelly, the recommended stake becomes 2.5% of bankroll. With a $1,000 bankroll, that means a $25 stake.
Worked Example: Strong Favorite With Bad Price
Suppose you believe a favorite has an 80% chance to win, but the bookmaker offers only 1.10.
- Bookmaker odds: 1.10
- Bookmaker implied probability: 90.91%
- Your estimated probability: 80.00%
The player may still be likely to win, but the price is too short. The expected value is negative, so the calculator recommends a stake of $0.
Understanding Variance and Losing Streaks
Tennis betting has real variance. Even a selection with a 55% win probability loses 45% of the time. A losing streak does not automatically mean the model is broken, but staking too aggressively can make normal variance financially damaging.
The calculator estimates the chance of a selected losing streak length using:
Losing Streak Probability = Loss Probability ^ Streak Length
It also estimates the bankroll drawdown after that losing streak if you keep staking the same percentage of your current bankroll.
Why Fractional Kelly Is Often Safer
Full Kelly assumes your probability estimate is accurate. In practice, tennis probabilities are uncertain. They can be affected by injury news, surface changes, fatigue, travel, draw difficulty, retirements, and market movement.
Fractional Kelly reduces the damage from model error. It also makes it easier to continue following a strategy through normal losing runs.
Limitations
This calculator does not estimate the true probability of a tennis match. It only converts your probability estimate and bookmaker odds into a staking recommendation. If your probability estimate is biased or too optimistic, the Kelly output will also be too aggressive.
Do not use Kelly staking on bets where you do not have a defensible probability estimate. Do not use the output as financial advice. Sports betting can result in total loss of bankroll.
Frequently Asked Questions
Why do bettors use fractional Kelly?
Fractional Kelly reduces volatility and protects against probability-estimation error. Full Kelly can create large bankroll swings when the edge estimate is wrong or too optimistic.
What is expected value in betting?
Expected value is the theoretical average return of a bet if the same edge were repeated many times. For decimal odds, EV per $1 staked equals probability multiplied by decimal odds, minus 1.
Does Kelly Criterion guarantee profit?
No. Kelly does not guarantee profit and does not prevent losing streaks. It only gives a mathematically defined stake size when your probability estimate is accurate.
Can I use Kelly for parlays?
Yes, but only if you can estimate the true probability of the entire parlay winning. Because parlays often have higher bookmaker margin and correlated legs, this is difficult to do accurately.
What happens if the calculator says no bet?
No bet means the expected value is zero or negative based on your inputs. The player may still win, but the offered price does not compensate for the risk under your estimated probability.