The Kelly Criterion is the formula professional bettors and quantitative traders use to decide exactly how much of their bankroll to stake on a bet with a known edge. It isn’t about picking winners — it’s about sizing the bets you already believe in correctly.
Loading calculator...
Bet too little relative to your edge and you leave long-term growth on the table. Bet too much and even a small string of losses can wipe out a large share of your bankroll. This calculator solves that balance for you instantly, in decimal, American, or fractional odds.
Below, we walk through exactly how the tool works, what each field and result means, the math behind the formula, and the mistakes that cause most bettors to misuse Kelly staking in practice.
📊 How to Use the Kelly Criterion Calculator
Start by selecting the odds format you have — Decimal, American, or Fractional — and enter the odds exactly as your bookmaker displays them. The calculator converts everything internally, so you never need to do the conversion yourself.
Your win probability estimate is the single input that matters most — if it’s wrong, every downstream number will be wrong too.
Next, enter your honest estimate of your win probability as a percentage, along with your total bankroll and preferred currency. Finally, choose a Kelly fraction — Full, Half, or Quarter — and the recommended stake updates immediately.
🔢 Calculator Fields Explained
Odds Format – Whether you’re entering Decimal (2.50), American (+150 / -120), or Fractional (5/1 / 3/2) odds.
Odds – The price your bookmaker or exchange is currently offering on the outcome you’re evaluating.
Win Probability – Your own estimate of how likely the outcome is to happen, expressed as a percentage from 1 to 99.
Bankroll – The total amount of betting capital you’re staking sizing decisions against.
Currency – The display currency for your bankroll and recommended stake.
Kelly Fraction – Full, Half, or Quarter Kelly, controlling how aggressively you apply the raw formula result.
💰 Understanding the Results
| Result | What It Means |
|---|---|
| Full Kelly Stake | The raw percentage of bankroll the formula recommends before any fractional adjustment |
| Applied Stake | The final percentage after applying your chosen Full/Half/Quarter setting |
| Recommended Stake | The Applied Stake percentage converted into your chosen currency amount |
| Betting Edge | Your expected profit per unit staked, as a percentage, given your probability estimate |
| Expected Growth Rate | The theoretical long-run geometric growth rate of your bankroll if this exact bet were repeated indefinitely |
The Recommended Stake is the number most people act on directly. It already reflects both your edge and your chosen level of aggressiveness.
A Kelly stake is only as good as the probability estimate behind it — treat the output as conditional, not guaranteed.
If the calculator shows zero edge, it means your estimated probability doesn’t justify a bet at these odds at all, regardless of bankroll size.
The Kelly stake shrinks toward zero as your edge shrinks toward zero — that relationship is the entire point of the formula.
📐 Calculation Formulas
The core formula is f* = (bp - q) / b, where b is the decimal odds minus one, p is your win probability, and q is 1 minus p. The table below compares the three fractional staking approaches this calculator offers.
| Staking Mode | Multiplier Applied | Typical Use Case |
|---|---|---|
| Full Kelly | 1.00× | Maximum long-run growth, highest short-term volatility |
| Half Kelly | 0.50× | Common professional compromise between growth and drawdown risk |
| Quarter Kelly | 0.25× | Conservative sizing for uncertain or unstable probability estimates |
Fractional Kelly sacrifices some theoretical growth in exchange for a meaningfully smoother bankroll curve.
Most professional bettors never run Full Kelly in practice, precisely because real-world probability estimates always carry some uncertainty that the pure formula doesn’t account for.
📝 Practical Examples
Example 1: Decimal odds of 2.10, a 55% win probability, and a $2,000 bankroll on Full Kelly produce a stake of roughly 9.3% of bankroll, or about $186.
Example 2: The same bet run on Half Kelly instead recommends staking around 4.65% of bankroll — about $93 — cutting the theoretical growth rate only modestly while sharply reducing volatility.
Switching from Full to Half Kelly typically halves your stake while giving up only a small fraction of long-run growth potential.
Example 3: American odds of +120 with a 50% win probability show only a small positive edge, producing a small recommended stake even on Full Kelly — a reminder that thin edges deserve thin stakes.
Example 4: Fractional odds of 5/1 with a genuine 20% win probability produce a much larger Kelly percentage than most bettors expect, because the payout multiplier is large relative to the risk.
A single overstated win probability of just a few percentage points can nearly double the recommended stake size.
💡 Tips & Best Practices
Treat your win probability estimate as the most important input you’ll provide, and be conservative rather than optimistic when you’re unsure. An overstated edge is the single most common cause of Kelly-related losses.
Use Half Kelly or Quarter Kelly as your default operating mode rather than Full Kelly, especially early in your betting career when your probability models haven’t been tested across many outcomes.
Recalculate your stake every time the odds move, even slightly — Kelly sizing is odds-sensitive, and a stake calculated on yesterday’s line may no longer be optimal.
Recording your estimated probability alongside the actual outcome over time is the single best way to calibrate your future Kelly inputs.
Keep a running log of your bets, your estimated probabilities, and the outcomes. Over dozens of bets, this reveals whether your probability estimates run optimistic or conservative.
- Never stake more than the Full Kelly amount, even on a “sure thing”
- Reduce your fraction after a losing streak rather than increasing it
- Reassess your bankroll figure regularly, not just at the start
Avoid the temptation to increase your Kelly fraction after a winning streak — the formula already accounts for compounding, and overriding it manually reintroduces the exact risk it’s designed to control.
⚠️ Common Mistakes to Avoid
Overestimating Win Probability
The most damaging mistake in Kelly staking isn’t the math — it’s feeding the formula an inflated probability estimate.
Even a 5-percentage-point overestimate in win probability can inflate your recommended stake far beyond what your true edge justifies.
Because the formula is highly sensitive to this input, small estimation errors compound into significantly oversized stakes.
Running Full Kelly With an Unstable Bankroll
Full Kelly assumes your bankroll figure is stable and accurately entered every time you calculate a stake.
Using a stale or inflated bankroll figure causes every subsequent Kelly stake to be miscalculated, not just the current one.
Update your bankroll input after every settled bet rather than relying on a figure from days or weeks earlier.
Ignoring Correlated Bets
Placing several Kelly-sized bets on outcomes that aren’t truly independent — like different markets on the same match — effectively concentrates risk the formula never accounted for individually.
Chasing Losses by Increasing the Fraction
Increasing your Kelly fraction after a losing streak is the fastest documented path to blowing up a bankroll. The formula is designed to be followed consistently, not adjusted emotionally.
🎯 When to Use This Calculator
This calculator is most useful whenever you have a genuine, testable edge — a probability estimate you can defend with data, not a hunch — and you want a disciplined, repeatable way to size your stake accordingly.
Kelly staking rewards bettors who already have an edge; it cannot manufacture one for bettors who don’t.
It’s particularly valuable for value bettors tracking closing line value, matched bettors evaluating recurring opportunities, and anyone managing a dedicated betting bankroll over a long time horizon.
🔗 Related Calculators
Risk Of Ruin Calculator, Compounding Growth Calculator, Unit Size Calculator, Sharpe Ratio Calculator, EV Calculator, Implied Probability Calculator.
📖 Glossary
Kelly Criterion – A formula for calculating the optimal fraction of a bankroll to stake given an edge and odds.
Edge – Your expected profit per unit staked, based on your probability estimate versus the odds offered.
Bankroll – The total pool of capital dedicated to betting.
Decimal Odds – Odds expressed as a single multiplier of the stake, including the original stake.
American Odds – Odds expressed as the amount won on a $100 stake (positive) or the stake needed to win $100 (negative).
Fractional Odds – Odds expressed as a ratio of profit to stake.
Full Kelly – Staking the entire raw percentage the Kelly formula outputs.
Fractional Kelly – Staking a reduced multiple (commonly Half or Quarter) of the raw Kelly output.
Expected Value (EV) – The average profit or loss expected per bet over many repetitions.
Geometric Growth Rate – The compounding rate at which a bankroll is expected to grow over repeated bets.
Overbetting – Staking more than the Kelly formula recommends for a given edge.
Risk of Ruin – The probability of losing a bankroll down to an unrecoverable level.
Implied Probability – The win probability suggested by a given set of odds before accounting for bookmaker margin.
❓ Frequently Asked Questions
What happens if I enter a win probability that’s too high?
The calculator will recommend a larger stake than is actually justified by your true edge, increasing both expected variance and drawdown risk.
For example, overstating a true 50% probability as 60% at even-money odds can turn a correctly-sized small stake into a dangerously large one.
Should I ever use Full Kelly?
Full Kelly is mathematically optimal only if your probability estimate is exactly correct, which is rarely guaranteed in practice.
Most professional bettors default to Half Kelly specifically because it tolerates estimation error far better than Full Kelly.
If you’re highly confident in a well-tested model, Full Kelly can be appropriate, but it should be the exception rather than the default.
Why does the calculator show a stake of zero?
A zero stake means your estimated edge at the given odds is zero or negative — the Kelly Criterion never recommends staking on a bet without positive expected value.
This is a feature, not a bug: it protects you from betting into markets where your estimate doesn’t actually beat the price on offer.
Can I use this calculator for accumulator or parlay bets?
The formula assumes a single, independent bet with a known probability and payout — it isn’t designed for the compounded probability structure of multi-leg accumulators.
For parlays, you’d need to calculate the combined probability and combined odds first, then treat the whole parlay as one Kelly input.
How often should I recalculate my stake?
Recalculate every time either the odds or your probability estimate changes — a stake calculated on stale inputs is no longer the optimal stake.
In fast-moving markets, this can mean recalculating multiple times before you actually place the bet.
Does bankroll size affect the recommended percentage?
No — the Kelly percentage itself depends only on your edge and the odds, not your bankroll size.
Bankroll size only affects the final currency amount, since the same percentage applied to a larger bankroll produces a proportionally larger stake.
What’s the difference between Half Kelly and Quarter Kelly in practice?
Half Kelly stakes twice as much as Quarter Kelly for the same edge, trading additional short-term volatility for slightly faster long-run growth.
Many bettors start with Quarter Kelly while building confidence in their probability model, then move to Half Kelly once their estimates prove reliable over time.
Is a negative Expected Growth Rate possible?
Yes — if the applied stake is set too aggressively relative to a thin or nonexistent edge, the geometric growth rate can turn negative even with a technically positive edge.
This is one reason fractional Kelly modes exist: they keep the applied stake safely within the range where growth stays positive.
⚖️ Legal Disclaimer
This calculator is provided for informational and educational purposes only. It does not constitute financial or betting advice, and no outcome or profit is guaranteed. Betting involves risk of loss, and you should never stake more than you can afford to lose. Please gamble responsibly and consult local regulations regarding legal betting activity in your jurisdiction.
Kelly Criterion is solid in theory but the payout mechanics matter way more than most calculators acknowledge. I’ve been routing withdrawals through Skrill for years and seeing instant transfers, but the moment you go above certain thresholds with some books, they force bank wire which takes 3-5 business days and hits you with $15-25 fees. Your bankroll management becomes pointless if you can’t access capital quickly when the next edge shows up. Also, check if your operator supports Visa Direct payouts—those are genuinely faster than debit card reversals and don’t trigger the same fraud flags. Crypto integration helps too. BTC withdrawals bypass the entire banking friction layer, though the volatility means you’re either holding through swings or taking an FX hit when converting back. The calculator assumes your bankroll sits static, but real money moves. Transaction fees and withdrawal delays compress your effective growth rate by 1-3% annually if you’re cycling capital monthly. Worth factoring into your Kelly fraction—maybe run 0.4x Kelly instead of 0.5x if your operator has slow payouts.
You’ve identified a critical blind spot in how most bettors apply Kelly—the withdrawal infrastructure layer. You’re right that payout velocity directly impacts effective bankroll growth. A 3-5 day wire delay on a $5,000 withdrawal genuinely costs you edge on subsequent opportunities, especially in sharp markets where lines move within hours. The crypto angle is particularly relevant here: while BTC volatility adds conversion risk, operators using stablecoin payouts (USDT on Polygon or Arbitrum) eliminate both the banking delay and FX slippage. Some books now process USDT to exchange within 15 minutes, which functionally extends your available capital. Your suggestion to adjust Kelly fraction for operator-specific friction is sound. If your average payout takes 72 hours and you’re cycling capital 8-10 times annually, you’re effectively running 1-2% drag that the formula doesn’t see. We’ve started tracking operator payout speeds in our compliance profiles—withdrawal delay data is becoming as important as licensing status for quant bettors.
Thanks, that stablecoin detail is exactly what I needed. I’ve been hesitant about crypto payouts because of the BTC volatility you mentioned, but USDT to an exchange in 15 minutes changes the calculation entirely. Been with my current book for three years and never realized their withdrawal speed was costing me more than their juice on some lines. Gonna audit their actual payout times this month and see if it justifies switching to one of the MGA-licensed operators that support faster rails.
Exactly—and when you audit those payout times, pay attention to the variance across methods. Most operators have tiered speeds: e-wallets fastest (Neteller usually 2-4 hours), then crypto, then bank transfers slowest. Some books artificially delay payouts during peak hours to manage liquidity, so check a weekend withdrawal vs. a Tuesday one. You might find their stated ‘instant’ claim is only real for deposits, not withdrawals. The MGA operators tend to be more transparent about this in their terms, though not all enforce it equally. Let us know what you find—we’re building a withdrawal speed database and real operator data is scarce.
This is way too mechanical for how betting actually works. Yeah the math checks out but you can’t just plug in a number and pretend that’s your real edge when you’re watching the team sheet an hour before kickoff. I had Liverpool at -110 yesterday, calculator said 2.3% of bankroll, but then I see Van Dijk’s out and suddenly the whole calculus changes. That one injury just shifted my probability estimate down 4-5 points. The formula doesn’t account for the fact that your read on a game isn’t static—it evolves. Also half the time you think you have an edge you’re just catching the line before sharp money moves it. Sold my stake down to 1% instead and watched it hit anyway. Would’ve lost less if I’d trusted the Kelly but made more by trusting my gut on squad news. Real talk though, I probably need this discipline because my gut is also how I turned 3k into 1.2k last season.
Regarding your Van Dijk example—you’re describing the core tension between static probability estimation and real-time information flow, which Kelly assumes away. The formula works perfectly if your probability estimate is accurate and stable, but you’re right that injury news, weather shifts, and line movement all compress or expand your actual edge within minutes. Where Kelly still helps you is in preventing the reverse problem: betting the same percentage on a 51% edge and a 62% edge because ‘it feels like a winner.’ Your 2.3% recommendation to 1% adjustment shows good intuition about variance, though it’s worth noting that pure emotional discipline and pure mathematical discipline often pull in different directions. The literature on this suggests hybrid approaches work best—use Kelly to set an upper bound on your stake (never exceed what the formula says, even on conviction plays), then apply personal adjustment downward based on probability uncertainty. Your 1.2k loss last season might’ve been prevented by letting Kelly enforce a ceiling rather than fighting it entirely. Consider it a floor-and-ceiling system rather than a point estimate.