This free Kelly criterion calculator turns two numbers from your trading record into an optimal risk size. Enter your winning-trade percentage and your payoff ratio. The tool returns the full Kelly fraction, plus the half and quarter Kelly sizes most traders actually use. It runs entirely in your browser. Just as important, this page explains why you should almost never risk the full Kelly number. The formula is elegant in theory and brutal in practice. Read the guide below before you size a single trade with it.
Kelly Criterion Calculator
The share of your closed trades that ended in profit. Pull it from at least 50 real trades.
Average winning trade divided by average losing trade, in money or R-multiples.
How to use the kelly criterion calculator
The tool needs two inputs, and both must come from your own trading record. Pull them from a broker statement or a trade journal. Guessed inputs produce a confident-looking number that means nothing.
- Count your closed trades and work out the winning-trade percentage. Sixty winners out of 120 trades gives 50.
- Compute the payoff ratio. Divide your average winning trade by your average losing trade. Wins averaging $150 against losses averaging $100 give 1.5.
- Enter both numbers and press Calculate.
- Read the full Kelly fraction first. That is the theoretical optimum, not a recommendation.
- Note the half and quarter Kelly lines. These are the sizes serious practitioners actually consider.
If the result box says your numbers show no positive edge, stop. A negative Kelly fraction means your system loses money over time. No sizing trick fixes that. Work on the strategy first and check it with the expectancy calculator. Kelly only has meaning once expectancy is positive.
A worked example: 50% winners at a 1.5 payoff
Take the default inputs. Half your trades win, and winners are 1.5 times larger than losers. The formula runs like this. Convert the winning percentage to a decimal: 0.50. Divide the losing share by the payoff ratio: 0.50 ÷ 1.5 = 0.333. Subtract: 0.50 − 0.333 = 0.167. Full Kelly says risk 16.7% of your capital on the next trade.
Pause on that number. On a $10,000 account, 16.7% is $1,670 at risk on one position. Three losers in a row, each resized to the shrinking balance, would cut the account by roughly 42%. That losing streak is routine for a system that wins half the time. The math is correct, and the outcome is still unlivable. Half Kelly at 8.3% softens it. Quarter Kelly at 4.2% softens it further. Most traders should go lower still, as the practice section below explains.
The Kelly formula explained
The formula is f* = W − (1 − W) / R. Here W is your winning probability as a decimal, and R is the payoff ratio. The output f* is the fraction of capital to risk per trade. John Kelly derived it at Bell Labs in 1956 for information theory. Blackjack teams and some fund managers later adopted it for bet sizing.
What does it optimize? Long-run compound growth. Kelly maximizes the expected logarithm of your wealth. In plain terms, it finds the risk size that grows an account fastest over many repeated bets. Bet less and you grow slower but smoother. Bet more and growth also falls, while the swings get worse. At exactly double Kelly, expected growth drops to zero. Beyond that, an edge-positive system compounds downward.
That asymmetry is the key insight. The penalty for betting too much is far harsher than the penalty for betting too little. Underbetting costs you some growth. Overbetting can destroy the account. Every practical rule in the next sections flows from this one fact.
Download the complete indicator database
Pair disciplined sizing with tested tools on your chart. One email unlocks the full library of 1,380+ indicators with compiled MT4 and MT5 files, plus my TradingView scripts. No paywall, no spam, unsubscribe any time.
Get free access to my indicator database
One email unlocks 1,380+ free MT4, MT5 and TradingView indicators — the complete library. No single-tool download; you get the whole database.
Why full Kelly is too violent for trading
Full Kelly maximizes growth, but it says nothing about the ride. The ride is savage. A full-Kelly bettor should expect deep drawdowns as a matter of course. The chance of halving your capital before doubling it is meaningful, not remote. Few traders can execute a system while watching half their account disappear. Most abandon the plan at the bottom, which locks in the damage.
Half Kelly is the standard compromise. It keeps about three quarters of the long-run growth at far less variance. Drawdowns shrink dramatically while the compounding engine keeps running. That trade-off is so favorable that even Kelly enthusiasts rarely bet full size.
The second problem is estimation error. The formula assumes you know W and R exactly. You never do. Your winning percentage and payoff ratio are noisy estimates from a limited sample. A 50-trade record can easily overstate a true 45% system as 55%. Feed an optimistic W into the formula and you are no longer betting Kelly. You are betting beyond it, in the zone where growth falls and ruin risk climbs. Since overbetting hurts far more than underbetting, the rational response to uncertainty is to shade down. Fractional Kelly is not timidity. It is the correct answer to inputs you cannot fully trust.
Kelly in practice: cap risk at 0.5% to 2% per trade
Here is how to use the output without getting hurt. Treat Kelly as a ceiling and a signal, never as a literal position size. Then cap your actual risk at 0.5% to 2% per trade, regardless of what the formula says. If quarter Kelly comes out at 4%, trade 1% and let the edge compound quietly. If quarter Kelly comes out below 1%, your edge is thin, so trade the low end of the range or not at all.
The cap exists because real trading adds risks the formula cannot see. Correlated positions, news spikes, slippage, and regime changes all stack on top of the modeled bet. A fixed 0.5% to 2% cap keeps any single surprise survivable. Once you pick a percentage, convert it into a lot size with the position size calculator. Then test the full plan in the risk of ruin calculator, which shows the probability of hitting a drawdown limit at your chosen risk. The three tools work as a chain: Kelly for direction, the cap for safety, position sizing for execution.
Kelly fractions at a glance
The table below runs the formula across common winning percentages and payoff ratios. Negative results are shown as no edge. Notice how fast the fraction moves with small input changes. That sensitivity is exactly why noisy estimates are dangerous.
| Winning % | Payoff ratio 1.0 | Payoff ratio 1.5 | Payoff ratio 2.0 |
|---|---|---|---|
| 40% | No edge (−20.0%) | No edge (0.0%) | 10.0% |
| 45% | No edge (−10.0%) | 8.3% | 17.5% |
| 50% | No edge (0.0%) | 16.7% | 25.0% |
| 55% | 10.0% | 25.0% | 32.5% |
| 60% | 20.0% | 33.3% | 40.0% |
Read the table alongside the breakeven calculator. The no-edge cells on the left are systems sitting below their breakeven winning percentage. Together, Kelly and breakeven describe the same boundary.
Kelly versus fixed fractional risk
Fixed fractional sizing risks the same percentage on every trade, such as 1%. Instead, Kelly scales risk to the measured edge. When the edge is strong, Kelly bets bigger. When it weakens, Kelly cuts back automatically. That sounds smarter, and with perfect inputs over a long horizon it is. Real accounts live in a finite, noisy world, and that changes the ranking.
Fixed fractional wins on robustness. It needs no probability estimates, so a bad sample cannot poison it. Its drawdowns are steadier and easier to plan around. A practical middle path combines the two. Run fixed fractional as your base, and let Kelly act as a monitor inside the cap. If quarter Kelly sits far above your fixed 1%, your edge supports that size comfortably. If quarter Kelly drops below your fixed size, cut risk before the account forces the decision. Used this way, the calculator becomes a monthly check. Recompute it from your journal and watch the trend, not the level.
Honest limitations of the Kelly criterion
Kelly assumes you know your probabilities and that trades are independent. Markets give you neither. Your true winning percentage drifts as volatility, spreads, and market regimes change. A number measured in a trending year can collapse in a ranging one. The formula has no way to know that.
Independence fails too. Forex positions correlate through shared currencies and shared sessions. Three EUR trades open at once are closer to one triple-sized bet than three separate bets. Kelly sized each one alone, so your real exposure is far above the formula's intent. Fat-tailed moves, gaps, and slippage add losses larger than your average loss, which quietly breaks the payoff ratio input as well.
Sample size is the last trap. Even 100 trades leave wide error bars on both inputs. Keep logging every trade in a trade journal, re-run the numbers quarterly, and stress-test the downside with the drawdown calculator. This is the same test-before-trust standard applied to every tool on this site, documented in the Editorial and Testing Policy. Kelly is a useful compass. It is not a map.
Want the full toolkit? Get the complete database.
FAQ
What is a good Kelly fraction for forex trading?
There is no universal number, because it depends on your recorded winning percentage and payoff ratio. In practice, most traders who use Kelly at all trade a quarter of it or less. Then they cap the final figure at 0.5% to 2% of the account per trade.
Should I use full Kelly or half Kelly?
Half Kelly or less. Full Kelly maximizes theoretical growth but produces drawdowns most people cannot sit through. Half Kelly keeps about three quarters of the growth at far less variance, and it also cushions errors in your inputs.
What does a negative Kelly fraction mean?
It means the strategy loses money over time as measured. The formula tells you the optimal bet is zero. Fix the system first, then revisit sizing once your expectancy turns positive.
How many trades do I need before the inputs are reliable?
Treat 50 trades as the bare minimum and 100 or more as far better. Small samples overstate or understate your edge easily. Recalculate as your journal grows and expect the Kelly output to move.
Can the Kelly criterion make my trading profitable?
No. Kelly only scales an edge that already exists; it cannot create one, and applied to noisy estimates it can amplify losses instead. Test any sizing rule on a demo account first. Results are not guaranteed; past performance is not indicative of future results.
Related tools: atr position size calculator, risk reward calculator and portfolio heat calculator, plus the full free forex tools directory.
External references
Kelly criterion on Wikipedia · The Kelly Criterion at Investopedia