Kelly Criterion Calculator
The Kelly Criterion Calculator finds the mathematically optimal fraction of your account to risk per trade to maximize long-term compound growth. It applies the formula K% = W − (1 − W) / R, where W is win rate (decimal) and R is your reward-to-risk ratio. It also computes a recommended fractional Kelly (default Half-Kelly, 0.5) and per-trade expectancy in R.
- Core formula: K% = W − (1 − W) / R, where W = win rate as a decimal and R = average win ÷ average loss. Example: 55% win rate at 1:1.5 R:R gives 0.55 − 0.45/1.5 = 0.25 (25%).
- The tool defaults to Half-Kelly (fraction 0.5), which retains roughly 75% of full-Kelly growth while cutting drawdown severity by about half — the recommended setting for most traders.
- Recommended size = full Kelly × your chosen fraction (1.0 Full, 0.5 Half, 0.25 Quarter). It also shows expectancy in R-multiples: W × R − (1 − W).
- A negative or zero Kelly means no edge — the calculator flags 'Don't Trade' and outputs N/A; values above 50% are flagged 'Extreme — Likely Overestimating Edge.'
- Betting at 2× Kelly drops expected geometric growth to zero, and above 2× it turns negative — so over-betting is far more dangerous than under-betting.
Calculate the mathematically optimal position size to maximize long-term growth. The Kelly Criterion tells you exactly how much to risk — and why most traders should use Half-Kelly.
How to Use the Kelly Criterion Calculator
Choose Your Input Mode
"Win Rate + R:R Ratio" is simpler — just enter your win percentage and average reward-to-risk ratio. "Win/Loss Amounts" lets you enter dollar amounts directly if you track your trades that way.
Enter Your Win Rate
Use your actual trading statistics from at least 100 trades. Be conservative — if you measured 60%, consider entering 55% to account for estimation error. Overestimating your edge is the most common Kelly mistake.
Set Your Risk:Reward Ratio
Divide your average winning trade by your average losing trade. If your average win is $150 and average loss is $100, your R:R is 1:1.5. Use actual results, not planned targets.
Choose Your Kelly Fraction
Start with 0.50 (Half-Kelly). This provides ~75% of optimal growth with dramatically lower drawdowns. Only use Full Kelly (1.0) if you have extreme confidence in your edge estimates.
Analyze the Results
Check the comparison table to see how different fractions perform over 1,000 trades. The equity curve chart shows the same trade sequence with different sizing — notice how Full Kelly has the wildest swings.
The Kelly Criterion Explained
The Kelly Criterion was developed by John Kelly at Bell Labs in 1956 for optimizing signal transmission. It was quickly adopted by gamblers and investors as the mathematically optimal strategy for sizing bets to maximize long-term wealth growth.
K% = W - [(1 - W) / R]
// Where:
W = Win rate (as decimal)
R = Win/Loss ratio (avg win ÷ avg loss)
K% = Fraction of capital to risk
// Example: 55% win rate, 1:1.5 R:R
K% = 0.55 - (0.45 / 1.5)
K% = 0.55 - 0.30 = 0.25 (25%)
// Half-Kelly recommendation
Suggested = 25% × 0.5 = 12.5%
The result tells you the fraction of your account to risk on each trade. Kelly maximizes the expected value of the logarithm of wealth — which means it maximizes geometric (compound) growth, not arithmetic growth.
Why Full Kelly Is Too Aggressive for Most Traders
Full Kelly maximizes long-term growth rate, but the ride is brutal. In theory, you'll end up richer than any other strategy. In practice, drawdowns of 50-80% are routine, and most traders — even professional ones — will abandon the strategy during these periods.
| Kelly Fraction | Growth Rate | Typical Max Drawdown | Psychological Impact |
|---|---|---|---|
| Full Kelly (1.0) | 100% (maximum) | 50-80% | Unbearable for most |
| Half Kelly (0.5) | ~75% of full | 25-40% | Tough but manageable |
| Quarter Kelly (0.25) | ~56% of full | 10-20% | Comfortable for most |
| Tenth Kelly (0.1) | ~34% of full | 5-10% | Very smooth equity |
The Half-Kelly Sweet Spot
Half-Kelly retains 75% of the maximum growth rate while roughly halving the drawdown. This is often called the "sweet spot" because you sacrifice relatively little growth for dramatically less pain. It's the most popular choice among quantitative traders.
Half-Kelly and Fractional Kelly Strategies
Fractional Kelly means using a percentage of the full Kelly recommendation. If Full Kelly says 20%, Half-Kelly says 10%, Quarter-Kelly says 5%.
The growth rate of any Kelly fraction f relative to full Kelly follows a parabolic curve. This has an important implication: the first half of Kelly costs relatively little growth, but the second half costs a lot of drawdown.
| Fraction | Relative Growth | What You Give Up | Best For |
|---|---|---|---|
| 0.10 (Tenth) | 34% | 66% of max growth | Institutional, multi-strategy portfolios |
| 0.25 (Quarter) | 56% | 44% of max growth | Conservative traders, beginners |
| 0.50 (Half) | 75% | 25% of max growth | Most active traders (recommended) |
| 0.75 (Three-Quarter) | 94% | 6% of max growth | Aggressive traders with high confidence |
| 1.00 (Full) | 100% | Nothing (but maximum volatility) | Theoretical optimum only |
When Kelly Goes Wrong — Overestimating Your Edge
The Kelly formula assumes you know your exact win rate and R:R ratio. In reality, these are estimates from limited data. The danger is that overestimating your edge leads to oversizing, and the penalty for betting above Kelly is severe.
The Asymmetry of Error
If your true Kelly is 10% but you calculate 20%, betting 20% puts you at 2x Kelly — where expected growth drops to zero. If you underestimate and bet 5% (half of true Kelly), you still get 75% of optimal growth. Under-betting is much safer than over-betting. This is the strongest argument for fractional Kelly.
Common sources of edge overestimation:
- Small sample size: 30 trades can easily show 60% win rate when true rate is 50%.
- Curve fitting: Optimized backtests inflate apparent edge.
- Regime change: Past win rates don't guarantee future performance.
- Ignoring costs: Spreads, commissions, and slippage reduce actual R:R.
Kelly vs. Fixed Fractional Position Sizing
| Feature | Kelly Criterion | Fixed Fractional (e.g., 1%) |
|---|---|---|
| Adapts to edge? | Yes — sizes up with stronger edge | No — same size regardless of edge |
| Requires accurate stats? | Yes — sensitive to estimation errors | No — robust to inaccuracy |
| Theoretical growth | Maximum (proven optimal) | Suboptimal but predictable |
| Drawdown behavior | Can be extreme at full Kelly | Predictable and bounded |
| Complexity | Requires ongoing statistics tracking | Simple — one number |
| Best for | Quantitative traders with robust stats | Most retail traders |
In practice, many traders combine both approaches: use Kelly to determine an upper bound, then cap position size at a fixed fraction (e.g., never exceed 2% per trade regardless of Kelly output).
Real-World Application in Forex Trading
Applying Kelly to forex requires honest self-assessment. Here's a practical workflow:
- Track at least 100 trades with consistent position sizing and strategy rules.
- Calculate your actual win rate and average R:R — not from backtesting, from live trading.
- Apply the Kelly formula with conservative estimates (reduce measured win rate by 5%).
- Use Half-Kelly or less as your position size.
- Cap at 2-3% per trade regardless of Kelly output.
- Recalculate quarterly as your trading stats evolve.
Practical Tip
If your Kelly calculation says you should risk more than 5%, you're probably overestimating your edge. Very few retail forex strategies have edges large enough to justify 5%+ risk per trade. When Kelly suggests extreme sizing, it's a signal to re-examine your statistics, not to trade bigger.
The Math: Why Kelly Maximizes Geometric Growth
Kelly maximizes E[log(W)], the expected value of the logarithm of wealth. This is equivalent to maximizing the geometric growth rate — the rate at which your account actually compounds over many trades.
G(f) = W × log(1 + f×R) + (1-W) × log(1 - f)
// Taking derivative and setting to zero:
dG/df = W×R/(1 + f×R) - (1-W)/(1-f) = 0
// Solving for f gives the Kelly formula:
f* = W - (1-W)/R = (W×R - (1-W)) / R
// Growth rate at fraction f relative to full Kelly:
// At Half-Kelly (f* × 0.5): ~75% of max growth
// At 2× Kelly: growth drops to ZERO
// Above 2× Kelly: negative expected growth
The critical insight: the growth function G(f) is a concave parabola. It rises from 0, peaks at the Kelly fraction, and drops back to 0 at twice Kelly. Beyond 2× Kelly, you're actively destroying wealth. This is why over-betting is so dangerous.
Frequently Asked Questions
-
The Kelly Criterion is a mathematical formula that calculates the optimal percentage of capital to risk on a bet or trade to maximize long-term geometric growth. Developed by John Kelly at Bell Labs in 1956, it balances the tradeoff between betting too much (risking devastating drawdowns) and too little (leaving growth on the table). It's widely used in professional gambling, hedge funds, and quantitative trading.
-
Kelly % = W - (1-W)/R, where W is your win rate as a decimal and R is your reward-to-risk ratio (average win divided by average loss). For example, with a 60% win rate and 1:1 R:R: Kelly = 0.60 - 0.40/1.0 = 0.20, meaning 20% of your account per trade. Most traders then apply a fraction (typically 0.5 for Half-Kelly) to reduce drawdown risk.
-
Full Kelly produces maximum long-term growth but with extreme drawdowns — often 50%+ peak-to-trough swings. Half-Kelly retains roughly 75% of the growth rate while cutting drawdown severity approximately in half. Most professional traders and hedge funds use Half-Kelly or less because: (1) the psychological pain of full-Kelly drawdowns causes most traders to abandon the strategy, (2) estimation errors in win rate/R:R are partially compensated, and (3) the growth sacrifice is relatively small.
-
A negative Kelly means your strategy has no mathematical edge — you lose money on average. The formula is telling you not to trade this system at all. Either improve your win rate, increase your R:R ratio, or find a different strategy before risking real capital. A negative Kelly is actually valuable information — it prevents you from slowly bleeding your account on a losing system.
-
Yes, but with important caveats. Kelly assumes you know your exact win rate and R:R ratio, which in forex are always estimates from limited data. Over-estimating your edge leads to dangerous over-sizing. Best practices: use fractional Kelly (25-50%), base calculations on conservative estimates from at least 100+ live trades, cap position size at 2-3% regardless of Kelly output, and recalculate quarterly as your statistics evolve.
-
The standard formula is K% = W - [(1-W)/R] where K% is the optimal fraction to bet, W is win probability (as decimal), and R is the win/loss ratio (average win / average loss). An equivalent form used in gambling is K% = (BP - Q)/B where B is net odds received on a win, P is win probability, and Q is loss probability (1-P). Both produce the same result.
-
At Full Kelly, the theoretical risk of ruin is 0% with infinite trades because bet size shrinks proportionally with account size — you can never hit exactly zero. However, short-term drawdowns can exceed 50%, which feels like ruin in practice. At Half-Kelly, drawdowns are roughly halved. Quarter-Kelly provides very smooth equity curves with minimal drawdown risk. The key insight: Kelly inherently prevents total ruin but not painful drawdowns.
-
Most professionals use 25-50% of the Kelly recommendation (Quarter to Half Kelly). Some hedge funds use even less — 10-20% of Kelly — because they prioritize capital preservation and investor comfort over maximum growth. Edward Thorp, who popularized Kelly in financial markets, has recommended Half-Kelly as the practical optimum. Very few successful traders use Full Kelly in practice.
-
Absolutely, as long as your R:R ratio compensates. A 35% win rate with a 1:3 R:R gives Kelly = 0.35 - 0.65/3 = 0.133 (13.3%). The requirement is positive expectancy: Win Rate × R:R must exceed (1 - Win Rate). Many trend-following strategies have win rates of 30-40% but excel because their winners are 3-5x their losers. Kelly works for any win rate and R:R combination that produces a positive edge.
-
Betting more than Full Kelly actually reduces your long-term growth rate — counterintuitively, you grow slower by betting bigger. At exactly 2× Kelly, your expected geometric growth drops to zero — the same as not trading. Above 2× Kelly, you are mathematically guaranteed to lose money over time despite having a positive edge. This is called "over-betting" and demonstrates why position sizing discipline matters more than having an edge.
-
At minimum 100 trades, ideally 200-500+ from live (not backtested) trading. With 50 trades, your win rate estimate has roughly ±10% error, which translates to massive Kelly estimation error. With 200+ trades, confidence intervals tighten significantly. Always use conservative estimates — if your measured win rate is 60%, consider using 55% in the formula. The cost of under-sizing is low growth; the cost of over-sizing can be catastrophic.
-
Fixed fractional sizing (e.g., always risk 1%) ignores your edge entirely — you risk the same percentage whether your strategy has a 60% or 51% win rate. Kelly sizing adapts to your edge: stronger edge means larger positions. Kelly is theoretically optimal but requires accurate edge estimates. Fixed fractional is simpler and more robust to estimation errors. Many professionals combine both: use Kelly to determine a ceiling, but cap at a fixed fraction (e.g., never exceed 2%) for safety.
Compare Top Forex Brokers
Trading CFDs carries significant risk of loss. Broker information is for comparison purposes only.