The Kelly Criterion is a position sizing formula that calculates the optimal percentage of your capital to risk on each trade, based on your strategy’s historical win rate and average payoff ratio.

What Is the Kelly Criterion?

The Kelly Criterion is a mathematical formula developed by John Kelly Jr. in 1956. It was originally designed for signal processing but was quickly adopted by gamblers and later by traders and investors as a way to size bets optimally.

In trading, the Kelly Criterion answers a specific question: given what you know about a strategy’s performance, what fraction of your capital should you risk on the next trade to maximise long-term growth?

The formula is straightforward:

Kelly % = W – [(1 – W) / R]

Where:

• W = win rate (percentage of trades that are profitable)
• R = reward-to-risk ratio (average winning trade divided by average losing trade)

For example: a strategy with a 55% win rate and a 1.5:1 reward-to-risk ratio gives Kelly % = 0.55 – (0.45 / 1.5) = 0.55 – 0.30 = 0.25. Full Kelly suggests risking 25% of capital per trade.

Why the Kelly Criterion Matters for Systematic Traders

Position sizing is one of the most underestimated variables in trading performance. Two strategies with identical win rates and payoff ratios will produce wildly different results depending on how much capital is risked per trade.

Size too small and you underperform — your edge barely compounds. Size too large and a losing streak destroys your account before the edge has time to play out. The Kelly Criterion sits at the mathematical optimum between these two failure modes.

For algorithmic traders, Kelly matters for a second reason: systematic strategies produce a reliable historical record of W and R. Unlike discretionary traders who struggle to estimate their own performance accurately, algorithmic traders can calculate Kelly inputs directly from their backtest data.

Should You Use Full Kelly?

In theory, full Kelly maximises long-term geometric growth. In practice, most traders use a fraction of it — commonly half Kelly or quarter Kelly.

There are three reasons for this:

• Estimation error: Your backtest win rate and payoff ratio are estimates. If your real-world performance is slightly worse than backtested, full Kelly oversizes your positions significantly.
• Drawdown tolerance: Full Kelly produces large drawdowns even when working correctly. Most traders cannot hold positions through a 40–50% equity drawdown without abandoning the strategy.
• Out-of-sample performance: Strategies often perform slightly worse live than in backtesting. Half Kelly gives you a buffer against this degradation.

Half Kelly — risking half the calculated percentage — retains around 75% of the theoretical growth rate while reducing drawdowns substantially. For most algorithmic traders, this is the practical sweet spot.

What Are the Limits of the Kelly Criterion?

Kelly assumes your win rate and payoff ratio are fixed and known. In real markets, both shift over time as market regimes change.

It also assumes independence between trades — that one trade’s outcome does not affect the next. Correlated positions in similar instruments violate this assumption and can lead to oversizing during periods when multiple trades move together.

Kelly does not account for transaction costs, slippage, or leverage constraints. A raw Kelly calculation on a high-frequency strategy ignoring fees will overestimate the viable position size considerably.

Finally, Kelly optimises for long-term geometric growth — not for any individual trader’s risk tolerance or drawdown limit. A strategy with a valid Kelly allocation of 30% is mathematically sound but practically unusable for most retail traders.

How to Apply the Kelly Criterion in Arrow Algo

Arrow Algo’s backtest reports give you the two numbers the Kelly formula needs: your win rate and your average reward-to-risk ratio. Both are visible directly in your results after running a backtest.

Once you have your Kelly percentage, apply it through Arrow Algo’s position sizing controls. Set your risk per trade as a percentage of equity, using half or quarter Kelly rather than full Kelly for live trading.

A sensible workflow:

1. Run a backtest over a representative period and note the win rate and average win/loss ratio from the results.
2. Calculate your Kelly percentage using the formula above.
3. Apply half Kelly as your starting position size in live or paper trading.
4. Re-run the calculation periodically as your strategy accumulates more live data — real-world W and R may differ from backtested values.
5. If live performance degrades significantly, reduce to quarter Kelly until you understand why.

Pairing Kelly sizing with a maximum drawdown limit is good practice. If your equity falls more than a set percentage from its peak, reduce position size further regardless of what Kelly suggests — this protects you during periods when your strategy’s edge may have temporarily disappeared.

For a deeper look at the metrics that feed into Kelly, see our guide on Profit Factor and how it relates to long-term strategy viability.

Key Takeaways

• The Kelly Criterion calculates the optimal position size to maximise long-term capital growth.
• It requires two inputs: win rate and reward-to-risk ratio — both available directly from Arrow Algo backtest results.
• Full Kelly is mathematically optimal but produces large drawdowns and is sensitive to estimation errors.
• Half Kelly is the practical standard — it retains most of the growth advantage with significantly lower drawdown risk.
• Kelly assumes fixed, independent trade outcomes — adjust downward for correlated positions or changing market regimes.
• Combine Kelly sizing with a maximum drawdown rule to protect your capital when strategy edge is uncertain.

Disclaimer: This content is for educational purposes only and does not constitute financial advice. Trading involves significant risk and you should only trade with capital you can afford to lose. Past performance is not indicative of future results. Always conduct your own research before making any trading decisions.

Ready to build your own automated trading strategies without writing a single line of code? Start for free at Arrow Algo and join thousands of traders who’ve made the switch to systematic trading.