Last updated: May 17, 2026
Capital Defense

Kelly Criterion: The Mathematical Limit of Risk

Trade-Charts IntelUpdate 2026.03

The Logic of the Betting Limit: What is the Kelly Criterion?

Developed in 1956 by John L. Kelly at Bell Labs, the Kelly Criterion is a formula used to determine the optimal size of a series of bets. In algorithmic trading, it is the 'Golden Rule' of position sizing. It tells you exactly what percentage of your account to risk on every trade to achieve the maximum possible long-term growth of your capital.

Unlike a fixed 2% risk (which is a generic retail rule), the Kelly Criterion is dynamic. It adjusts your risk based on the specific Win Rate and Risk-Reward Ratio of your strategy. If your EA has a 70% win rate and a 1:1 reward ratio, the Kelly formula will tell you to risk a much larger percentage than an EA with a 30% win rate and a 1:5 reward ratio.

The Formula: (Win Rate - (Loss Rate / RR Ratio))

The most common version of the formula for traders is: Kelly % = W - [(1 - W) / R]. Where W is the win rate (e.g., 0.55 for 55%) and R is the risk-reward ratio (e.g., 2.0 for a $2 win vs $1 loss).

If the result is positive, the system has a 'Positive Expectancy' and is tradable. If the result is negative, the system will mathematically blow up the account over time, regardless of what lot sizing you use. The Kelly % is the 'Speed Limit' of your account growth. If you risk more than the Kelly %, your account's 'Compounded Growth' will actually decrease as you increase risk.

⚠️Danger Zone / Risk Check

Kelly Criterion Execution Rules

  • Primary: Calculate the Win Rate (W) and Reward Ratio (R)

  • Formula: W - ((1 - W) / R) = Optimal Risk Fraction

  • Threshold: Always use 'Half-Kelly' or 'Fractional Kelly' for safety

  • Constraint: Never risk more than the Full-Kelly result

  • Market State: Account for slippage in your Reward Ratio calculation

  • Verified: Recalculate the Kelly % every 50-100 trades for accuracy

The 'Half-Kelly' Advantage: Avoiding the Blowout

The 'Full-Kelly' percentage is mathematically perfect but practically dangerous. In the real market, your Win Rate is never 100% stable, and slippage can occur. If the Kelly formula tells you to risk 10% of your account per trade, a small losing streak will result in a massive emotional and financial drawdown.

Professional quantitative developers almost always use 'Half-Kelly' (50% of the calculated result) or 'Fractional Kelly'. This significantly reduces the volatility of the equity curve while still capturing most of the compounded growth benefits. It creates a 'Safety Buffer' against the uncertainty of future market conditions.

MT4 Implementation: Coding the Kelly Filter

In MQL4/MQL5, an EA can calculate its own Kelly % based on its AccountHistory(). By scanning the last 100 closed trades, the EA can determine its current win rate and average win/loss ratio. It then automatically updates the AutoLot function using the Half-Kelly formula. This allows the bot to 'Self-Correct' its risk level as its performance fluctuates over different market regimes.

Frequently Asked Questions

Is Kelly better than fixed 2% risk?

For a professional quant, yes. A fixed 2% risk is a 'one size fits all' rule that ignores the specific efficiency of your strategy. If your strategy has a high win rate, a 2% risk is actually 'Under-leveraging' your account. If your strategy is low-win-rate/high-reward, 2% might be 'Over-leveraging'. The Kelly Criterion provides a customized, mathematically optimal risk profile.

What if the formula gives a result of 20% risk?

This is a common issue with high-win-rate scalpers. Even if the math says 20% is optimal for growth, the Drawdown of such a risk would be intolerable for most humans. Professional traders usually set a 'Maximum Risk Cap' (e.g., 5%) regardless of what the Kelly formula suggests. Mathematical optimality must always be balanced with psychological survival.

Recommended Reading