Calcady™
Home / Financial / Kelly Criterion Calculator

Kelly Criterion Calculator

Use the exact Kelly Criterion formula to mathematically determine the optimal percentage of your trading bankroll to risk on a single event to maximize compounding without ruin.

Trading Strategy Metrics

%

The exact percentage of times your setup wins over a large sample size.

1 :

If you risk $100 to make $250, enter 2.5.

Extremely High Risk

Suggested Risk (Half Kelly)

16.25%
Professional Standard Bankroll allocation
Volatility Curve Adjustments:
Full Kelly (Max Growth):32.5%
Quarter Kelly (Safe):8.13%
Email LinkText/SMSWhatsApp

What is The Mathematics of Optimal Bet Sizing?

Developed by John Kelly at Bell Labs in 1956, the Kelly Criterion calculates the exact optimal fraction of your bankroll to risk on a single trade or bet. It represents the absolute mathematical limit for maximizing the long-term compound growth rate of a portfolio while reducing the risk of total ruin to zero. In quantitative finance, it bridges the gap between having a statistical edge and mathematically surviving the variance required to execute that edge.

Mathematical Foundation

K%=W−(1−W)RK\% = W - \frac{(1 - W)}{R}
K%K\%= The Kelly Fraction: The exact percentage of your current bankroll you should risk.
WW= Your Historical Win Probability (e.g., 0.55 for 55%).
RR= Your Reward-to-Risk Ratio (e.g., 2.0 if you risk $1 to make $2).

Laws & Principles

  • The Negative Edge Law: If the formula outputs 0% or a negative number, your trading system has a mathematically negative expected value. If you trade this system, you are mathematically guaranteed to eventually go bankrupt. You must improve your win rate or hold out for larger winners.
  • The Over-Betting Death Spiral: If Kelly outputs 10%, and you bet 20%, you will eventually go bankrupt. Betting above Full Kelly drastically increases 'Drawdown Drag' — forcing your portfolio into holes so deep that regular compounding mathematics cannot dig you out.
  • Dynamic Resizing: Kelly is a fraction of your current bankroll, not your starting bankroll. As you lose money, your actual dollar risk scales down automatically, which is why it mathematically prevents total ruin.

Step-by-Step Example Walkthrough

" A disciplined day trader backtests a system. Their historic Win Rate is a mediocre 40% (W = 0.40). However, their risk management is elite: they cut losses at $50 and let winners run to $150, creating a 3.0 Reward-to-Risk ratio (R = 3.0). "

  1. Identify variables: W = 0.40, R = 3.0.
  2. Calculate 'Loser Drag': (1 - W) / R = (1 - 0.40) / 3.0 = 0.60 / 3.0 = 0.20.
  3. Subtract from Win Rate: 0.40 (Win Rate) - 0.20 (Loser Drag) = 0.20.
  4. Output Kelly Percentage: 0.20 = 20.0%.
Final Result: The trader's Full Kelly is an aggressive 20.0%. On a $100k account, they are mathematically cleared to risk exactly $20,000 on the next setup without fearing systematic ruin, despite losing 60% of their trades.

Quick Answer: How does the Kelly Criterion Calculator work?

The Kelly Criterion Calculator merges your historical Win Rate with your average Reward-to-Risk Ratio to calculate optimal bet sizing. The engine outputs the exact, mathematically safe percentage of your portfolio you should systematically risk per trade to hit maximum long-term compounding speeds while fundamentally guarding against account blowups during losing streaks.

Optimal Risk Equation

Standard Kelly Formula

Kelly % = WinProbability − (LossProbability / RewardRatio)

⚠ The Variance Warning

The mathematics of the Kelly Criterion assume that when you lose, you lose exactly what you expect. In the real world, massive overnight stock gaps or sudden illiquidity can cause your stop-loss to 'slip', resulting in a loss much larger than your planned risk. If you bet Full Kelly, a single severe slippage event can destroy your portfolio. Always use fractional Kelly.

System Expectancy Scenarios

✓ The High-Probability Grind

High Win Rate | Low R-Multiple

  1. Model: A high-frequency market maker wins a staggering 70% of his executions.
  2. Data: He scalps for small targets, risking $10 to make $10 (1.0 Reward Ratio).
  3. Calculation: 0.70 - ((1 - 0.70) / 1.0) = 0.40.

→ Kelly validates betting a terrifying 40% of his bankroll per trade. Because his win rate is astronomically high, the sheer volume of his wins mathematically protects him from deep drawdowns.

✗ The Home-Run Swing

Low Win Rate | Massive Payout

  1. Model: A venture capital investor funds chaotic startups, succeeding on only 15% of his checks.
  2. Data: When a startup succeeds, it returns 10 times his investment (10.0 Reward Ratio).
  3. Calculation: 0.15 - ((1 - 0.15) / 10.0) = 0.065.

→ Despite being wrong 85% of the time, his system has a positive edge. The Kelly strictly limits him to risking 6.5% per startup, ensuring he survives the inevitable 85% failure streak.

Fractional Kelly Applications

Risk Tier Characteristics
Full Kelly (100%)Maximum growth speed, but suffers brutal 30-50% equity drawdowns.
Half Kelly (50%)Sacrifices ~25% of compounding speed, but suppresses 75% of the volatility.
Quarter Kelly (25%)Slow, steady growth curve with virtually nonexistent drawdown risk.

Execution & Risk Management Tactics

Do This

  • ✓Use a massive sample size. A minimum of 100 closed trades is required to calculate a valid Kelly percentage. If you only have 5 trades, your win rate will be statistically invalid due to short-term variance. Garbage data into the Kelly formula will output mathematically dangerous bet sizes.
  • ✓Default to Half Kelly. Because human traders make execution errors, suffer slippage, and deal with changing market regimes, the theoretical 'Full Kelly' is too volatile for reality. Half Kelly provides the optimal balance of compounding velocity and psychological safety.

Avoid This

  • ✗The Fixed Percentage Fallacy. Books often advise \"always risk exactly 2% of your account.\" If your Win Rate is 40% and Reward Ratio is 1.2, your true Kelly is negative. Risking exactly 2% on that system absolutely guarantees you will blow up. Fixed percentages cannot save a weak statistical edge.
  • ✗Ignoring correlation. If the Kelly tells you to risk 5%, you cannot buy 5% Apple, 5% Microsoft, and 5% Google simultaneously. They are highly correlated — in a market crash, they will all trigger stop-losses simultaneously, meaning your true risk was 15%.

Frequently Asked Questions

What happens if I bet exactly double the Kelly Criterion?

If you consistently bet more than the Full Kelly threshold (e.g., betting 25% when Kelly is 12%), you enter the theoretical zone of 'Mathematical Ruin'. The extreme volatility of your losses will actively outpace your positive edge. In the long run, even with an incredibly profitable trading strategy, betting double Kelly will drive your account perfectly to $0.00.

Why use Quarter Kelly instead of Half Kelly?

If you are utilizing heavy leverage, trading in exceptionally illiquid markets prone to flash crashes, or simultaneously holding multiple correlated positions (e.g., holding 5 different tech stocks), your true risk is much higher than mathematically modeled. Quarter Kelly introduces massive psychological comfort and shields your bankroll against unpredictable black swan volatility shocks.

How many past trades do I need to calculate a valid Kelly percentage?

A completely minimum sample size is 100 closed trades, but aggressive quants prefer up to 300. If you only have 6 trades, your win rate will be incredibly skewed (possibly showing an 83% win rate due to simple variance). If you plug bad data into the Kelly formula, it will arrogantly tell you to bet 50% of your account, guaranteeing your own destruction when the variance balances out.

What does it mean if my Kelly percentage is negative?

A negative Kelly indicates that your trading system lacks a mathematical edge. Your combination of win rate and reward-to-risk ratio results in a negative Expected Value over time. A negative Kelly is a mathematical directive telling you to immediately stop trading that setup — no money management system can turn a negative expectation into a long-term profit.

Related Risk & Expectancy Tools