The Kelly formula determines the exact percentage of your bankroll to risk on each bet or investment by balancing the probability of winning against the potential payout. This mathematical approach, developed at Bell Labs in 1956, protects you from betting too much while ensuring you don’t miss growth opportunities by betting too little.
Core Formula and Definitions
The Kelly bet uses a specific formula to calculate your optimal wager size. For simple bets where you either win or lose everything, the formula is: f* = (bp – q) / b.
In this equation, f* represents the Kelly fraction—the portion of your bankroll you should wager. The variable b shows how much you gain relative to your bet (if you bet $10 and win $20, then b = 2). The letter p stands for your win probability, while q equals your loss probability (which is simply 1 – p).
The formula maximizes log wealth over time rather than focusing on any single bet. This approach uses logarithmic utility, which means it prioritizes steady growth over risky attempts at huge gains.
When you have no edge (meaning the bet is perfectly fair), the Kelly criterion tells you to bet nothing. If the calculation gives you a negative number, you should take the opposite side of the bet.
Key Variables: Edge, Odds, and Probability of Winning
Your edge represents the advantage you hold in a betting situation. You calculate it by comparing your win probability to the odds being offered.
Consider a biased coin that lands on heads 60% of the time with even-money odds. Your win probability is 0.6, your loss probability is 0.4, and b equals 1. Plugging these into the formula gives you: f* = (1 × 0.6 – 0.4) / 1 = 0.2, or 20% of your bankroll.
The odds determine how much you receive when you win. Better odds mean you can bet a smaller fraction while maintaining optimal growth. Your probability of winning must be accurate for the Kelly formula to work properly—overestimating your chances leads to overbetting and potential ruin.
The expected value alone doesn’t tell you how much to bet. Two bets might have identical expected returns but require different position sizes based on their odds and win probabilities.
Historical Background and Development
John Larry Kelly Jr., a researcher at Bell Labs, published the Kelly criterion in 1956. He originally developed it to solve problems with long-distance telephone signal noise, but the formula’s applications extended far beyond telecommunications.
Kelly built upon work by Bernoulli and connected his betting formula to information theory. This link showed that optimal betting strategies share mathematical foundations with how information travels through communication systems.
The formula gained wider attention through Fortune’s Formula, a book by William Poundstone that documented how professional gamblers and investors applied the Kelly criterion. The law of large numbers underpins why the strategy works—over many bets, your results converge toward the expected probabilities, and the Kelly fraction maximizes your geometric growth rate during this process.
Applying the Kelly Criterion in Gambling and Investing
The Kelly Criterion guides you to calculate specific percentages of your bankroll to wager or invest based on your edge and odds. This mathematical approach addresses both optimal bet sizing and risk management through various implementation strategies, while accounting for real-world constraints like estimation error and transaction costs.
Optimal Bet Sizing and Bankroll Management
The Kelly Criterion formula determines your optimal bet size by calculating the percentage of your bankroll to risk on each opportunity. You multiply your win probability by the odds, subtract the loss probability, then divide by the odds. This gives you the exact position sizing for maximum geometric mean growth over time.
Your bankroll management becomes systematic rather than emotional. In sports betting, if you have a 55% win probability with even money odds, the Kelly formula tells you to bet 10% of your bankroll. For investing, the same principle applies to portfolio weights across different assets.
The strategy maximises your expected growth rate while preventing risk of ruin. You never bet more than your edge justifies, which protects against overbetting. Online casinos and traditional betting markets both respond to this approach, though you must account for the house edge reducing your actual win probability.
Key metrics for optimal betting:
- Win probability (W)
- Win/loss ratio (R)
- Current bankroll size
- Kelly percentage = W – [(1-W)/R]
Fractional Kelly Strategies and Risk Management
Full Kelly betting creates substantial volatility and large drawdowns that most people cannot tolerate psychologically or financially. Fractional Kelly approaches reduce variance of return by betting a portion of what the formula suggests.
Half-Kelly means you bet 50% of the calculated optimal bet, which significantly decreases your standard deviation while maintaining strong long-term returns. Simulations show that half-Kelly reduces your expected growth rate by only 25% but cuts variance by 50%. This trade-off improves your Sharpe ratio and makes the betting strategy more practical.
Quarter-Kelly goes even further in risk management, creating smoother equity curves with fewer dramatic swings. You sacrifice some growth potential but gain consistency and reduced stress. Many professional gamblers and value investors prefer these fractional approaches over full Kelly.
Double Kelly, betting twice the formula amount, dramatically increases leverage and risk. This aggressive approach leads to higher volatility and greater chances of devastating drawdowns. Most experts warn against exceeding full Kelly under any circumstances.
Limitations and Practical Considerations
The Kelly Criterion assumes you know your true win probability and odds with perfect accuracy. Estimation error represents the biggest practical challenge since overestimating your edge by even small amounts causes significant overbetting and potential losses.
Transaction costs reduce your effective returns, especially with frequent rebalancing. Short selling constraints, correlation between positions, and varying rates of return across assets complicate the mathematics beyond simple Kelly calculations. Portfolio diversification requires more complex approaches like quadratic programming to determine optimal portfolio allocation across multiple positions.
Expected utility theory suggests that maximising geometric growth may not align with your personal risk tolerance. You might prefer lower volatility even if it means accepting a reduced expected growth rate compared to the growth optimal portfolio.
Practical implementation challenges:
- Accurately estimating win probabilities from historical data
- Accounting for changing market conditions and odds
- Managing correlation between multiple positions
- Incorporating the risk-free rate into calculations
- Building models in Excel or other tools for complex scenarios
- Balancing geometric Brownian motion assumptions with real market behaviour
You need to reassess your inputs regularly and adjust position sizing as your bankroll changes and new information emerges about your true edge.
Real-World Use Cases and Impact
The Kelly Criterion has moved from theoretical mathematics into actual investment practice, where professional money managers use it to size positions and build portfolios. Notable investors have applied Kelly-based strategies to achieve long-term growth while managing risk in their investment decisions.
Applications in Portfolio Construction
When you apply the Kelly Criterion to portfolio construction, you’re calculating what percentage of your capital should go into each investment based on your edge and the odds of success. The formula helps you build a growth-optimal portfolio that maximizes your expected return over time without risking ruin.
Institutional investors face unique challenges when implementing Kelly strategies. Many prioritize maintaining specific risk levels over maximizing returns regardless of risk. This means they often use a fractional Kelly approach, betting less than the full Kelly ratio suggests.
The Kelly strategy works best when you can:
- Accurately estimate win probabilities
- Calculate realistic reward-to-risk ratios
- Rebalance positions as market conditions change
- Accept short-term volatility for long-term growth
Practical implementation requires complex statistical analysis, especially in markets where odds aren’t as clear as casino games. You need to continuously update your probability estimates as new information becomes available.
Famous Practitioners and Case Studies
Edward Thorp stands as the most famous Kelly Criterion practitioner. He used the formula to beat blackjack casinos, then applied those same principles to build a successful hedge fund. His work inspired the book Fortune’s Formula, which documented how Kelly-based betting could generate superior long-term returns.
Renaissance Technologies, George Soros, and Stanley Druckenmiller have used Kelly-like models in their trading strategies. These managers scale their positions dynamically based on their perceived edge and market volatility. Bill Gross also incorporated position-sizing principles similar to Kelly during his tenure managing bond portfolios.
The approach differs from Warren Buffett’s concentrated value investing style. While Buffett makes large bets on high-conviction ideas, Kelly practitioners calculate precise position sizes based on expected utility and probability estimates. Both methods aim for long-term growth, but Kelly offers a mathematical framework for risk management that adapts to changing market conditions.