In prediction markets, knowing what to bet on is only half the battle. The other half—often more crucial—is knowing how much to bet. The Kelly Criterion is a mathematical formula that determines the optimal fraction of your bankroll to wager on each bet in order to maximize long-term wealth growth while minimizing the risk of ruin. For traders on platforms like Kalshi and Polymarket, mastering Kelly is the difference between compounding gains and blowing up your account.
Originally developed by John Kelly Jr. at Bell Labs in 1956 and published in the Bell System Technical Journal, the formula has become the gold standard for bankroll management among professional gamblers, traders, and increasingly, prediction market participants.
What Is the Kelly Criterion for Prediction Markets?
The Kelly Criterion is a position-sizing formula that calculates the exact percentage of your bankroll to allocate to a bet, given your estimated edge over the market's implied probability. It was originally designed for information theory but was quickly adopted by gamblers and investors because it provably maximizes the expected logarithm of wealth—meaning it produces the highest long-run compounding rate of any fixed-fraction betting strategy.
Prediction markets are an especially strong fit for Kelly because contracts have binary outcomes, defined maximum payouts, and short resolution windows—all conditions that make the formula's inputs clean and calculable.
What Is the Kelly Criterion Formula?
The Kelly Criterion formula is:
f = (bp − q) / b
Where:
- f = fraction of bankroll to bet
- b = net odds received on the bet (profit per $1 wagered)
- p = your estimated probability of winning
- q = your estimated probability of losing (1 − p)
If f is negative, Kelly tells you there is no edge and you should not bet. If f exceeds 1, Kelly is effectively telling you the edge is enormous—but in practice, you should treat any result above ~25–30% with skepticism and apply fractional Kelly (covered below).
How Do You Apply the Kelly Criterion to a Prediction Market Contract?
The following example uses a real market type available on Kalshi to demonstrate how to plug numbers into the formula step by step.
Suppose you are analyzing a contract on whether the Federal Reserve will raise interest rates at their next meeting. The market prices the "Yes" contract at $0.40 (implying 40% probability), but your research suggests the true probability is 55%.
- Market price: $0.40 (40% implied probability)
- Your probability estimate (p): 0.55
- Your loss probability (q): 0.45
- Net odds (b): You pay $0.40 to win $1.00, so your profit on a win is $0.60. Net odds = 0.60 / 0.40 = 1.5
Plugging into the formula:
f = (1.5 × 0.55 − 0.45) / 1.5 = (0.825 − 0.45) / 1.5 = 0.375 / 1.5 = 0.25
Kelly recommends betting 25% of your bankroll on this contract—a large allocation that reflects the significant edge you have identified. In practice, most traders would apply a fractional Kelly multiplier and bet 6–13% instead (see below).
Why Is the Kelly Criterion Well-Suited to Prediction Markets?
The Kelly Criterion works especially well in prediction markets for three structural reasons:
Binary Outcomes Simplify Probability Estimation
Most prediction market contracts resolve as Yes or No, Over or Under. Binary outcomes make it far easier to assign a single probability estimate than in continuous markets like equities, where the range of outcomes is unbounded.
Defined Payouts Make "b" Precise
Unlike stocks where potential gains are theoretically unlimited, prediction market contracts pay out a fixed maximum (typically $1.00 on Kalshi and Polymarket). This means the "b" variable in the Kelly formula can be calculated exactly from the contract's current price, eliminating a major source of estimation error.
Short Resolution Windows Enable Rapid Compounding
Many prediction market events resolve within days or weeks—Fed meetings, election nights, earnings announcements. Faster resolution means capital can be recycled more frequently, and the compounding advantage of Kelly accumulates faster than in annual-horizon investment strategies.
What Is Fractional Kelly and Why Do Professionals Use It?
Fractional Kelly is a conservative variation of the Kelly Criterion in which traders bet only a set fraction—commonly 25% to 50%—of the full Kelly recommendation. Fractional Kelly reduces short-term volatility and protects against the most dangerous failure mode of full Kelly: overconfident probability estimates.
In our Fed rate example, full Kelly recommends a 25% bankroll allocation. Applying fractional Kelly:
- Half Kelly (50%): Bet 12.5% of bankroll
- Quarter Kelly (25%): Bet 6.25% of bankroll
Research by Thorp (2006) and others has shown that half Kelly captures roughly 75% of the long-run compounding benefit of full Kelly while cutting variance approximately in half—a highly favorable trade-off for most traders.
Fractional Kelly is especially important in prediction markets because it accounts for:
- Estimation uncertainty: Your probability model is never perfectly accurate
- Correlated positions: Multiple related markets can move together, amplifying effective exposure
- Transaction costs: Spreads and fees erode the edge that Kelly assumes is riskless
- Personal risk tolerance: Full Kelly can produce 50%+ drawdowns even when your edge is real
For a deeper treatment of fractional Kelly and portfolio-level applications, see our guide on Advanced Kelly Criterion: Fractional Kelly & Multi-Market Applications.
What Are the Most Common Kelly Criterion Mistakes in Prediction Markets?
Mistake 1: Overconfident Probability Estimates
The most dangerous Kelly error is overestimating your edge. If you believe an event has a 70% chance of occurring but the true probability is 50%, the formula will recommend a position size far larger than warranted—and repeated over-betting is the primary path to ruin. Always stress-test your estimates by asking: "What would I need to be wrong about for this probability to drop by 10 percentage points?"
Mistake 2: Ignoring Correlations Between Positions
The standard Kelly formula assumes each bet is independent, but prediction markets frequently feature correlated outcomes. If you hold positions on multiple political events, economic indicators, or markets tied to the same underlying variable, your true portfolio risk is higher than the sum of individual Kelly allocations suggests. For a framework to handle this, see our post on Dynamic Position Sizing: Beyond Kelly for Multi-Market Portfolios.
Mistake 3: Failing to Adjust for Transaction Costs
Kelly calculations should use your net edge after spreads and fees, not your gross edge. On thinly traded markets, the bid-ask spread alone can eliminate an edge that looks attractive on paper. Recalculate "b" and "p" using realistic fill prices before sizing any position.
Mistake 4: Applying Full Kelly to Illiquid Markets
In liquid markets you can exit a position if your view changes. In illiquid prediction markets, you may be locked in until resolution. Full Kelly sizing assumes you can rebalance continuously; when you cannot, the practical risk is much higher than the formula implies.
How Does Kelly Criterion Fit Into a Broader Risk Management Framework?
Kelly is a position-sizing tool, not a complete risk management system. It answers the question "how much to bet" but does not address which markets to enter, when to exit early, or how to manage overall portfolio drawdown limits. For a complete framework that integrates Kelly with stop-loss rules, diversification constraints, and platform-specific considerations, see our Risk Management for Prediction Markets: A Complete Guide.
You should also consider how Kelly-sized positions affect your portfolio's risk-adjusted return profile. Maximizing geometric growth (what Kelly does) is not the same as maximizing the Sharpe ratio. Our guide on Risk-Adjusted Returns in Prediction Markets: Sharpe Ratio Optimization explains how to balance both objectives.
For traders who have moved beyond basic contracts and want to apply these techniques to more complex Kalshi markets, see Advanced Kalshi Strategies: Beyond Your First Trade.
Quick Reference: Kelly Criterion Checklist for Prediction Markets
- Identify the contract's current market price (implied probability).
- Form your own probability estimate using independent research or a model.
- Calculate net odds: b = (1 − price) / price.
- Apply the formula: f = (bp − q) / b.
- If f ≤ 0, skip the trade—there is no edge.
- Apply a fractional Kelly multiplier (0.25–0.50) to the raw f value.
- Check for correlation with existing positions and reduce further if needed.
- Recalculate after accounting for transaction costs and realistic fill prices.
Frequently Asked Questions
What is the Kelly Criterion in simple terms?
The Kelly Criterion is a formula that tells you what percentage of your bankroll to bet on a given opportunity based on your estimated edge and the odds offered. It is designed to maximize the long-run compounding rate of your capital. Betting more than Kelly recommends actually reduces your long-term growth rate, even if you have a genuine edge.
Can you use the Kelly Criterion on Kalshi or Polymarket?
Yes. Prediction market contracts on platforms like Kalshi and Polymarket are ideal for Kelly because they have binary outcomes and fixed $1.00 payouts, which makes both "p" (your probability estimate) and "b" (net odds from the contract price) straightforward to calculate. The main challenge is forming an accurate probability estimate that is better than the market's implied probability.
What fraction of Kelly should prediction market traders use?
Most experienced prediction market traders use between 25% and 50% of the full Kelly recommendation. Half Kelly (50%) is a common starting point because it captures roughly 75% of the theoretical compounding benefit while cutting variance significantly. Traders with less confidence in their probability models, or with correlated positions, should move toward quarter Kelly or lower.
What happens if you consistently bet more than Kelly recommends?
Over-betting relative to Kelly is mathematically guaranteed to produce a lower long-run growth rate than optimal Kelly, even with a real edge. In the extreme, betting twice the Kelly amount produces the same expected growth rate as betting nothing at all—but with much higher variance. Sustained over-betting eventually leads to ruin regardless of edge size.
How do you handle the Kelly Criterion when you have multiple simultaneous prediction market positions?
Standard Kelly assumes independent bets evaluated one at a time. When holding multiple positions, you need to account for correlations between outcomes and apply a portfolio-level Kelly or fractional Kelly framework that ensures your total bankroll allocation across all positions remains within safe limits. Our guide on Advanced Kelly Criterion: Fractional Kelly & Multi-Market Applications covers this in detail.