The Kelly Criterion is a bankroll management formula that calculates the optimal fraction of capital to risk on a single bet — maximizing long-run portfolio growth while minimizing ruin risk. Developed by Bell Labs scientist John L. Kelly Jr. in 1956, it remains the foundational position-sizing tool for professional prediction market traders on platforms like Kalshi and Polymarket.
What Is the Kelly Criterion and Why Does It Matter for Prediction Markets?
The Kelly Criterion was originally published by John L. Kelly Jr. in his 1956 Bell System Technical Journal paper. Designed initially for signal transmission theory, it was rapidly adopted by gamblers, and later by professional investors, because it answers one of the most important questions in any probabilistic endeavor: how much of your capital should you risk on any single bet?
In prediction markets like Kalshi and Polymarket, where you're trading binary contracts (will X happen: yes or no?), the Kelly Criterion is uniquely powerful. Every contract has a price — say, 62¢ on Kalshi — that implies a market probability of 62%. If your research tells you the true probability is 75%, you have an edge. Kelly tells you exactly how large a position that edge justifies.
Without a structured framework like Kelly, traders either over-bet (blowing up accounts on high-confidence losses) or under-bet (leaving massive returns on the table). Both errors compound painfully over time. As quantitative finance researcher Edward Thorp demonstrated in his landmark work on optimal betting, Kelly-based sizing consistently outperforms fixed-fraction strategies across long time horizons.
What Is the Kelly Criterion Formula and How Does It Work?
f* = (bp − q) / b
- f* — the fraction of your bankroll to wager
- b — net odds received (profit per $1 risked). On a binary contract priced at 60¢, if you're right you receive 40¢ profit, so b = 0.40/0.60 ≈ 0.667
- p — your estimated probability the event occurs
- q — probability it does not occur (1 − p)
How Do You Calculate Kelly Criterion for a Prediction Market Contract?
Suppose a Kalshi contract asks: "Will the Fed hold rates steady at the May 2026 meeting?" The contract is trading at 65¢ (implying a 65% market probability). After reviewing Fed minutes, employment data, and CME FedWatch data, you estimate the true probability at 78%.
- Contract price: $0.65 → net odds b = (1 − 0.65) / 0.65 = 0.538
- Your estimated probability: p = 0.78, q = 0.22
- Kelly fraction: f* = (0.538 × 0.78 − 0.22) / 0.538 = (0.420 − 0.22) / 0.538 = 0.372, or ~37% of bankroll
Full Kelly recommends betting 37% of your account on this single contract. That's theoretically optimal — but in practice, almost no serious trader uses full Kelly. Here's why.
Why Should You Almost Never Use Full Kelly in Prediction Markets?
Full Kelly maximizes the long-run geometric growth rate of your bankroll, but it assumes your probability estimate is perfectly accurate. In prediction markets, your edge estimate is always uncertain. A small error — say, you estimated 78% but the true probability was 68% — can turn a Kelly-optimal bet into a dangerous over-bet that dramatically increases ruin risk.
The standard professional solution is Fractional Kelly: bet a fixed fraction of the Kelly-recommended size. Most quantitative traders use Half Kelly (50% of the Kelly output) as a default. Half Kelly preserves roughly 75% of the geometric growth rate of Full Kelly, while cutting variance — and drawdown risk — by approximately half. For a deeper breakdown of fractional sizing strategies and advanced Kelly variants, see our guide to Kelly Criterion mastery and complete position sizing for prediction markets.
What Is Fractional Kelly and Which Fraction Should You Use?
Fractional Kelly means multiplying your calculated f* by a fixed scaling factor before placing a bet. Common professional benchmarks are:
- Full Kelly (1.0×): Theoretically optimal only if your probability estimates are perfectly calibrated — rarely the case.
- Half Kelly (0.5×): The most widely used professional default. Cuts variance significantly while retaining most of the growth advantage.
- Quarter Kelly (0.25×): Used when edge confidence is low or when managing a large number of simultaneous positions.
In the Fed rate example above, Half Kelly would recommend betting approximately 18.6% of your bankroll rather than 37%. For most prediction market traders holding multiple open positions simultaneously, Quarter Kelly or Half Kelly are the rational defaults.
How Does Kelly Criterion Apply Differently Across Prediction Market Platforms?
The core formula is platform-agnostic, but two practical differences matter depending on where you trade:
Kalshi: As a CFTC-regulated exchange, Kalshi contracts resolve at $1.00 or $0.00, making the binary Kelly formula a direct fit. Contract prices reflect consensus probabilities from real-money traders. As of Q1 2026, Kalshi lists over 200 active markets across economics, politics, and weather categories.
Polymarket: Polymarket operates using USDC on the Polygon blockchain, with contract mechanics identical to Kalshi for Kelly purposes. However, liquidity varies more significantly across markets, which can affect your ability to enter or exit at your target price — a practical constraint the Kelly formula itself doesn't account for.
In both cases, the biggest variable is not the formula — it's the accuracy of your probability estimate p. Even a perfectly executed Kelly calculation produces poor results if your edge estimate is miscalibrated.
What Are the Most Common Kelly Criterion Mistakes in Prediction Markets?
- Overconfidence in p: Estimating 80% when the true probability is closer to 65% inflates your Kelly fraction dangerously. Conservative edge estimates are nearly always better than optimistic ones.
- Ignoring correlation across positions: If you hold five contracts that all resolve on the same Fed decision, your effective exposure is much higher than individual Kelly fractions suggest. Treat correlated positions as a single bet.
- Applying Kelly to illiquid markets: Kelly assumes you can trade at the stated price. Thin order books on smaller markets mean your actual fill price may differ enough to invalidate the calculation.
- Recalculating too infrequently: As new information arrives and market prices move, your edge changes. Kelly fractions should be recalculated whenever the contract price or your probability estimate shifts materially.
How Does Kelly Criterion Compare to Fixed-Fraction Betting in Prediction Markets?
Fixed-fraction betting — wagering a constant 2%, 5%, or 10% of your bankroll on every trade regardless of edge — is simpler but mathematically suboptimal. It ignores the key insight that position size should scale with the size of your edge. A contract where you have a 3% edge does not deserve the same allocation as one where you have a 15% edge.
Kelly-based sizing dynamically allocates more capital to higher-edge opportunities and less (or nothing) to marginal ones. Over a large sample of trades, this dynamic allocation produces meaningfully higher geometric growth than any fixed-fraction approach — which is why institutional traders, quantitative funds, and professional sports bettors have adopted Kelly-based frameworks since the 1970s.
Frequently Asked Questions: Kelly Criterion in Prediction Markets
What does a negative Kelly fraction mean in a prediction market?
A negative Kelly fraction means you have no edge — or a negative edge — on that contract at its current price. The formula is signaling you should not bet, or that the opposing side of the contract offers value. Never bet when Kelly outputs a negative number.
Can you use the Kelly Criterion on Polymarket and Kalshi simultaneously?
Yes, but treat your total capital across both platforms as a single bankroll. Kelly fractions represent a share of total account equity, not per-platform balances. Running separate Kelly calculations per platform while ignoring the combined exposure is a common and costly mistake.
How accurate does my probability estimate need to be for Kelly to work?
Kelly degrades quickly with miscalibration. If your estimated edge is 10% but your true edge is 5%, Full Kelly will roughly double your optimal position size and significantly increase drawdown risk. This is the core argument for using Half Kelly or Quarter Kelly as a buffer against estimation error.
Is the Kelly Criterion legal to use on CFTC-regulated prediction markets?
Yes. Kelly is a position-sizing methodology, not a trading strategy — it carries no regulatory implications. Kalshi is a fully CFTC-regulated designated contract market, and using mathematical frameworks to size positions is standard practice on all regulated exchanges.
What is Half Kelly and why do professionals prefer it?
Half Kelly means multiplying your Kelly-calculated fraction by 0.5 before betting. It retains approximately 75% of Full Kelly's geometric growth rate while cutting variance and maximum drawdown risk roughly in half. Most professional prediction market traders and quantitative bettors use Half Kelly as their default position-sizing rule.
Does the Kelly Criterion account for trading fees on Kalshi or Polymarket?
The standard Kelly formula does not automatically account for fees. To incorporate them, reduce your net odds variable b by the effective fee rate. On Kalshi, trading fees as of Q1 2026 range from 1–7¢ per contract depending on volume tier — small but meaningful enough to adjust your Kelly input on thin-edge trades.