Kelly criterion for event contract sizing
The Kelly criterion is a formula for sizing bets that maximizes the long-run growth rate of your bankroll. Applied to binary event contracts, it tells you exactly what fraction of your capital to deploy on a given trade given your estimate of the true probability and the price offered by the market.
Kelly criterion for event contract sizing
The formula
For a binary outcome (win or lose), the Kelly fraction is: f* = (b * p - q) / b, where b is the net odds on a winning bet (if you stake $1 and win, you receive $b plus your stake back), p is your estimate of the true probability of winning, and q = 1 - p is the probability of losing. In prediction market terms, if a YES share is priced at $0.40 and you think the true probability is 55%, then b = (1 - 0.40) / 0.40 = 1.5, and f* = (1.5 * 0.55 - 0.45) / 1.5 = 0.25. You should stake 25% of your bankroll on this trade.
Estimating true probability vs market price
The Kelly formula is only as good as your probability estimate. If the market price already reflects all available information, your edge is zero and Kelly tells you to bet nothing. Your estimate must be genuinely independent of the market price to generate a positive Kelly fraction. Good sources include base rates from historical data, proprietary models, or information the market has not yet absorbed. Overconfidence in your own estimate is the most common Kelly mistake: traders systematically overstate their edge, leading to position sizes that are far too large.
Fractional Kelly in practice
Full Kelly maximizes long-run growth but produces dramatic short-run volatility. A single bad bet at full Kelly can cost you a quarter of your bankroll. Most professional gamblers and traders use fractional Kelly, typically 25% to 50% of the full Kelly fraction. Half-Kelly roughly halves both the variance and the long-run growth rate relative to full Kelly. For most prediction market traders, 25% Kelly is a reasonable starting point: it keeps individual losses manageable while still compounding capital meaningfully over many trades.
How fees change the math
Platform fees reduce your effective edge and therefore reduce the Kelly fraction. On Polymarket, a 2% fee on winnings means your net payout on a winning YES share is $0.98 rather than $1.00. On Kalshi, fees can reach 5% on some markets. To account for fees, reduce the net odds b by the fee percentage before calculating f*. If you calculated a Kelly fraction of 20% before fees and the fee is 5% of winnings, the post-fee Kelly fraction will be noticeably smaller. Always calculate Kelly on your after-fee expected value, not the gross payout.
Artículos relacionados en esta serie
¿Qué es la probabilidad implícita en los mercados de predicción?
Por qué una acción SÍ que cotiza a $0.62 significa una probabilidad implícita de mercado del ~62 %, y cuándo esa lectura falla.
SerieLibros de órdenes frente a AMM en mercados de predicción
Kalshi y Polymarket usan libros de órdenes; las plataformas on-chain más antiguas usan creadores de mercado automatizados. Cómo cada uno fija el precio del riesgo de forma diferente.
Serie¿Qué es LMSR (Logarithmic Market Scoring Rule)?
La fórmula de creación de mercado que Robin Hanson diseñó para los mercados de predicción, por qué acota las pérdidas y quién sigue usándola.