# Expected value in prediction markets: is a trade worth it, and how much to stake? > How to tell if a prediction market trade is worth it after fees, and how much to stake with the Kelly criterion. Simple formulas and worked examples. Oct 8, 2026 · Markets · Sikt Intelligence · https://www.siktintelligence.com/blog/expected-value-prediction-markets **A prediction market trade is worth making only if your probability is higher than the price you pay, after fees.** That gap is your expected value. How much to stake on it is a separate question, and the classic answer is the Kelly criterion. Both rest on one number you have to supply yourself: an honest probability. This guide shows the formulas, works through examples, and explains why that one number decides everything. ## Expected value in one line A "Yes" contract pays $1 if the event happens and nothing if it does not. If you buy it at price **c** and believe the true chance is **p**, then per contract: > **Expected value = p − c − fee** - **Example:** you think an event is 60% likely, and Yes trades at 50¢. Before fees, your expected value is 0.60 − 0.50 = **10¢ per contract**, a 20% expected return on what you pay. - **Break-even:** a trade is worth it only if your probability is above the price plus fees. At 50¢ with a 1.75¢ fee, you need to believe the chance is above about 51.75%. The same logic works for "No": buy it if you think the event is *less* likely than the market says. ## Fees change the answer Fees are small per contract but large relative to small edges. On Kalshi, the trading fee is 0.07 × contracts × price × (1 − price), rounded up to the next cent, so it is highest near 50¢, at about 1.75¢ per contract ([Kalshi fee schedule](https://kalshi.com/docs/kalshi-fee-schedule.pdf); [Covers](https://www.covers.com/betting/prediction-sites/polymarket-vs-kalshi)). Redo the example with fees: 0.60 − 0.50 − 0.0175 = **8.25¢**, not 10¢. And on a 52%-versus-50¢ "edge", the fee eats most of it. Add the spread, the gap between the buying and selling price, and many apparent edges vanish. We explain both in [prediction market odds, explained](/blog/prediction-market-odds-explained). ## How much to stake: the Kelly criterion In 1956, the Bell Labs scientist John L. Kelly worked out the stake that makes your money grow fastest over many bets ([Kelly, 1956](https://www.adrian.idv.hk/2020-11-18-k56-inforate/)). For a prediction market contract that pays $1, it simplifies to: > **Kelly stake = (p − c) ÷ (1 − c)** of your bankroll - **Example:** p = 60%, c = 50¢. Kelly says stake (0.60 − 0.50) ÷ 0.50 = **20%** of your bankroll. Include the fee as part of the price (c = 51.75¢), and it drops to about **17%**. - **No edge, no bet:** if p is at or below the price, Kelly says stake nothing. ### Why most people use half Kelly or less Full Kelly is aggressive: it produces big swings, and it assumes your probability is right. If you overestimate your chances, which almost everyone does, full Kelly overbets. The mathematician Ed Thorp, who used Kelly to beat blackjack and later to run a hedge fund, recommended half Kelly or less: halving the stake keeps about three-quarters of the long-run growth with far smaller swings ([Thorp](https://gwern.net/doc/statistics/decision/2006-thorp.pdf)). In the example, half Kelly means staking about 8–10% of your bankroll, and many careful traders go lower still. ## The number everything depends on Every formula above takes your probability **p** as given. That is the hard part: - **Overconfidence is expensive.** If your "60%" events happen only 50% of the time, every trade above has negative expected value, and Kelly makes it worse by telling you to bet big. - **Long shots fool people.** Cheap contracts feel like value, but contracts under 10¢ have historically been overpriced: on Kalshi, their buyers lost more than 60% of their money on average ([Bürgi, Deng and Whelan](https://www.karlwhelan.com/Papers/Kalshi.pdf); see the [favorite-longshot bias](/blog/favorite-longshot-bias)). - **Check your calibration.** Record your probabilities and what happened, and check whether your 60%s come true 60% of the time. Our free [Brier score calculator](/tools/brier-score-calculator) does it, and with the market's price as a third column it shows whether you actually beat the market. That is why most traders lose money ([why most prediction market traders lose money](/blog/why-prediction-market-traders-lose-money)): the formulas are easy; an honest probability is not. ## Where Sikt Intelligence fits Sikt Intelligence is building that number: an [AI superforecaster](/blog/what-is-an-ai-superforecaster) that reads the news and data that exist today, checks every source, and gives an honest probability next to the market's price, graded every time, misses included. When the two disagree, that gap is where expected value starts. Leave your email below for early access. Nothing here is financial or investment advice, and no formula removes the risk of losing money. ## Key takeaways - Expected value per contract = your probability − price − fee. Trade only when it is clearly positive. - Fees and spreads erase small edges; near 50¢ on Kalshi, the fee is about 1.75¢ per contract. - Kelly stake for a $1 contract = (p − c) ÷ (1 − c) of your bankroll; most careful traders use half Kelly or less. - Everything depends on an honest, calibrated probability. Overconfidence turns positive expected value negative. ## FAQ ### How do you calculate expected value in a prediction market? Subtract the price and the fee from your probability. If you think an event is 60% likely and Yes costs 50¢ with a 1.75¢ fee, your expected value is about 8 cents per contract. ### What is the Kelly criterion? A formula from 1956 for the stake that makes your money grow fastest over many bets. For a prediction market contract that pays $1, stake (p − c) ÷ (1 − c) of your bankroll, where p is your probability and c the price. ### Should I bet full Kelly? Usually not. Full Kelly assumes your probability is exactly right and produces large swings. Half Kelly keeps about three-quarters of the growth with much less risk, and many traders bet less still. ### Is a cheap contract good value? Often not. Contracts under 10¢ tend to be overpriced, because people overpay for small chances of big payouts. Check the expected value with your own honest probability first. ## Sources - Kelly, J. L. (1956): A New Interpretation of Information Rate, *Bell System Technical Journal* 35(4) ([summary](https://www.adrian.idv.hk/2020-11-18-k56-inforate/)) - Thorp, E. O. (2006): [The Kelly Criterion in Blackjack, Sports Betting and the Stock Market](https://gwern.net/doc/statistics/decision/2006-thorp.pdf) - Kalshi: [Fee schedule](https://kalshi.com/docs/kalshi-fee-schedule.pdf); Covers: [Polymarket vs. Kalshi: fees](https://www.covers.com/betting/prediction-sites/polymarket-vs-kalshi) - Bürgi, Deng and Whelan (2026): [Makers and Takers: The Economics of the Kalshi Prediction Market](https://www.karlwhelan.com/Papers/Kalshi.pdf)