How to Use the Kelly Criterion for Prediction Markets

The Kelly Criterion is a mathematical formula developed by John L. Kelly Jr. at Bell Labs in 1956 that determines the optimal fraction of your bankroll to wager on a bet with a positive expected value. It maximizes the long-term growth rate of your capital while minimizing the risk of ruin. In prediction markets, where every position resolves to $0 or $1, the Kelly Criterion provides a principled answer to one of the hardest questions in trading: how much should I bet?

The beauty of the Kelly approach is that it balances aggression with caution. Bet too little and you leave money on the table. Bet too much and a losing streak can wipe you out. Kelly finds the exact point in between that maximizes your geometric growth rate — the rate at which your bankroll compounds over many bets. This is why it has been adopted by professional gamblers, hedge fund managers, and now prediction market traders as the gold standard for position sizing.

The Kelly Criterion assumes you know your true probability better than the market. If you do not have an edge — if the market price already reflects the true probability — Kelly tells you to bet nothing. This is a feature, not a bug. It forces you to be honest about whether you actually have superior information before committing capital. Many traders discover, after running the Kelly calculation, that their perceived edge does not justify the position size they were planning.

For a binary prediction market, the Kelly formula simplifies to: f = (p - c) / (1 - c), where f is the fraction of your bankroll to bet, p is your estimated true probability of the outcome occurring, and c is the current cost of the share (the market price). This formula tells you to bet a fraction equal to your edge divided by the potential payout. If p equals c — meaning you agree with the market — the formula returns zero. You should not bet.

Consider a concrete example. You believe an event has a 70% chance of occurring, and Yes shares cost $0.55. Your edge is 0.70 minus 0.55 equals 0.15. The Kelly fraction is (0.70 - 0.55) / (1 - 0.55) = 0.15 / 0.45 = 0.333. Kelly says bet 33.3% of your bankroll. On a $10,000 account, that is $3,333 worth of shares. If you are right about the 70% probability, this bet size maximizes your long-run growth rate.

For No shares, the formula adjusts accordingly. If you believe the event has only a 30% chance of occurring but the market prices Yes at $0.55, you would buy No shares at $0.45. Your edge on No is (0.70 - 0.45) / (1 - 0.45) = 0.25 / 0.55 = 0.455. In practice, buying No when you think the probability is lower is often mathematically equivalent — the key is that the formula always quantifies whether and how much to bet based on your informational advantage.

Binary prediction markets are the ideal setting for Kelly because the payoff structure is simple: you either receive $1 per share or $0. There are no partial payouts, no early exits in the formula (though you can always sell before resolution), and no continuous price paths to model. This simplicity means you can calculate the exact Kelly bet for any market in seconds, as long as you have an honest estimate of the true probability.

To apply Kelly in practice, start by writing down your probability estimate before looking at the market price. This prevents anchoring — the cognitive bias where the market price influences your estimate. Then compare your estimate to the market price. If there is a meaningful gap (your estimate is at least 5 to 10 percentage points away from the market), calculate the Kelly fraction. If the gap is small, the Kelly bet will be tiny, which is the formula's way of telling you the edge may not be worth the risk.

One important nuance: the Kelly formula assumes you are correct about your probability estimate. In reality, you are never 100% certain about your estimate. If you think there is a 70% chance of an event, your actual confidence in that 70% number is itself uncertain. This uncertainty in your estimate is the primary reason most practitioners use fractional Kelly rather than full Kelly. Overestimating your edge is the most common and most costly mistake in Kelly-based position sizing.

Full Kelly betting maximizes long-run growth rate under the assumption that your probability estimates are perfectly calibrated. In practice, nobody's estimates are perfect, and full Kelly can produce stomach-churning volatility. A trader using full Kelly might bet 30% or more of their bankroll on a single market. If they are wrong about the probability, the resulting loss can take months to recover from. The mathematical growth rate is optimal, but the psychological toll and practical risk of a large drawdown make full Kelly impractical for most traders.

Fractional Kelly — typically half-Kelly or quarter-Kelly — reduces position sizes proportionally. Half-Kelly cuts the recommended bet in half, achieving approximately 75% of the full Kelly growth rate with substantially less volatility and drawdown risk. Quarter-Kelly further reduces size, achieving roughly 50% of the growth rate but with very smooth equity curves and minimal risk of ruin. The choice between fractions depends on your risk tolerance, the accuracy of your probability estimates, and how many simultaneous positions you hold.

Most professional prediction market traders use half-Kelly as their default. The reasoning is practical: if your probability estimate is off by even a few percentage points, half-Kelly keeps you in a safe zone where the damage is manageable. With full Kelly, a small estimation error can turn a positive-expected-value bet into a portfolio-damaging overbet. Half-Kelly provides a comfortable margin of safety while still capturing most of the long-run compounding benefit. If you are new to Kelly-based sizing, start with quarter-Kelly and work up as you validate your calibration.

The Kelly Criterion is only as good as your edge estimate, which makes accurate probability estimation the most critical skill in prediction market trading. Your edge is the difference between your estimated true probability and the market price. If you estimate 65% and the market says 55%, your perceived edge is 10 percentage points. But how confident are you in that 65%? If you are uncertain, your effective edge is smaller, and your Kelly bet should be reduced accordingly.

One practical method for calibrating your estimates is to keep a prediction journal. Before each trade, write down your probability estimate, the market price, your reasoning, and the Kelly bet you calculated. After the market resolves, review your estimates. Over 50 to 100 markets, you can calculate your calibration: if you said 70% and the event occurred 70% of the time, you are well-calibrated. If events you rated at 70% only occurred 55% of the time, you are systematically overconfident and should use a more conservative Kelly fraction.

You can also benchmark your estimates against the track records of top traders on <a href="https://0xinsider.com/leaderboard">0xInsider's leaderboard</a>. If the highest-rated traders are positioned in the opposite direction of your trade, it should prompt you to re-examine your probability estimate. This does not mean you should always follow the crowd, but it does mean you should have a specific, articulable reason for disagreeing with traders who have demonstrated consistent skill. If you cannot articulate why your estimate differs, your edge may be illusory.

The most dangerous mistake is overestimating your edge. If you think you have a 15-point edge but your actual edge is 5 points, full Kelly on the perceived edge is a massive overbet relative to the real edge. This is not a theoretical concern — most traders are systematically overconfident in their probability estimates. The solution is to use fractional Kelly and to rigorously track your calibration over time. If you have not resolved at least 30 markets, you do not have enough data to know how good your estimates are.

Another common error is applying Kelly independently to each position without considering portfolio-level risk. If you have 10 open positions, each sized at half-Kelly, your total portfolio allocation can be extremely high. Kelly assumes each bet is independent, but in prediction markets, many events are correlated — political markets move together, crypto markets move together, and macro events affect multiple markets simultaneously. A portfolio-level Kelly approach accounts for these correlations by reducing individual position sizes when the total portfolio risk exceeds a target threshold.

A third mistake is ignoring the cost of being wrong about the direction of the bet. Kelly tells you the optimal size assuming you have a positive edge. If your edge is negative — if the market is actually more accurate than your estimate — Kelly gives a negative number, which means do not bet. Some traders see a slightly negative Kelly result and bet anyway because they feel the opportunity is too good to pass up. This is a recipe for slow, steady losses. If Kelly says do not bet, the disciplined response is to move on and find a market where you have a genuine edge.

Example 1: A market on whether a central bank will raise rates is priced at $0.40 for Yes. You have deep expertise in monetary policy and estimate a 55% probability. Kelly fraction: (0.55 - 0.40) / (1 - 0.40) = 0.15 / 0.60 = 0.25. Full Kelly says bet 25% of your bankroll. At half-Kelly, that is 12.5%. On a $20,000 account, you would buy $2,500 worth of Yes shares — approximately 6,250 shares at $0.40 each. If the rate hike happens, you collect $6,250, netting $3,750 in profit.

Example 2: An election market prices a candidate's Yes at $0.72. You think the true probability is 80%. Kelly fraction: (0.80 - 0.72) / (1 - 0.72) = 0.08 / 0.28 = 0.286. Half-Kelly: 14.3%. On a $15,000 account, that is $2,143 in Yes shares — about 2,976 shares at $0.72. The edge here is smaller (8 points vs 15 in Example 1), so even though the probability is high, the Kelly bet is smaller relative to the bankroll. This is the formula working correctly — smaller edge means smaller bet, regardless of how confident you feel.

Example 3: A crypto market prices Yes at $0.50. You estimate 52%. Kelly fraction: (0.52 - 0.50) / (1 - 0.50) = 0.02 / 0.50 = 0.04. Full Kelly is just 4% of your bankroll. At half-Kelly, that is 2% — a very small position. This is Kelly telling you that a 2-point edge is not worth a big bet. The expected value is slightly positive, but the variance is enormous relative to the edge. Many experienced traders would skip this trade entirely, reserving their capital for opportunities where the edge is wider and the Kelly bet is more meaningful.

These examples illustrate a key insight: Kelly naturally concentrates your capital in your highest-conviction, widest-edge opportunities and keeps you small or out of markets where your edge is thin. Over time, this capital allocation discipline is one of the biggest drivers of long-term compounding in prediction markets.