Kelly Criterion Calculator for Prediction Markets
Size a Polymarket, Kalshi or HIP-4 position with the Kelly criterion. Enter your probability, the contract price and fees to get a stake in dollars and contracts.
How much of your bankroll should this contract get?
Enter the price you would pay for a contract that pays $1 if it wins, and your own probability that it wins. For a NO position, enter the NO price and your probability of NO.
Use the average price you expect to fill at, not the midpoint. Fees default to zero; estimate yours with the profit calculator and divide by the number of contracts.
Suggested stake
$83.20
208 contracts, 8.32% of your bankroll.
- Cost per contract incl. fees
- 40.00¢
- Break-even probability
- 40.00%
- Edge per contract
- 10.00¢
- Expected return per $1 staked
- 25.00%
- Full Kelly fraction
- 16.67%
- Selected fraction of bankroll
- 8.33%
- Profit if it wins
- $124.80
- Loss if it loses
- $83.20
- Expected profit
- $20.80
- Expected log growth per bet
- 1.537%
Kelly assumes your probability is right, the contract settles at $1 or $0, and this is your only open bet. Overestimating your edge makes full Kelly oversized; that is why many traders use half or quarter Kelly. This is a sizing formula, not a forecast or financial advice.
The Kelly formula for prediction markets
A binary contract bought at cost c (price plus fees) pays $1 if it wins. A winning contract returns 1 − c on c staked, so its net odds are b = (1 − c) ÷ c. With your probability p, the Kelly fraction (p × b − (1 − p)) ÷ b simplifies to (p − c) ÷ (1 − c). That is your edge divided by the most a contract can earn.
Worked example: a contract trades at 40¢ and you think it wins 50% of the time. Full Kelly is (0.50 − 0.40) ÷ 0.60 = 16.67% of your bankroll. Half Kelly is 8.33%: on $1,000 that is 208 contracts for $83.20. If it wins you make $124.80; if it loses you lose $83.20.
Why most traders bet a fraction of Kelly
Full Kelly maximizes long-run growth only if your probability is accurate. Overestimating it makes full Kelly too large, and drawdowns at full Kelly are deep even when you are right. Half Kelly keeps about three quarters of the growth with much less variance. If you are new to forecasting, measure your accuracy first with the calibration journal.
The formula sizes one bet at a time. Several open positions on related events, such as strikes on the same price or teams in one league, are not independent; size them together and reduce each stake. Kelly also ignores liquidity: a thin order book can fill you at a worse average price than the one you entered.