Find the bet size that maximizes the long-run growth of your bankroll. Enter your bankroll, your edge over the market, and the decimal odds — the calculator returns full Kelly plus the safer fractional sizes most real-world traders use.
This calculator is useful only after you have a thesis. If you still need to compare price, probability, and expected value, start with the implied probability calculator and the EV calculator first.
For a binary bet, the Kelly fraction of bankroll to wager is:
f* = ( odds × p − 1 ) / ( odds − 1 )
Where p is your probability of winning (market-implied probability + your edge) and odds are decimal. If f* comes out negative or zero, you don't have an edge and shouldn't bet.
Kelly maximizes the geometric growth rate of your bankroll. Bet less, you grow slower. Bet more, you grow faster on average but eventually go broke from variance.
Full Kelly assumes you know your probability exactly. In real prediction markets you don't — you estimate it. Overestimating your edge by even a small amount can turn full-Kelly sizing into ruin sizing. Halving or quartering Kelly cuts variance roughly proportionally while keeping most of the growth, which is why the pros do it.
For a deeper treatment of bankroll management, see our portfolio management guide and our risk management guide.
To find your edge first, use the probability calculator. To verify a single trade is +EV before sizing it, use the EV calculator.
A Kelly criterion calculator estimates how much of your bankroll to risk when you know your probability estimate and the market odds. It is a sizing tool, not a signal generator: bad probability estimates still create bad bets.
Convert the YES price to decimal odds, compare the market probability with your researched probability, enter the edge, and use the fractional Kelly output as a conservative position-size ceiling.
Usually no. Full Kelly assumes your probability estimate is correct, which is rarely true in live prediction markets. Half Kelly or quarter Kelly reduces volatility and gives room for fees, spread, slippage, and model error.