Picking the right side is only half the job. How much you stake on it determines how fast your bankroll grows — and how likely you are to blow it up. Bet too little and a real edge barely moves the needle; bet too much and variance ruins you. Position sizing is the bridge between having an edge and actually profiting from it.
Why sizing matters as much as picking
Two traders can make exactly the same correct calls and end up worlds apart purely because of how they sized. The principle behind good sizing is that your stake should scale with two things: the size of your edge (how far your estimate is from the price) and the odds on offer. A fixed-dollar bet ignores both. The Kelly criterion formalises this.
The Kelly criterion, simply
The Kelly criterion gives the stake that maximises the long-run growth of your bankroll. For a binary prediction-market contract, it takes a clean form. If you can buy a Yes contract at price p (in dollars, so 55¢ is 0.55) and you believe the true probability is q, the Kelly fraction of your bankroll to stake is:
where p is the contract price and q is your estimated probability. If q is not greater than p, you have no edge and Kelly says stake nothing.
A worked example
Suppose a contract trades at 50¢ (p = 0.50) and, after your research, you estimate the true probability at 60% (q = 0.60). Plugging in: (0.60 − 0.50) ÷ (1 − 0.50) = 0.10 ÷ 0.50 = 0.20. Full Kelly says stake 20% of your bankroll. On a $1,000 bankroll that is $200 — which, for most people, immediately illustrates why almost nobody bets full Kelly.
Translate the price into a probability first with our odds converter, then check the payoff with the profit calculator — those are the raw inputs for any sizing decision.
Why use fractional Kelly
Full Kelly is mathematically optimal only if your probability estimate is exactly right — and it never is. It is also brutally volatile, with large swings that most people cannot stomach. The standard fix is fractional Kelly: stake a half or a quarter of what the formula suggests. Half-Kelly captures most of the long-run growth with roughly half the volatility, and it cushions the damage when your estimate of q is off. In the example above, half-Kelly would stake 10% and quarter-Kelly 5% — much closer to sensible bankroll limits.
Estimating your edge honestly
Every Kelly calculation depends on q, your probability estimate — and that is where most of the risk lives. If you systematically overestimate your edge, even fractional Kelly will overbet. Be conservative with q, and only size up when you can clearly explain why the market price is wrong. The discipline of finding a genuine edge is covered in how to find value.
Kelly is a guide, not a guarantee. Combine it with a hard per-position cap from your bankroll plan, and never let a formula talk you into a position larger than you can afford to lose.
Related strategies
Build on this approach with the adjacent playbooks:
Frequently asked questions
What is the Kelly criterion in simple terms?
It is a formula for how much to stake based on your edge: the bigger the gap between your estimated probability and the price, the more you stake — but always as a fraction of your bankroll, never everything. For a contract at price p with true probability q, the Kelly fraction is (q - p) / (1 - p).
Should I bet full Kelly?
Almost no one does. Full Kelly is optimal only if your probability estimate is exactly right and is very volatile. Most traders use fractional Kelly — a half or quarter of the suggested amount — which captures most of the growth with far less risk.
How do I size a position on a prediction market?
Estimate the true probability (q), compare it with the contract price (p), and if q is higher you have an edge. Apply the Kelly fraction (q - p)/(1 - p), then take a half or quarter of it and cap it within your bankroll limits.