The Kelly criterion gives the stake size that maximises the long-run growth rate of a bankroll, given a known edge. The phrase doing the work in that sentence is known edge.
The formula
With b as the profit per unit staked at the offered price, p as your estimated probability of winning, and q as 1 minus p, the Kelly fraction is (b times p, minus q) divided by b. If that number is zero or negative, the correct stake is zero.
Why full Kelly is rarely correct
- It assumes your probability estimate is exact. It never is.
- Overestimating your edge causes overstaking that compounds, and the damage is asymmetric: staking twice the optimal amount is far worse than staking half.
- The optimal-growth path includes drawdowns most people will not sit through, which means they deviate from it at the worst moment.
- Correlated positions break the single-bet assumption entirely.
Fractional Kelly — a half, a quarter, or less — sacrifices a modest amount of theoretical growth for a large reduction in variance and a large margin of safety against your own estimation error. Most careful practitioners stake a quarter Kelly or below.
The failure mode that matters most
If your true edge is zero but you believe it is 4 percent, Kelly does not protect you. It tells you to stake confidently into a negative-expectation bet, and it tells you to do so repeatedly. Staking math cannot rescue a wrong probability estimate; it can only limit how fast a wrong estimate destroys a bankroll.
Note
The VigPulse Kelly calculator clamps negative results to zero and defaults to a fractional multiplier for these reasons. It is an educational tool and it is not a recommendation to stake anything.