Staking: how much, and why not more
Kelly, fractional Kelly, and the arithmetic that makes a bad run survivable.
Updated September 27, 2026 · TrueEdge Academy
The Kelly criterion gives the stake that maximises long-run growth: bet a fraction of your bankroll equal to your edge divided by the odds. It is provably optimal and almost nobody should use it at full size, because it assumes you know your edge exactly and you do not.
- Edge
- 5%
- Full Kelly stake
- 5% of bankroll
- Quarter Kelly (typical)
- 1.25% of bankroll
- Growth rate kept
- about 44% of full Kelly's
- Volatility
- about a quarter of it
Betting a fraction c of Kelly keeps about c × (2 − c) of its growth: 44% at a quarter, 75% at a half. Quarter Kelly is the common choice anyway, because it cuts the swings to a quarter and survives the usual mistake — an edge estimated higher than it really is. See the Kelly criterion guide.
The asymmetry that matters
Losing 50% of a bankroll requires a 100% gain to recover. Losing 80% requires 400%. Drawdowns compound against you harder than gains compound for you, which is the entire argument for staking below what the maths says is optimal.
And your edge estimate is wrong
Kelly is optimal given a known edge. Yours is estimated from a fair price that is itself an estimate. Overstating your edge by half and betting full Kelly puts you well past the point where growth turns negative — fractional staking is the margin for that error.