The Kelly Criterion maximizes the expected long-run growth rate of wealth, but full-Kelly betting implies large position sizes and correspondingly large drawdowns in practice. Fractional Kelly betting — most commonly half-Kelly (f = 0.5 × f*) — scales the position size down proportionally.
The trade-off is well-characterized: a strategy at fraction f of Kelly earns approximately f × (full-Kelly growth rate) in expected log-wealth terms, but its variance is proportional to f². Half-Kelly therefore retains roughly 75% of the full-Kelly expected growth rate with half the variance.
Why practitioners use fractional Kelly
- The Kelly formula requires accurate estimates of expected return and variance, which are uncertain in practice. Kelly betting is sensitive to estimation error: overestimating expected return leads directly to overbetting.
- Full-Kelly drawdowns are severe enough to trigger institutional risk limits or investor redemptions before the strategy can recover.
- Most risk management frameworks (VaR, drawdown limits, notional caps) constrain position sizes to well below full-Kelly allocations.