Position Sizing: Fixed Fractional Basics

Fixed fractional position sizing allocates a constant percentage of current account equity to each new position, scaling exposure up after gains and down after losses. It is the baseline against which all other sizing schemes — Kelly, volatility-targeted, risk parity — are measured, because it isolates the compounding behavior of geometric returns without introducing volatility estimation noise.

The method traces to Ralph Vince's optimal-f work and has two common parameterizations: fraction of equity allocated to the position, or fraction of equity risked per trade given a stop distance. The second form is what most systematic traders mean when they say "I risk 1% per trade."

The computation

For the risk-based form, position size in units of the traded instrument is:

where f is the fixed fraction of equity placed at risk per trade (e.g., 0.01 for 1%), and the denominator is the per-unit loss if the stop is hit. For the simpler exposure-based form:

Equity is recomputed before every entry. This is the property that produces geometric compounding: a 50% drawdown halves the next trade's size, and a 100% gain doubles it.

Interpreting f

Typical values for discretionary and systematic retail trading sit between 0.005 and 0.02 of equity at risk per trade. Values above 0.02 imply either high conviction in edge or tolerance for deep drawdowns. Values below 0.005 generally indicate either a high-frequency strategy with many concurrent positions or a portfolio overlay where individual trades are not the primary risk unit.

The relationship between f and terminal wealth is non-monotonic. As f increases, expected geometric growth rises, peaks at the Kelly fraction, then declines and eventually goes negative — even with positive expectancy. The Kelly value for a strategy with win probability p and win/loss ratio b is f* = (p(b+1) − 1) / b. Most practitioners size at one-quarter to one-half Kelly because the full Kelly fraction produces drawdowns that are mathematically optimal but psychologically and operationally unacceptable.

What fixed fractional does not capture

It assumes trades are independent. Two simultaneous positions each sized at 1% of equity are not 1% risk events if their returns are correlated — in a regime shock they behave as a single 2% position. Fixed fractional sizing has no mechanism to detect or adjust for this.

It assumes the stop is reliable. Slippage, gaps, and overnight risk mean realized loss frequently exceeds entry_price − stop_price. Sizing for a 1% loss can produce 3% realized losses on bad fills. Backtests that assume stop execution at the stop price systematically understate the variance of fixed fractional outcomes.

It ignores volatility regime. A fixed-percentage stop placed at 2 ATR will produce smaller positions in high-volatility regimes and larger positions in quiet ones — but the dollar risk per trade remains constant, which is the opposite of what most portfolio-level risk models prescribe. Volatility-targeted sizing addresses this; fixed fractional does not.

Finally, it does not account for the discretization of position sizes. For futures contracts or low-priced equities purchased in whole shares, the floor function on units produces meaningful tracking error against the theoretical fraction, particularly at small account sizes.

In Kestrel Signal

Kestrel Signal exposes fixed fractional as the default sizing primitive in the strategy configuration, with both exposure-based and risk-based forms selectable per-strategy. Realized f is reported on every backtest alongside the configured value so discretization drift is visible. Drawdown distributions are computed across the full grid of f values you specify, making the Kelly-fraction tradeoff between growth and drawdown explicit rather than hidden behind a single summary statistic.