Volatility targeting is a position sizing method that aims to keep a portfolio’s expected volatility near a predefined level.

Volatility targeting is a position sizing method that aims to keep a portfolio’s expected volatility near a predefined level. Instead of holding a fixed dollar exposure through changing market conditions, the portfolio scales exposure up when forecast volatility is lower and scales exposure down when forecast volatility is higher. In that sense, it converts fixed-size exposure into fixed-risk exposure.[1]

The basic mechanism is a scaling rule. A portfolio first produces a set of signals or weights. It then estimates the portfolio’s forward-looking volatility from recent returns, model-based forecasts, or another volatility estimator. The portfolio exposure is multiplied by a factor equal to target volatility divided by forecast portfolio volatility. When forecast volatility is above target, the factor is less than one and gross exposure is reduced. When forecast volatility is below target, the factor is greater than one and exposure is increased.[1]

This framing matters because portfolio risk is not constant through time. A fixed allocation can experience very different realized volatility across regimes even if the nominal weights do not change. Volatility targeting addresses that instability by making sizing responsive to the estimated risk of the portfolio rather than to dollar notional alone.[1]

A practical implication is that realized volatility should cluster more tightly around the chosen target than it would under a fixed-size implementation, assuming the volatility forecast is informative. The quality of the outcome depends on forecast quality, rebalancing frequency, transaction costs, and any leverage or position constraints. If the volatility estimate is stale or noisy, the scaling rule can underreact or overreact, which moves realized risk away from target.[1]

Forecast error is therefore central to the method. If forecast volatility is underestimated, the portfolio will scale up too much and realized volatility can exceed the target. If forecast volatility is overestimated, the portfolio will scale down too much and realized volatility can fall below target. The formula is simple, but its effectiveness depends on how well the volatility model tracks changing conditions and on whether the estimate is robust to regime shifts.[1]

Volatility targeting is often evaluated with backtests, but any such evidence needs careful statistical interpretation. A backtest can look better after adding a volatility scaling rule for reasons that do not survive out of sample, especially if the rule, the lookback window, or the estimator choice was tuned repeatedly. Sonar’s overfitting audit framework emphasizes testing whether observed performance is statistically credible after accounting for multiple trials and selection effects rather than accepting an apparently strong backtest at face value.[2]

One tool for that evaluation is the deflated Sharpe ratio. The deflated Sharpe ratio adjusts the interpretation of a Sharpe ratio by accounting for non-normal returns, limited sample length, and the fact that many variants may have been tried before reporting the best result. In research on volatility targeting, this matters because scaling parameters and forecast choices can create a large family of nearby specifications. A conventional Sharpe ratio alone may overstate confidence in the improvement. The deflated Sharpe ratio is intended to test whether the observed Sharpe ratio is high enough to be statistically distinguishable from results that could arise from luck under repeated testing.[3]

That makes the right conclusion narrower and more useful. Volatility targeting is a risk budgeting technique. It is designed to make the portfolio’s risk profile more predictable by linking exposure to forecast volatility. Whether it improves risk-adjusted results in a given strategy is an empirical question that should be examined with robust out-of-sample testing and with statistics that address backtest overfitting, including the deflated Sharpe ratio.[2][3]

Covered in depth in the Strategy research fundamentals pillar hub.