Inverse Volatility Weighting

Inverse volatility weighting sets each asset's portfolio weight proportional to $1/\sigma_i$, where $\sigma_i$ is that asset's own volatility, then rescales so the weights sum to one. A jumpy asset gets a small weight and a calm asset gets a large one, so no single position dominates the portfolio's swings just because it happens to be more volatile than the rest. It ignores correlations between assets entirely and uses no expected-return forecasts — it is purely a risk-based sizing rule, not an attempt to maximize return.

The formula for two assets is $w_i = \frac{1/\sigma_i}{1/\sigma_1 + 1/\sigma_2}$. Suppose asset A has volatility 30% and asset B has volatility 10%. Then $1/\sigma_A = 3.33$ and $1/\sigma_B = 10$, giving weights of $3.33/13.33 \approx 25\%$ in A and $10/13.33 \approx 75\%$ in B — the calmer asset gets three times the allocation of the volatile one, which is the ratio $\sigma_A/\sigma_B$ inverted.

Because it ignores correlation, inverse volatility weighting is not the same as true risk parity, which additionally accounts for how assets move together; it is best thought of as risk parity's simplest special case, used when correlation estimates are unreliable or the portfolio is small enough that pairwise correlations matter less. It also says nothing about expected returns, so an inverse-volatility portfolio can hold a large weight in an asset with a poor return outlook simply because that asset happens to be calm — the method optimizes risk contribution, not expected profit, and is usually paired with a separate view on returns rather than used as a complete allocation rule on its own.