A sleeve at zero volatility has no inverse, so its weight is withheld and it leaves the denominator.
Risk share is weight times volatility, in percentage points of STANDALONE risk, and it is equal across sleeves by construction. Portfolio volatility is not computed here: it needs the correlations between the sleeves, which this calculator does not take.
Hypothetical illustration, computed only from the volatilities entered. These are inverse-volatility weights. Equal risk contribution equalizes each sleeve’s MARGINAL contribution to portfolio risk, needs the full covariance matrix, and is solved numerically; the two coincide only where every pairwise correlation is the same.
Volatility is estimated from a past window and moves, so a weight set on it is only as stable as that estimate. Portfolio volatility is not computed, because it needs the correlations. Nothing accounts for rebalancing cost, taxes or leverage, and the method uses volatility alone and says nothing about expected return.
Splitting money equally across sleeves does not split risk equally: the most volatile sleeve dominates the swings. Inverse-volatility weighting sizes each sleeve in proportion to one over its volatility, so every sleeve contributes the same standalone risk. The quiet sleeve gets the most money and the wild one the least.
Three sleeves at 15%, 5% and 25% annualized volatility weight to 21.74%, 65.22% and 13.04%. Each then carries 3.26 percentage points of standalone risk, which is what makes the split even.
Reading inverse-volatility weights as equal risk contribution. They match only where every pair of sleeves is equally correlated. Two sleeves that move together contribute more joint risk than their standalone volatilities suggest, and only the covariance matrix sees that.