Using a clustering procedure, we classify Italian funds ex-post on the basis of the composition of their portfolios and find that the optimal number of clusters is equal to 4. The four groups which result from the statistical classification closely match the 4-level aggregation of the 20 ex-ante...