VISUALIZING AND CLUSTERING FINANCIAL PORTFOLIOS USING INTERNAL COMPOSITIONS

A method to cluster financial portfolio according to their internal composi- tion is introduced. Portfolio internal compositions are estimated through a style analysis model. Starting from such compositions, we cluster portfolios exploiting archetypal data analysis. Visualization tools are based on parallel coordinate plots and on asymmetric maps. The proposed procedure is based both on static and dynamic visualization tools. Final aim of such a procedure should be to compare and classify portfolios according to their similarities and to compare the evolution of the portfolio management along time.