Time Series Clustering for Portfolio Selection

Nowadays there is an increasing interest to study phenomena that change over time. Within the framework of cluster analysis, a large number of methods dealing with time series have been proposed. In this paper we use a new clustering proposal that exploits the properties of P-splines ensuring excellent performance of resulting partitioning procedure in a reduced computational time. Focusing our attention on financial time series, we show how this algorithm can be used to build a financial portfolio. Our aim is to propose a methodology that can be helpful in the investment decisions of portfolio manager.