A Combined Clustering and Multi-Criteria Approach for Portfolio Selection

One of the key role of a portfolio manager is to identify a suitable asset allocation strategy. The portfolio composition task has to take into account the return and the risk of each asset along with several other factors such as investor’s aims, market expectations and risk tolerance. Thus the final decision making can be naturally viewed as a multiple criteria problem whose solution can benefit from Multiple Criteria Decision Making strategies. In this paper, we propose a two-step multi-criteria approach to support the selection of equity portfolios. Our strategy exploits a varying-coefficient Capital Asset Pricing Model framework. First, we identify clusters of stocks having similar systematic risk factors, then we rank these assets using an ELECTRE III method. We also present a real data example on stocks of S&P500 to illustrate the proposed methodology.