K-Means Clustering Approach for Stock Risk Assessment and Portfolio Construction: A Case Study Based on the EU-EV Risk Model

The purpose of this work is to explore the approach of K-means clustering for stock risk classification and efficient portfolio construction through a case study, using data from the PSI index from January 2019 to December 2022. The classification of stock risks will be based on the expected utility, entropy and variance (EU-EV) risk model. The methodology consists in applying K-means clustering with EU-EV risk related attributes to the stocks of the Portuguese PSI index in order to obtain two classes of stocks categorized with low and with high risk. Equally weighted cluster based portfolios are built for each risk class. The performance of these portfolios will be compared with the performance of portfolios constructed by selection of the best EU-EV risk ranked stocks.