Sectorial portfolio analysis with genetic algorithms: case applied to the Mexican Stock Exchange

The type of industry, size of company, number of employees, etc. are variables that are considered as control variables in a large number of articles. In this research we consider the sector variable as a determinant of financial performance (Baird et al. 2012) and the risk (Artikis and Nifora, 2011) rather than as a control variable. This paper analyzes six sectors of the Mexican economy divided according to the Mexican Stock Exchange: industrial, basic consumer products, materials, non basic consumer products, telecommunications and financial services. The sample consists of Mexican companies, that is, 30 companies in the 2007-2012 period. To measure portfolio performance two classic indicators are used: (1) Jensen alpha and (2) Sharpe ratio, and also conditional metrics are used that measures the number of times the portfolio return exceeds the market average. The goal is to find a portfolio that maximizes these parameters and compare the results between the different sectors under study. Due to a nonlinear programming problem, genetic algorithms are used to obtain the optimal portfolio that maximizes these metrics. The results show a better risk-adjusted financial performance in the field of materials and financial services and a lower performance in such sectors as the industrial and telecommunications ones.