Multi-objective optimization of portfolio selection using evolutionary algorithms: an empirical analysis involving return and skewness

Resumo

This article aims to compare the performance of 04 evolutionary algorithms in themulti-objective optimization of researched problem in the area of Finance: the portfolio selection.The multi-objective algorithms NSGAII, MOEAD, IBEA and GDE3 were used to simultaneouslyoptimize the mean return and the portfolio skewness. The data collected in the survey were from the companies listed in the Dow Jones Index and an in sample (2010 - 2014) and out-of-sample (2015 - 2017) analysis of the optimal portfolios of each algorithm was done.Statistical tests showed that the GDE3 algorithm presented better in sample and out-of-sample performance of the optimized portfolios.