Determining the Optimal Stock Portfolio in Tehran Stock Exchange Based on Multi-Objective Evolutionary Algorithm with ϵ Error Level (ϵ-MOEA)

Classical statistical models can solve the problem of portfolio optimization and can determine the efficient frontier of investment when there are few investable assets and constraints. But these models cannot easily solve optimization problems when we consider real-world constraints. Therefore, data mining techniques such as evolutionary algorithms are important in portfolio optimization. The purpose of the present research was to solve the mean-variance cardinality constrained portfolio optimization (MVCCPO) problem using -MOEA. Thus, optimal portfolios were created using the monthly returns data of 185 companies listed in Tehran Stock Exchange (TSE) during the period 2009-2010 and the performance of these companies were evaluated. The results showed that -erorr multi-objective evolutionary algorithm ( -MOEA) can successfully solve the optimization problem. 

Keywords:

Optimal Portfolio , Mean-variance Cardinality Constrained Model , Efficient frontier , evolutionary algorithms , ϵ-erorr Multi-objective Evolutionary Algorithm