Application Investigation of Genetic- Nelder-Mead Hybridized-Heuristic Algorithm in Portfolio Optimization

Authors

1 Assistant Professor, Department of Industrial Engineering and Management, Shahrood University of Technology, Shahrood, Iran

2 MBA Student, Department of Industrial Engineering and Management, Shahrood University of Technology, Shahrood, Iran, Corresponding Author

Abstract

Markowitz portfolio model still is the dominant approach in the investment profession and
scientific approaches. Contrary to the growing use of portfolios and in spite of the rich
literature on the subject, there are some problems and unanswered questions. How to
select the stocks of a portfolio is a matter of controversy; besides, the approach to
optimize the selected potfolio is a sub-group of this controversy. The aim of this work is
to be a useful instrument for helping finance practitioners and researchers with the
portfolio selection problem. While investigating major methods ever used in optimization,
classics and heuristics, a hybridized algorithm, consisting of the combination of two
heuristic algorithms, is applied to the portfolio selection problem in this paper. Portfolios
are selected and optimized in Tehran stock exchange for the stocks of top 35 companies
out of top 50 companies. The results indicate that the hybridized algorithm is adaptable to
the portfolio selection problem, and in contrast to optimization via genetic algorithm, the
hybridized algorithm holds a better convergence speed and owns a more reasonable riskreturn
performance. The research findings also show that in a comparison between the
hybrid-based constructed portfolios, although the convergence speed and degree of
diversification for the monthly selected portfolio outperforms the other one, the annual
selected portfolio holds a better performance from risk-return point of view.

Keywords