Comparison of Metaheuristic Techniques for Portfolio Optimization under Semi-Variance Risk Criterion using t-test.

With the introduction of mean-variance model Markowitz took a giant step in modeling and optimizing portfolio type problems. But his model is based upon some assumptions that rarely they can hold in practice. For this reason, many researchers have taken steps both theoretical and practical to make some improvements to his standard mean-variance model. Up to now different risk criteria models such as semi-variance model, the absolute deviations mean model, and the variance with skewness model are introduced by various researchers. The most famous risk criterion is semi-variance model studied well and took into consideration again in this article as well. Here, authors have taken steps to optimize the semi-variance model taking the number of the portfolio shares and the ratio of portfolio in each share as the model constraints using simulated annealing and Tabu-Search meta-heuristic approaches for optimization purposes. The efficient frontier line of main constraint is drawn and the capability of these algorithms in drawing these lines using two-sampling t- test is investigated. The model uses the historical data from DAX, Hang Seng, and S&P100 from years 2007 through2009 as its input for calculation purposes as well as model analysis