Portfolio Optimization Based on Semi Variance and Another Perspective of Value at Risk Using NSGA II, MOACO, and MOABC Algorithms

This study examines the criterion of value at risk from another perspective and presents a new type of mean-value at Risk model. To solve the portfolio optimization problem in Tehran Stock Exchange, we use NSGA II, MOACO, and MOABC algorithms and then compare the mean-pVaR model with the mean-SV model. Given that, finding the best answer is very important in meta-heuristic methods, we use the concept of dominance in the discussion of multi-objective optimization to find the best answers and show that, at low iterations, the performance of the NSGA II algorithm is better than the MOABC and MOACO algorithms in solving the portfolio optimization problem. As the iteration increases, the performance of the algorithms improves, but the rate of improvement is not the same, in a way, the performance of the MOABC algorithm is better than that of the NSGA II and MOACO algorithms. Then, to compare the performance of the “mean-percentage of Value at Risk” model and the “mean-semi variance” model, we examine both models in the standard mean-variance model and show that the mean-pVaR model, compared to the mean-SV model, Has better performance in stock portfolio optimization.