Meta heuristics for dependent portfolio selection problem considering risk

Research highlights? We develop a new mathematical model for portfolio selection problem in which the revenue of projects is stochastic. ? Two Meta heuristic algorithms (Electromagnetism-like and Genetic algorithms) are developed to solve the problem. ? Genetic algorithm shows better performance in quality of solution and CPU run time. Portfolio selection is one of the most important problems which human, companies and organizations are in dealing with. Dependency between projects is a major issue, which has not been considered widely according to previous researches. Dependency often occurs in industrial and constructional projects. In this paper, we developed a model that included cost dependency. Moreover, some technological constraints have been considered. Stochastic revenue and risk are some other major points of the model. An Electromagnetism-like (EM-like) and a Genetic Algorithm (GA) are developed to solve the proposed model. Some experimental tests are developed to examine the influence of algorithm's parameters on their performance. In addition, the results of the algorithms are reported. Comparison of GA and EM-like algorithm with optimum answer shows the efficiency of the algorithms. In addition, it reveals that GA has better performance in comparison with EM-like algorithm.