چکیده:
Portfolio selection problem which is one of the most important issues in finance, using a model that considers conditions of the real world is important. In financial markets, severe and frequent fluctuations cause frequent changes in the portfolio selection models outputs, which increases the number of times to change the weight of portfolio's assets, and so that incurs high management and transaction costs. In the literature of portfolio selection models, one of the approaches to prevent this kind of high costs is robust optimization approach. In this study, in order to optimize the portfolio, genetic algorithm and shuffled frog-leaping algorithm are used to solve robust probablistic planning model presented by Amiri and Heidari (1399) in higher dimensions. To this end, 15 specific problems with different dimensions (number of companies and time periods) are designed and processed. The results of the implementation of two algorithms on the above 15 problems were compared using T-test, which shows no significant difference between two algorithms in portfolio selection problem, but the combined approach of TOPSIS and entropy weighting selects the genetic algorithm as superior algorithm.
خلاصه ماشینی:
Stock portfolio optimization based on a robust chance programming model using Genetic and Hybrid Frog Leap algorithms Mohammad Saeed Heidari 1 Date of receipt: 99/08/23 Date of acceptance: 99/09/28 Javad Validy 2 Seyyed Babak Ebrahimi 3 Abstract In the investment portfolio selection problem, which is one of the most important issues in the financial field, it is important to use a model that can consider the conditions of real environments.
In this research, an attempt has been made to use the Genetic algorithm and the Hybrid Frog Leap algorithm to solve the robust chance programming model presented by Amiri and Heidari (2020) in larger dimensions and for the purpose of stock portfolio optimization.
Table 8- Averages of objective functions and computational time of sample problems with metaheuristic algorithms (Refer to page image) Considering the proposed levels provided for the Genetic algorithm and based on Figures 2 and 3, to increase the efficiency of the Genetic algorithm in obtaining the optimal solution, the values of ݐ݅ ݔܽܯ and ܰ must be set on the numbers 100 and 80 respectively, and the values of ܿܲ and ݉ܲ on the number 0.
The results indicate the high efficiency of both algorithms in solving the stock portfolio optimization problem; however, in comparing these two algorithms, it must be said that no statistically significant difference was found at a 95 percent confidence level between the two algorithms, but considering the TOPSIS method, which is one of the multi-criteria decision-making techniques, and the entropy method, the genetic algorithm was selected as the superior algorithm.
(
, 
, 
)
