| Abstract [eng] |
Effective management of financial assets is becoming more and more important worldwide. Financial portfolio formation is becoming more complex and requires attention to the growing number of indicators and parameters. For this reason, this work is attempting to apply genetic algorithm to optimize the securities portfolio in large volumes and multiparameter portfolio problem. The paper shows that the genetic algorithm optimizing the portfolio is able to do it efficiently enough, within a reasonable period of time. The genetic optimization algorithm is easily adaptable to the problem and is usually modified by adding new variables affecting the portfolio. The paper presents all of the genetic algorithm design approaches and discusses the most important portfolio methods described in the previous written papers. The results showed that the genetic optimization method is accurate enough in both the classical Markowitz return - risk model and complexity multiparametrinius portfolio optimization problems. Simulation of optimized portfolios of securities over time and comparison of them with the market indicators such as the S&P 500 and Nasdaq revealed that the majority of experiments has higher returns than the market at the end of the period. However, the results of the calculations also showed that this algorithm does not guarantee an exact answer. It is therefore necessary to weigh the scales the desired accuracy of the solution and the time used for the solutions obtain. |
