STOCK PORTFOLIO OPTIMIZATION IN THE MEAN-VARIANCE RISK MODEL USING GENETIC ALGORITHMS

Abstract

The Mean-Variance model is a popular model in the investment field. The MeanVariance model uses parameters in the form of returns, variants and covariance of selected shares. More time is needed when more and more assets are selected from the portfolio, so an artificial intelligence is needed to help the calculation faster. One artificial intelligence that can be used is the Genetic Algorithm. Genetic algorithm is a metaheuristic search algorithm based on the mechanism of natural selection and genetic operations to get a solution. Genetic algorithm uses operator selection, crossover and mutation. In the selection process in the search for fitness values calculated by the Mean-Variance risk model. The application of the MeanVariance risk model with genetic algorithms shows optimal portfolio performance. The results obtained are by using a crossover operator in the form of Arithmetic crossover, mutation using Reciprocal Exchange Mutation, and selection using Elistism Selection. From these operators, optimal results were obtained with a population of 170, a generation of 250, and a combination of a crossover rate of 0.6 mutation rate of 0.4. Keywords: genetic algorithm, mean-variance, portfolio, risk