Algoritmo genético compacto e modelo de Markowitz aplicados à otimização de portfólio em carteiras de criptomoedas.

The selection of investment portfolios is a complex task that involves choosing the assets that make up a portfolio. In the context of cryptocurrencies, this activity becomes particularly challenging due to the high volatility and complexity inherent in this market. Despite these characteristics, cryptocurrencies have gained increasing popularity in recent years, becoming an attractive investment option for many. In this scenario, there is a pressing need to address the optimization of cryptocurrency portfolio selection by exploring the potential of Estimation Distribution Algorithms (EDAs). The fundamental purpose of this work is to implement a type of EDA called Compact Genetic Algorithm, employing the Modern Portfolio Theory of Markowitz as a foundation. The goal is to optimize the composition of the investment portfolio, focusing on maximizing returns and minimizing associated risks. To achieve this, the closing prices of selected cryptocurrencies were comprehensively obtained through the CoinGecko platform, covering a historical period of three years (from 2020 to 2022) segmented into three experiments. The results obtained demonstrate that the technique was able to generate good results. Using the Sharpe ratio metric, the best portfolio identified by the Compact Genetic Algorithm in the training set ranked among the top 100 returns in the test set during the three experiments conducted in a solution space containing 1000 generated portfolios.