Cryptocurrency portfolio optimization through Grey-Box Gene Pool Optimal Mixing Evolutionary Algorithms

Abstract

Portfolio optimization of cryptocurrencies with Evolutionary Algorithms is a fairly new topic in financial literature. New and upcoming studies are addressing portfolio optimization problems through a wide array of evolutionary and other novel algorithmic approaches. This study compares the Gene-Pool Optimal Mixing Evolutionary Algorithm (GOMEA) with the Genetic Algorithm and the Particle Swarm Optimization through an evaluation of each algorithm’s capabilities for portfolio risk management. Specifically, we use the Conditional Value at Risk (CVaR) as our risk metric for optimization and construct an efficient frontier for the portfolios generated to examine the performance of the algorithms. Making use of both simulated and historical data, our analysis focuses on these algorithms’ capacity to manage the intricate risk/reward trade-off inherent in cryptocurrencies. We construct a theoretical framework that supports the assumption behind the preference of GOMEA and conduct an empirical analysis to test whether our assumptions hold under the two distinctive datasets. Our results suggest that GOMEA presents an overall better performance in portfolio risk management through its optimization approach of the cryptocurrency portfolios. These results underscore the potential benefits of employing advanced evolutionary algorithms that exploit the inherent interdependencies found in cryptocurrencies.