Cardinality-Constrained Portfolio Optimisation with Evolutionary Algorithms = Kardinalitáskorlátos portfólióoptimalizáció evolúciós algoritmusokkal

Absztrakt (kivonat)

Modern Portfolio Theory originated in 1952 from the ground-breaking research work of Harry Markowitz, which made a significant contribution to financial decision making by quantifying the risk-return relationship. The Markowitz model became an important foundation for many researchers seeking solutions to increasingly complex portfolio optimisation problems, however, there is no single method that would be considered the best. This study investigates evolutionary computing as a potentially effective method for portfolio optimisation. Evolutionary computing is an optimisation technique inspired by the natural evolution process with the idea of ‘survival of the fittest’ at its core. To assess the effectiveness of the proposed method, a genetic algorithm has been designed for a cardinality constrained portfolio optimisation problem based on the Markowitz model. In the same simulation environment, the performance indicators of the genetic algorithm were evaluated in relation to those of a mixed integer programming algorithm, a more widely used approach to solving the Markowitz portfolio optimisation problem. The results demonstrate that while genetic algorithms underperform mixed integer programming in less complicated environments, in larger investment universes they are capable of finding better portfolios in the presence of time constraints. The research findings suggest that evolutionary computing can be a good alternative to conventional tools in more complex optimisation problems, opening the possibility to further research with more diverse investment preferences as well.