An Approach for Diverse Group Stock Portfolio Optimization Using the Fuzzy Grouping Genetic Algorithm

Previously, an algorithm has been proposed for optimizing a diverse group stock portfolio using the grouping genetic algorithm. However, it is not easy to set appropriate parameters for genetic operations when datasets are different in the previous approach. In this paper, we propose an approach that can not only obtain a diverse group stock portfolio but also tune parameters dynamically for the genetic operations using the fuzzy grouping genetic algorithm. Three parts, grouping, stock, and stock portfolio parts, are used as encoding scheme. The fitness function defined in the previous approach is utilized to evaluate quality of a chromosome. To deal with parameter setting for genetic operations, a fuzzy logic controller is designed and employed to dynamically adjust them using the predefined knowledge base for deriving a better solution. Experiments on a real dataset were conducted to show the effectiveness of the proposed approach.