GPS: A graph-based approach to portfolio selection

Abstract

Investment success depends on the performance of the stocks selected for investment. In this study, we present a novel graph-based method for selecting appropriate stocks for an optimal portfolio by utilizing the ability of graphs to model relationships between different stockssOur approach consists of presenting the performances of many different random portfolios in the form of a graph, and then applying the community detection technique to this graph for the purpose of determining suitable stocks. In a comparison of the proposed method with competing models, the evaluation results demonstrate its superiority over the top 50 stocks in the S&P 500 index.