The Strength of Influence Ties in Stock Networks: Empirical Analysis for Portfolio Selection

This paper presents a novel method for portfolio selection in financial markets using a network-based approach. The proposed method, named k-Influential Nodes (k-IN), leverages complex network structures to model the linear influences among stocks in the financial market. We used directed graphs to capture the interdependencies among assets and applied them to perform a portfolio selection. To calculate the influence between two assets, we employed different time-dependent cross-correlation measures. The k-IN method uses shifted log-return data to measure the influence of the co-movement of stocks and examine the strength of the links to select influenced stocks and create the portfolio. The empirical analysis conducted in this study highlights the efficacy of leveraging influence ties within financial networks for portfolio selection. We provided an empirical analysis of six market indices from America, Europe, and Asia to assess the performance of the proposed method. We also presented financial results using k-IN for filtering stocks to perform Mean-Variance Analysis (MVA) for portfolio selection, named k-IN++. Results suggested that the proposed method, despite being simple and easy to compute, presents superior results compared to the results of the three baseline methods. The proposed k-IN method offers a valuable contribution to the field of data mining and presents a promising approach for enhancing investment strategies in financial markets.