The construction of multilayer stock network model

Abstract

Stock market analysis is an interesting topic and has attracted more attentions in the financial domain. Recently, complex networks are wildly applied to analysis and describe stock market, such as Pearson correlation network and Granger causality correlation network. However, the existing networks are constructed by considering only one kind of relationship among stocks, which would neglect some significant information. In this paper, we first construct three single-layer networks for stock market based respectively on Spearman correlation coefficient, grey relational analysis and maximum information coefficient. Furthermore, to describe the stock market accurately, we propose a multilayer network model by combining three single-layer networks. In the proposed multilayer network model, we first combine the Spearman correlation coefficient network and the grey relational analysis network together, and then combine the new compound network with the maximum information coefficient network. Finally, extensive experiments are conducted to demonstrate the effectiveness of the proposed networks.