LAUSR

In this paper, we discuss the stock network construction problem under simultaneous consideration of linear and nonlinear relations between stocks. A novel method based on the conditional probability is proposed to describe the nonlinear relation between stocks. Furthermore, by considering both the linear and nonlinear relations between stocks, a multilayer network is constructed to characterize stock market, in which Pearson correlation network, Granger causality network, and our proposed nonlinear relation network are combined. Finally, several experiments are conducted to illustrate the effectiveness of the proposed approaches. The results show that the proposed multilayer network not only covers more nodes than the Pearson correlation network, but also better balances the relation between prediction accuracy and the number of predictable nodes.