Identifying influential nodes based on fluctuation conduction network model

Abstract

Identifying influential stocks and determining the diffusion mechanism in a complex stock network are helpful for recognizing and avoiding the risk of a financial crisis. We define the influence of a stock according to the price fluctuation that it triggers in other stocks. Meanwhile, we propose the fluctuation conduction network (FCN) model, a novel dynamic model that can provide an econometrics basis. Through the data of the closing price, we analyze the price fluctuation influence capacity (PFIC) of stocks. First, we test the validity of our method and compare the PFIC of stocks with other features of stocks; second, we rank the stocks and make an empirical analysis of the influential stocks. From this, we find the following: (1) The closeness centrality has a tight correlation with the PFIC. (2) From the individual stock level, the 10 most influential stocks in the giant component of the stock network have strong leadership and reputations in China. (3) From the industry sector level, “Finance” and “Electric, thermal, gas, water production and supply” are the most influential sectors. (4) Most stocks reach their maximum influence range at step 3 of the price conduction.