Directed association network analysis on the Standard and Poor's 500 Index

In this paper, we study the association between the core American listed companies by analysing the stock data of the Standard and Poor's 500 Index. During the network analysis, we use a new correlation coefficient (Chatterjee in J Am Stat Assoc 116(536):1-21, 2020) to construct the directed association network and apply the directed spectral clustering on ratios of eigenvectors method (DSCORE) (Ji and Jin in Ann Appl Stat 10(4):1779-1812, 2016) for community detection. The obtained three communities are: "traditional" community, "intermediate" community, and "advanced" community respectively. We continue to analyse the entire directed association network and three communities by the node degree, and further study the companies of the central location of networks or associating within their own community or through the entire directed association network. Our results present a rational and particular community detection analysis of the financial market network. The microeconomic information hidden in stocks is successfully reflected in the associations between the American listed companies. The findings are also helpful to understand the United States market.