Research On Multi-correlation Network In Stock Market

With the rapid development of China’s economy,the stock market has become more complicated and diversified.Considering that the Chinese stock market is still in the emerging and transitional stage,the immature stock market itself makes the research of relevance more necessary.As an emerging research tool,network theory provides a new perspective for studying the stock market.In this paper,three different correlation coefficients are used to define Euclidean distance.The planar maximum filtering graph algorithm and sliding window technology are used to construct a dynamic and complex network of stock market.Through the analysis of the multicorrelation network topology of stock market,the dynamic changes of China’s stock market are studied.And compare the differences in structural characteristics between different correlation networks and the relationship between the actual operating conditions of the stock market.At the same time,according to the degree value of the nodes in the multi-correlation network,the stock construction stock index is selected,and the stock indexes generated by different networks are compared to reflect the actual operating conditions of the stock market.According to the centrality of the stock network,the stock structure investment portfolio is selected to compare various portfolio investment returns.Give investors advice.Through the study of multiple correlation networks in the stock market,it is found that the multi-correlation network has small cosmopolitan and scale-free characteristics.There is a strong correlation between different correlation networks,and the correlation coefficient changes have the same trend.There is a certain degree of negative correlation between the clustering coefficient and the index return rate of the network,and there is a significant positive correlation between the average shortest path and the index return rate.When the stock market is in a bull market,the degree distribution of the multi-correlation network can well satisfy the power-law distribution,and as the stock market becomes worse,the power-law distribution fit begins to deteriorate.When the stock market is stable,the network based on Kendall correlation coefficient can better reflect the operation of the stock market.When the stock market fluctuates sharply,the network based on Tail correlation is a better choice.The Pearson correlation coefficient-based network shows insensitivity to the overall state of the stock market when the clustering coefficient is analyzed.The overall stability is relatively small and the fluctuation range is small,so it is not suitable for the forecast of stock market changes.In addition,this paper analyzes the stocks based on the network model to select stocks with greater influence,and then constructs stock indexes to study the actual stock market,and compares the differences between different related networks.The results show that the stocks selected by the correlation coefficient based on the generous value method can better reflect the stock market.Secondly,this paper analyzes the selection of stocks according to the centrality of the network nodes to determine the investment portfolio,and compares the returns of different investment portfolios.The results show that the portfolios of stocks in the relevant network center position have higher expected returns.