Research Of Quantitative Stock Selection Strategy Based On Complex Network

In recent years, complex networks become one of the focus research on financial market, but most of them focus in studying the stock market association network statistical properties and topological structure, the application of complex network theory in the quantitative selection strategies in literature is still relatively small. This paper used the theoretical model of complex networks to build the constructed network the relationship between stock price of CSI 300 Index constituent stocks, and then made cluster analysis. Next, this paper used information entropy index to select stocks, and constructed block mean variance model based on complex network to solve portfolio problem. Last, this paper used the complex network theory in the quantitative selection strategies.Research shows that the stock market is a complex network. The associative network constructed with constituent stocks of the CSI 300 Index is showing tree distribution. Stock center of the network is the core, and other stocks is around the stock center to the surrounding stretch. The phenomenon shows that the local area around the stock and stock center has a very strong correlation, indicating that local aggregation properties of the CSI 300 constituent stocks are strong.ln the period of 2010 January 4 to 2016 April 25, the research construct complex network and make clustering analysis besed on constituent stocks of CSI 300 Index, and then use information entropy index to select 35 stocks to invest portfolio. The portfolio strategy get maximum benefit rate of 201.34%, and compared with strategy of selecting cluster center stocks, the profit rate of the strategy increases 22.88%, significantly outperforming the CSI 300 index number of rate of return -10.56%. The mean variance model based on complex network block structure is constructed to solve the optimal frontier of investment portfolio, and the covariance matrix based on block structure can deal with the situation when covariance matrix is singular matrix, and the computation efficiency is greatly improved. The results show that the optimal portfolio can be obtained when the covariance matrix of the 8 block structure is constructed.The research for quantitative investment provides a brand new way of stock, which on the application of the future of complex networks in the quantitative stock has important theoretical significance.