Research On Stock Investment Strategy Based On Complex Network

With the continuous development of China's stock market,the regulations of two major stock exchanges in Shanghai and Shenzhen are becoming more and more perfect.The number of listed companies in China has sprung up,showing an intricate and inexplicable relationship network.The emergence of complex network theory and the continuous emergence of research outcomes have made the research on the connection of abstract things more effective,and set off a wave in the academic world.Many scholars have constructed stock market networks based on the complex network theory and have reached many valid conclusions by further analysis.However,this field is still in the preliminary stage of exploration,and there is still a lot of research space and value.Therefore,the paper uses complex network theory to model the stock market,bringing a new perspective to the methods of the construction of investment portfolios.The paper takes the Shanghai and Shenzhen A-shares of China as the research object,so we collect the closing price data of stock index of the Shanghai Stock A and Shenzhen Stock A and their constituent stocks from January 04,2016 to December 31,2018.Listed below is the work done: Firstly,the structural characteristics of each index are tested in order of stability,normality and auto-correlation,etc.According to the test results,we select the R/S method to determine the non-cyclic period length T of the stock index sequence,and short-term trends are forecast based on the stock index included in the time window t(10)-t T],1[ in order to incorporate limited predictability in the stock market into the portfolio building process.Secondly,according to the results above,periodic stocks are selected as an alternative to the portfolio when the predicted stock index trend shows an upward trend,otherwise defensive stocks are selected as an alternative.At this point,the first level of screening of the portfolio has been completed.Thirdly,based on the correlation coefficient threshold method and planar maximally filtered graph(PMFG)method,the stocks selected through the first layer are used as nodes to construct networks,and the dynamic evolution analysis is performed on them.By comparing the advantages and disadvantages of the two network construction methods described above,we finally select the network constructed based on the PMFG method as the basis for the subsequent analysis.In addition,the FN algorithm is applied to divide the PMFG network constructed above into communities.Combined with complex network indicators of node degree values,we propose a stock selection strategy taking investor risk appetite and risk tolerance into consideration.So far,we complete the second layer of portfolio selection.Associated with the results of community division,the third layer of portfolio selection is completed.The above is the entire portfolio construction process.The research shows that the Shanghai and Shenzhen A-share markets have the characteristics of "spikes and fat tails".The non-cyclic period length of the Shanghai A Index(Shenzhen A Index)is 95(126)trading days.The stock market network has significant linkage and the correlation between stock nodes is strong and there are key nodes.What's more,there is a clear positive correlation between the degree value of the node and the investment risk.Investors can use the complex network indicators such as the degree value of the node and the results of community division to screen out of the corresponding stocks.By this way,a portfolio that meets their own risk appetite and risk tolerance can be built.According to the research process and research results of the paper,corresponding suggestions are finally given to stock market regulators and stock investors.