Research On The Leading Factors Stock Selection Model Based On Characteristics Of Stock Return Correlation Network

With the prosperous development of capital market,quantitative investment has been gradually developed in our country.And the most commonly used multi-factor stock selection model has also been continuously developed in practice.Due to the continuous changes in market systems and investors' styles,the factors in the multi-factor stock selection model have also been required timely enrichment in accordance with current market characteristics.Recently," Institutional investors huddle together for investment" has become a hot phenomenon.Under such circumstances,it's of great significance for enriching the factors system to describe the investment logic and find new impact factors implied in this phenomenon through technical means.Based on the trading data of stocks in China stock market between January 2012 and December 2017,monthly stock return correlation networks have been constructed and some network characteristics are calculated from the networks.Subsequently,we define five characteristics which are degree,weighted degree,closeness centrality,harmonic closeness centrality and eigenvector centrality as a new kind of factors which is named as “leading factors”.Furthermore,we test the effectiveness of the leading factors to explore their explanatory power to stock return.The result shows that these features have significantly positive explanatory power to stock return.Based on these new factors,we build a stock selection model with scoring method and design a trading strategy.During the period of January 2018 to September 2020,we conduct a back testing to test the effectiveness of the strategy,the result shows that the “leading factors” can achieve a return of 33.57%over the benchmark index.What's more,considering the risk dispersion principle,the stock selection model is optimized according to the heterogeneity of the stock concepts,and the optimized model achieves a return of 37.15% over the benchmark index.Finally,in order to compare the performance differences between the “leading factors” and the traditional stock selection factors,we select 15 traditional factors from four categories which are valuation,growth,capital structure and technology.Following the same steps,we get three effective factors which are turnover rate,volatility rate and the growth rate of operating profit,and then build a similar stock selection model,design a trading strategy and conduct a back testing.During the same back testing period,the“leading factors” is better than the traditional factors in the field of excess return,alpha,beta,sharp ratio and other indicators.The excellent performance of the “leading factors”indicates that this kind of factors are robust.