Research On Stock Portfolio Strategy Based On Complex Network Theory

China’s stock market is in a period of rapid development and change,and quantitative stock selection strategies are being used by more and more investors in their investment portfolios.The volatility of stock prices is not only affected by intrinsic value,but also interacts with each other to build a giant complex system.As a perfect tool for studying complex systems,complex network theory can comprehensively and systematically describe the complex relationships of the stock market,and understand the correlation and dynamic evolution characteristics of stocks as a whole.Therefore,how to extract the investment logic contained in the stock market is of great significance to enrich investment theories and provide investors with more investment strategies.This paper takes the CSI 300 constituent stocks as the research object,uses the complex network theory to construct the stock correlation network,and analyzes the dynamic evolution characteristics of the stock network.Using the theory of community division and factor scoring analysis,based on the statistical characteristics of the network and the financial indicators of stocks,the investment portfolio is constructed,and the performance of the final portfolio is back-tested in six-month,one-year,two-year and three-year cycles.The research shows that:(1)During the period of large fluctuations in the stock index,the aggregation of the stock network is significantly improved,and the financial industry stocks occupy the core position of the network,which can transmit risks to stocks in other industries.The regulatory authorities should focus on monitoring such stocks.The community division is carried out on the minimum spanning tree network,and the backtest results show that the community division can effectively reduce the volatility risk of the stock portfolio.(2)The validity of the financial indicators of the stock is tested,and it is found that the price-to-book ratio,price-to-earnings ratios,the return on equity,the growth rate of net profit and the rate of return are all in direct proportion,while the degree of stock is inversely proportional to the rate of return.Metrics such as ESG scores,and price-to-earnings-to-earnings-to-earnings growth(PEG)ratios did not show a significant correlation with yields.(3)The stocks screened by the community division and index scoring method are used for portfolio backtesting.The highest yield is 268.95%,which is higher than the 191.27% of the industry’s diversified portfolio,and significantly outperforms the CSI 300 index yield of 64.10% over the same period.The longer the holding time of the experimental group,the more obvious the excess return and the higher the Sharpe ratio.This paper combines the statistical characteristics of complex networks with stock financial indicators,and proposes a novel investment portfolio strategy,which has certain guiding significance for the application of complex networks in quantitative stock selection.