Establishment And Optimization On Multi-factor Quantitative Model

The stock market is constantly changing,so from the current point of view,it is of practical significance to explore which indicators are more effective in quantitative stock selection.This is also the original intention of the research work in this paper.Maintaining data rigor,neutrality,and the scientific nature of data processing methods throughout the research process is critical to the accuracy of research findings.In this paper,a comprehensive candidate factor library is constructed firstly.After a series of processing such as outliers,missing values,standardization,market value neutralization,and industry neutralization.the validity of the candidate factors is tested.And re-screening,finally we think EP TTM(net profit TTM/total market value).BP_LF(net asset TTM/total market value),operete_profit_gr_TTM(operating profit growth rate TTM),sales_gr_TTM(business income growth rate),BOOK LEVEL(book leverage),stock_to_market volatility(the ratio of individual stocks to market volatility).REV_LASTIM_MAX(maximum return to daily yield),and RSI(relative strength)are more effective factors.After the effective factors are obtained,the scoring method is used to construct the eight-factor stock selection model.The backtest interval is selected from January 2016 to April 2019.Through the back test of the model,we found that the eight-factor model constructed in this paper has a relatively stable excess return compared to the Shanghai and Shenzhen 300.During the entire back test period,the total return of the eight-factor model was 41.02%.and the annualized return was 11.07%.Compared with the benchmark income.the eight-factor model has a relative total return of 32.28%and an annualized average excess return of 8.92%.In order to conduct in-depth research,we optimized the original model.The optimization scheme is:"Before using the eight-factor model for formal stock selection.the stock pool is first screened by the growth rate-net cash flow generated by operating activities(TTM)indicator.".By back testing the optimized eight-factor model,we found that the optimized eight-factor model was better than the original eight-factor model.Compared with the Shanghai and Shenzhen 300.the optimized eight-factor model has a relatively stable excess return.During the entire back testing period,the total benefit of the optimized eight-factor model was 47.89%,and the annualized return was 12.70%.Compared with the benchmark income,the optimized eight-factor model has a relative total return of 39.15%and an annualized average excess return of 10.62%The final work of this paper is to compare and analyze the eight-factor model,the optimized eight-factor model and the classic Fama-French three factors.After back testing,we found that the optimized eight factors established in this paper perform better in the backtesting interval than the classic three-factor model.The total return.the annualized return,the relative total return,the annualized average excess return,the Alpha,the annualized Sharpe ratio,and the Sotino ratio of the eight-factor model are better than the optimized three-factor model's corresponding indicators.The investment strategy established in the interval based on the optimized eight-factor model has stronger profitability.At the same time,the optimized eight-factor model has smaller annual volatility,the absolute value of maximum drawdown and downside risk,indicating that the investment strategy based on the optimized eight-factor model in the backtesting interval is more stable than the Fama-French three-factor model.