Performance Evaluation Of Multi-factor Stock Selection Strategy By Information Ratio

The mature of China Capital Markets has led to a growing number of rational investors,Quantitative Stock Selection is becoming more and more widely used.Multi-factor Stock Selection Model has the advantages of clear logic and outstanding stock selection effect so it is widely spreaded in both the theoretical field and the capital market.At some future date,we believe that with the data of China Capital Market become more abundant and transparent,and the high-speed computing power is easier to achieve,there will be a better application prospect of Multi-factor Stock Selection Model.The research of this paper is based on the Multi-factor Quantitative Stock Selection Model.In the study,we make some improvements about these respects,including the release system of financial statements of China,selection of data,the analysis of factor robustness,how to determine the factor weight,and the comparative analysis of each model,which can improve the profitability of the model.This paper choose 187 stocks from pharmaceutical industry as research objects,we devide the data from 2006 to 2019 into training dataset and verification dataset.We use training dataset to find the optimal factor and calculate the Comprehensive Scoring Model,and in the validation dataset we compare the investment performance of each Comprehensive Scoring Factor.First of all,we evaluate 50 initial factors from the perspectives of correlation,the profitability of portfolio,the probability of factor's earning outperform the CSI 300 Index,Sharpe Ratio,Information Ratio and maximum retraction,and we get 12 efficient factors.Then,according to the correlation coefficient matrix and the P-value matrix of the effective factors,we compare the performance of each factor to remove 4 extra factors,so the remaining 8 factors are of the best overall quality.Next,we use these 8 factors to construct a Comprehensive Scoring Model,and use the Comprehensive Scoring Model to calculate 4 Comprehensive Scoring Factors.Finally,we compare the effect of 4 Comprehensive Scoring Factors on the verification dataset and get the best performing stock selection model.After a series of study,we find that the Comprehensive Scoring Factor of S1 which use the Information Ratio to calculate the weight of factor and use the proportion of factor ranking to calculate the value of factor has the best effect of stock selection.In the data of validation set,this factor's average annualized return is 23.72%,while the CSI 300 Index's average annualized return is 9.64%.The probability of beating the market of the return of the long combination is 72.22%.On the annualized basis,its Sharpe Ratio is 0.97,and the Information Ratio is 0.45.During this period,the maximum loss of the long combination only 19.93%,while the same indicator of the CSI 300 Index is 31.86%.These results show that the effect of the improved model in China Capital Market is worthy of recognition and we can deeply study this model more in the future.