Research On Pair Trading Strategy Based On Factor Model

With the rise of computer technology and the continuous improvement of China's financial market,quantitative investment as a scientific and effective investment method has been highly valued and widely applied.Among them,factor strategy and pairing transaction are very classic and well-established strategies.This paper attempts to introduce the concept of paired trading into factor stock selection.Based on the CSI 800 Index and using its constituent stocks as a sample,the Alpha factor and trading strategy were constructed to investigate whether the alpha factor constructed by the paired trading idea can effectively obtain excess returns.Firstly,this paper constructs the MSCI Barra CNE5 multi-factor model as a structured risk model.The stock factor exposure is calculated based on the A-share historical market and fundamental data,and the stock return rate is decomposed into factor returns by weighted least squares.The correlation coefficient matrix is estimated by the 63-day time window rolling,and the alpha factor is constructed by combining top50,threshold setting and cointegration test.Then test the alpha factor,first calculate the factor grouping income by investigating the monotonicity of the income,and then calculate the information coefficient of the factor's future earnings information coefficient and the residual factor of the risk model orthogonality to the risk model excess return.The discovery factor outperformed 2015 in 2016,and the factor has a relatively stable predictive power for excess returns.Finally,we use the alpha factor to build a trading strategy based on factor ordering.Similar to the factor test,the strategy performs better in 2016 than in 2015.Under the premise that the annualized income remains stable,the retracement and volatility are significantly improved compared to 2015.At the same time,the strategy has achieved considerable benefits relative to the benchmark.From the results,the alpha factor constructed with the paired trading idea has the ability to effectively select stocks,and the trading strategy constructed with the alpha factor can achieve considerable gains,and the factor is a relatively qualified alpha factor.The current method of constructing factors is relatively simple,and the factors need to be optimized in individual parts.In future research,factor performance will continue to be optimized and a more accurate and effective trading strategy will be designed in combination with risk models.