Empirical Study On Pairs Trading Strategy Based On Stochastic Control Method

Pairs trading strategy as a kind of statistical arbitrage strategy,over the years has been widely used in various markets abroad.With the development of margin trading and short selling business,as well as Shanghai-Hong Kong Stock Connect and Shenzhen-Hong Kong Stock Connect business,The pairs trading strategy has the application condition in the Chinese market,but there are few researches on pairs trading strategies for my country’s stock market.Based on this background,this paper studies the pairs trading strategy based on the stochastic control method,and conducts an empirical analysis in the A-share market.This paper constructs a pairs trading model based on the stochastic control method,assuming that the spread sequence of the stock pair satisfies the Ornstein-Uhlenbeck process,with the goal of maximizing the net present value of the return,and setting the transaction threshold to control the transaction.The optimal threshold problem is transformed into a stochastic control problem,and the corresponding HJB equation is solved to obtain the optimal transaction threshold.Three parameter estimation methods are used: least squares estimation,maximum likelihood estimation and maximum likelihood estimation of Jackknife,and it is verified that the maximum likelihood estimation of the knife-cut method is the most suitable estimation for the model in this paper.Then,it conducts empirical analysis on the underlying Stocks for margin trading and short selling of A-share,using correlation and cointegration to screen stock pairs,and conducts transaction backtesting based on cointegration method to compare their backtest results.The empirical results show that the annual return rate of the stochastic control method is significantly better than that of the co-integration method and CSI 300 index,and the average annual return rate of the stock pairs in the same industry is 22.763%.The annual volatility of stochastic control method is slightly better than that of cointegration method and CSI 300 index,and the maximum retracing rate is only 14.405%,while the maximum retracing rate of CSI 300 index is 32.5%,indicating that the stop-loss level set in the model can effectively control the risk.This paper also analyzes the industry distribution of high-yield stock pairs.Different from some previous research results,the analysis shows that stocks in different industries are more likely to achieve high returns.