The Analysis And Empirical Research On Statistical Arbitrage Of Shanghai Hong Kong Stock Connect A+H Shares Based On High Frequency

In recent years,with the gradual improvement of China's financial system and the continuous development of capital market,domestic financial market has been widely concerned in the world.As a neutral strategy based on the mean value recovery of price difference sequence,statistical arbitrage is favored by investors because it can obtain relatively stable investment income at a lower risk level.With the formal implementation of the Shanghai Hong Kong Stock Connect program,the degree of integration between the mainland and Hong Kong stock markets has become increasingly close,and the linkage between A-shares and H-shares has been continuously strengthened.At the same time,it also provides the feasibility for the realization of cross market arbitrage.In this paper,64 pairs of stocks which are listed in both A-share and H-share markets are selected,and the correlation coefficients of each frequency are calculated based on the highfrequency data of 5 minutes,15 minutes and 30 minutes.Finally,after comprehensive consideration,Zijin mining,which has the most high correlation,is selected as the research object.Followed by cointegration test,the sample data of Zijin Mining is proved that it has a long-term equilibrium relationship,and an error correction model is established to describe its short-term volatility characteristics.By using the cointegration regression equation,the spread series of A+H stock pairs is obtained,and the arbitrage position ratio is determined according to the cointegration coefficient.This paper mainly uses two methods to implement statistical arbitrage strategy.The first method is to build GARCH model to realize the dynamic statistical arbitrage trading strategy.The GARCH model is used to describe the heteroscedasticity fluctuation characteristics of the spread series.According to the time-varying standard deviation,the optimal threshold is established to maximize the return,and the opening trading signal is set.The second method is to divide the market into two states: low volatility and high volatility,and establish Markov regime-switching model.At the same time,according to the time-varying variance characteristics of the spread series,the Markov regime-switching model and GARCH model are combined to construct the MS-GARCH model.According to the two different volatility characteristics of spread series and different parameters in each state,trading signals are established for arbitrage.This paper analyzes the arbitrage strategy by setting up the income evaluation indexes such as arbitrage success rate,total return,annualized return,and risk evaluation indexes such as annualized volatility,sharp ratio and max drawdown.The results show that the statistical arbitrage strategy based on GARCH model and Markov regime-switching model are both effective,and the arbitrage opportunities in the volatility spread can be well captured.At the same time,the use of high frequency data can better capture the arbitrage opportunities in the volatility spread and obtain relatively considerable income.This also reflects that under the background of the smooth implementation of the "Shanghai Hong Kong Stock Connect" program,the linkage between the mainland market and Hong Kong market is significantly enhanced,and there are still a lot of cross market arbitrage opportunities between A shares and H shares.In addition,this paper analyzes the high frequency data at 5 minutes,15 minutes and 30 minutes interval of Zijin Mining,and the results show that the higher the frequency,the worse the effect of arbitrage.Considering the return indexes and risk indexes,30-minute frequency data performs the best while 5-minute frequency data performs the worst with two different arbitrage strategies,which breaks the traditional impression that the higher the data frequency is,the better the statistical arbitrage effect is.Comparing these two arbitrage strategies,we find that they have their own advantages.From the perspective of return,the dynamic statistical arbitrage strategy based on GARCH model performs better,while from the perspective of risk,the effect of Markov regimeswitching model is better.Therefore,we can make targeted choices according to different risk preferences of investors.