Research On High Frequency Statistical Arbitrage Strategy Based On GARCH-OU Process

Since 2014,China has ushered in a golden era of quantitative hedging.The fund industry association has promoted the comprehensive sunshine of private equity funds.This year is also the first year of quantitative trading.The major institutions have issued more than 600 quantitative hedging products throughout the year.Although China’s stock market suffered a stock market crash in 2015,the quantitative scale is still in a rapid growth period.In the past two years,China’s financial market has entered a divergent market,and the traditional statistical arbitrage strategy has become less and less suitable for the current changing market.In order to obtain stable investment returns,investors need to continuously improve the quantitative model,add time-varying signals,use high-frequency data to find arbitrage opportunities,flexibly set trading strategies,combine mathematical and statistical optimization parameters,etc.,will become quantitative investment.The main research method of strategy.The Shanghai and Shenzhen 300 stock index futures trading target is the Shanghai and Shenzhen 300 Index.The index has market representation and has a bilateral trading mechanism.Therefore,this paper selects the current month contract and the next month contract of the Shanghai and Shenzhen 300 stock index futures as the matching target.Statistical arbitrage research.Research on statistical arbitrage has been carried out for many years,and the arbitrage model based on cointegration theory has become a classic.With the further development of the technology era,an Ornstein-Uhlenbeck process that can be used to describe the mean return characteristics of a sequence is widely used in the analysis of financial time series.Therefore,the arbitrage of the statistical arbitrage model based on the Otnstein-Uhlenbeck process is also studied.Based on the cointegration theory model,it is usually assumed that the time series obeys the normal distribution,but in practical applications,the financial time series usually has the characteristics of sharp peaks and tails,and some even have sequence autocorrelation and heteroscedastic effects,while the Otnstein-Uhlenbeck process only considers In the case of the first-order autocorrelation of financial time series,combining the limitations of the above two models,this paper proposes the idea of combining the OU process with the GARCH model,and then constructs a third statistical arbitrage model based on the GARCH-OU process.By constructing three different arbitrage models,one is to observe the difference in arbitrage conditions of different arbitrage models;on the other hand,it is tested whether the threshold of data optimization within the sample will have a positive impact on arbitrage outside the sample.The empirical analysis of different arbitrage models reveals:(1)Within the sample: The statistical arbitrage strategy based on the GARCH-OU process has the best arbitrage performance,and the annualized rate of return is as high as 365.89%.Outside the sample: The statistical arbitrage strategy based on the OU process has the best arbitrage performance and achieved an annualized rate of return of 91.64%.The statistical arbitrage strategy based on cointegration is the worst in both cases.The strategy outside the sample runs out of the same period of the Shanghai and Shenzhen 300 index;(2)the conditions for the three arbitrage models to maximize the Sharpe ratio in the sample.The optimal threshold obtained under the test is applied to the outside of the sample for backtesting,and both have positive yields.Therefore,the threshold of optimization within the sample has reference and reference significance for the results outside the sample.