Research On High Frequency Matching Strategy Of Chinese Stock Market Based On Mean Reversion Jump Diffusion Model

Paired trading has earned the spread through the mean recovery feature of spreads,and its effectiveness has been unanimously recognized at home and abroad.For paired trading,the most important thing is to fit the spread sequence of the two assets and determine how the spread sequence fluctuates,so as to predict the trend of the spread in advance and make arbitrage behavior.For the description of the spread sequence,foreign scholars have done a lot of research.On the other hand,the domestic description of the spread sequence is still in the exploratory stage.Therefore,this article will focus on the statistical arbitrage of domestic stocks,explore the volatility form of the spread between stock assets,and count Arbitrage strategy is applied to the Chinese stock market and its effectiveness is explored.On the other hand,when describing the spread sequence,this article focuses on exploring the impact of the jump component in the spread sequence on the spread sequence.Because it has been proved that the price difference sequence is a random fluctuation process,this article adopts the classic model describing the random process:the OU process,plus the jump diffusion model describing the jump term,to construct the mean-reverting jump-diffusion model,namely MJD(mean-reverting jump-diffusion)Model to describe the sequence of price differences between two ETF assets.This way,on the one hand,it can reproduce the randomness of the spread sequence and capture the jump phenomenon on the spread sequence,thereby better estimating the parameters of the spread model,which is conducive to better realization of statistical arbitrage,and on the other hand,it can ensure the stability of the spread sequence,To meet the premise assumptions of the OU process.When estimating the parameters of the spread sequence,since the obvious jumps in asset spreads mainly come from overnight changes,this article will be divided into two cases:considering intraday jumps and not considering intraday jumps.On the basis of theoretical analysis,choose not to consider intraday jumps.Follow-up research on the MJD model.On the whole,based on domestic and foreign research and the background of the Chinese market,this article selects the Shanghai and Shenzhen 300 ETF,CSI 500 ETF,and SSE 50 constituent stocks for research.Because these assets are large and active,they are very suitable for the research direction of this article.At the same time,this article selects 5 minutes of highfrequency data.This is the first time that the MJD model is applied to the high-frequency trading of my country’s stock market.The MJD model is constructed for the spread sequence,parameter estimation and pairing transactions are performed,and the traditional minimum distance model,The cointegration model is compared with the paired trading strategy constructed by the pure OU process.