The Design Of Stock Pairing Trading Strategy Based On Jump Regression

Since recent years,China’s economy has maintained a sustained and stable growth trend.The 19 th National Congress of the Communist Party of China pointed out in the report that China’s economy "from the stage of rapid growth to high-quality development stage,China’s economy once again made a clear path choice-change the way of development,achieve intensive economic growth,improve total factor productivity." As the largest economic power in the world,China’s securities market has experienced more than more than 20 years of development,from the system is not perfect,the system is imperfect,the scale is not strong,to the present relative norms perfect,the system risk is significantly reduced.However,as an important part of the financial system,the overall scale of the securities industry is still small relative to banks and trusts,and it is still in the initial stage of development.A large number of domestic and foreign scholars and experts have proved that China’s stock market has the characteristics of frequent fluctuations,large amplitude and long lasting.There is a relatively high chance of speculative arbitrage in China’s stock market.In this paper,the Jump regression theory is applied to China’s securities market to find the phenomenon of jumping regression in the securities market,and the applicable financial scene is found for the jump regression theory.It not only takes into account the existence of the phenomenon of jump regression,but also takes into account the stability of the phenomenon of jump regression,which makes the application of its theory more scientific.In the design of stock pairing trading strategy in the stable industry,it is found that not all stocks can be paired successfully,whether based on cointegration theory,minimum distance or enterprise index,combined with the correlation of stock price sequence,and not all shares are successful.I n the design of stock pairing trading strategy between industries,the number of successful pairs of shares is even smaller.On the characteristics of paired stocks,there is also asymmetry in the shares of paired stock groups,and there will be multiple matching phenomena,especially in the matching trading strategy between industries,this phenomenon is more obvious.For the selection of the spread pattern through the paired stock group,it is also found that not all paired shares can be filtered.In the two-year period from June 1,2016 to June 1,2018,the number of paired shares selected through the spread is not ideal,whether it is paired in the industry or traded between industries.Comparing the performance of three traditional pairing trading strategies,it is found that pairing trading strategies based on enterprise indicators perform better.Therefore,as the goal of optimization,the tendency score matching model and relative spread are used to optimize the strategy.In the process of optimizatio n,it is found that neither of these optimization models has improved the well-performing pairing trading model.In the industry strategy,the tendency score matching model has obvious improvement effect,but the relative spread model not only does not enhance the effect but also weakens the strategy effectiveness;In the strategy between industries,the tendency score matching model and relative spread model not only improve the strategy yield,but also increase the number of transactions.By comparing the strategies in the stable industry with the strategy effect in the unstable industry,it is found that the pairing strategy in the unstable industry is underperforming in four aspects of yield,risk,feasibility and stability,which further explains the po sitive role of the jump regression phenomenon in the application of the securities market.Finally,the LSTM model is used to predict the spread of paired stocks,and the prediction results have certain reference significance for investors’ behavior in the future.