Research On Paired Trading Strategy Based On Cointegration Theory And GARCH Model

In 2018,A-shares plummeted,market systemic risks increased,various funds sought shelter,or bought bond assets,or bought precious metals such as gold to hedge,and some institutions developed various quantitative strategies in response to changes in market style,paired trading strategies are one of them.This strategy has been widely used in mature European and American markets.Specifically,although the stock price series with cointegration relationship are random,their price difference is in a reasonable range.When the price difference is higher than this price range,the short-selling combination is used.when the price difference is lower than the price range,the pairing combination is made,thereby earning the benefit of the spread convergence.This strategy has a significant advantage in effectively avoiding market systemic risks through hedging mechanisms,similar to choosing short-selling stock index futures in order to hedge risks in a market unilateral falling environment.There are two important aspects to this strategy.The first is to choose the right paired stock and establish the relevant trading model.The second is to respond to the measures that the cointegration relationship will be invalid at any time,to continuously optimize the strategy and make it realistic..The trading strategy is mainly divided into the following steps: selecting a stock pool,screening suitable matching combinations from the stock pool,establishing a trading model,and historical back testing.In the past literature,scholars often used three methods to construct paired trading strategies,namely,cointegration theory,minimum distance method and random price difference method.This paper chooses the cointegration theory to analyze.Although the time series itself is non-stationary,the linear combination between them may be stable.If the combination has a long-term stable equilibrium relationship,they have a cointegration relationship.In this way,the sequence of the spread can be calculated and used as the basis for opening the position.Taking the data from 2012 to 2018 as the research sample,the first two years are the Pairing combination.the second half is the trading period,and the matching combination is updated every six months,taking into account the securities lending and other transaction costs.Taking 18 years as an example,the paper introduces the relevant steps and methods.Through the data selection of 2016.01.01—2018.01.01,the combination is tested with the data of 2018.01.01—2018.07.01,and the transaction results are summarized and analyzed.And test the other years one by one.It turns out that it is effective to use the cointegration theory to construct a paired trading strategy.At the same time,using the GARCH model to optimize the opening and closing conditions can not get better results.