Empirical Research On Statistical Arbitrage Strategies Based On Threshold Cointegration And State-space Model

With the implementation of China's capital market opening-up strategy,the commodity futures market is expected to meet a historic opportunity for rapid development,which also means that statistical arbitrage strategies based on paired trading have huge investment opportunities in the commodity futures market.Finding statistical arbitrage strategies that can be effectively applied to the commodity futures market has become the focus of many institutional investors.Based on the traditional co-integration theory,this paper first improves the two points in the application of cointegration equations to statistical arbitrage,and proposes a statistical arbitrage model based on threshold cointegration and a statistical arbitrage model based on state space,then discusses the parameter estimation algorithm of the two models.And then the corresponding trading rules are designed based on practical experience,including the opening and closing signal and the stop signal generation method.Finally,a complete arbitrage trading program is obtained,which further expanded the types of the statistical arbitrage strategy.In the empirical analysis section,we take soybean and soybean meal futures as examples to study the effectiveness of the two statistical arbitrage models proposed in this paper for the cross-variety arbitrage of commodity futures.First of all,the stability test and cointegration test are performed on these two trading targets.The empirical results show that there is a cointegration relationship between them,that is,a long-term equilibrium relationship,so statistical arbitrage can be carried out.The data in the sample is used to calculate the parameter estimate of the statistical arbitrage model.Then based on the designed trading rules,the arbitrage transactions within and outside the sample are simulated and tested to obtain the final performance evaluation results.The results show that both of the models can be effectively applied to cross-variety arbitrage of commodity futures,achieving stable and substantial returns,but risk control still needs to be done.Cross-variety arbitrage failures may also cause greater losses.Through comparison,it is also found that the statistical arbitrage based on threshold cointegration model is more suitable for cross-variety arbitrage than the statistical arbitrage based on state space model.After analyzing the paper,it is believed that the latter may play an advantage in the case of a very short mean half cycle.In summary,although the empirical results verify the feasibility of the statistical arbitrage strategy proposed in this paper,there are still some deficiencies in the study of this paper that are worthy of in-depth exploration and improvement.At present,domestic research on stati-stical arbitrage lags behind foreign frontiers.Therefore,more scholars are needed to join this field and develop more effective statistical arbitrage strategies to provide important reference for domestic institutional investors to make decisions.