基于我国期货市场的统计套利研究

Empirical Study of Statistical Arbitrage in Chinese Future Market

不同于股票市场,期货市场存在天然的做空机制,非常方便进行套利。本文以我国的期货市场为研究对象,对期货市场上的统计套利进行实证研究.首先,我们提出了协整模型、误差修正模型和基于协整关系的GARCH等3个统计套利模型,设计了相应的交易策略,然后,对样本外数据进行检验,分析不同开仓位、止盈、止损位置与累计收益率的关系。结果表明,基于协整关系的GARCH模型是三个统计套利模型中最优的.实证结果表明,本文所提出的统计套利模型以及相应的策略在我国的期货市场是可行的。

Different from stock market,there exists natural shorting mechanism in future market.So it is very convenient to conduct arbitrage.This paper focuses on statistical arbitrage in the future market.Firstly,we introduce risk-less arbitrage and propose an operable set of strategies in arbitrage trading.We study cotton future contracts in Zhengzhou commodity exchange,and use historical data from 15 portfolios in 6 contract months to do empirical analysis.As about statistical arbitrage,we propose three models,that is,cointegration model,Error Correction model and the new GARCH model based on the conintegration relationship.Based on those three models,we design corresponding transaction strategies and conduct tests on the out-of-sample data,analyzing the relationships between accumulative returns and different open positions,surplus positions and stop-loss positions.The empirical results show that,the ARCH model based on the conintegration relationship is the best model in our three statistical arbitrage models.Hence,we can get a conclusion that the models and strategies in statistical arbitrage are feasible in our future market.