基于协整-GARCH模型最优阈值统计套利研究

Statistical arbitrage based on GARCH model and the optimal threshold cointegration theory

依据协整和GARCH模型理论,以历史数据作为样本,建立预测未来标准残差序列的模型。利用GARCH模型标准残差序列预警套利信号,搜索标准残差套利的最优阈值和风险测度,建立最优套利方案。检验结果表明,以历史数据建立的最优套利方案对样本外数据进行套利,与传统利用标准正态分布置信水平确定阈值进行套利相比效果更好,其收益和套利成功率均更高,适用于未来短期跨期套利;最优阈值套利方案量化风险值,可有效控制套利风险,适用于低风险投资爱好者。

The prediction model is established to predict future standard residual sequence based on cointegration theory and GARCH model,using historical data as samples. GARCH model standard residual serial is used as warning signal to arbitrage in search for optimal threshold and measure of risk in standard residual serial to establish the optimal arbitrage scheme. The results show that the optimal arbitrage scheme established by historical data is good for arbitraging and better than the traditional probability level of the standard normal distribution to determine the threshold. The earnings and arbitrage success rate are suitable for future short-term arbitrage.The optimal arbitrage scheme could quantify the risk and easy to control the risk of arbitrage,especially for low risk investors.