摘要
本文以双指数复合泊松分布的Lévy过程捕捉价差的跳跃性,以Ornstein-Uhlenbeck(OU)过程刻画价差的均值回复特征,通过均值回复速度和已实现波动率(realized volatility,RV)筛选股票对,依据异质自回归(heterogeneous autoregressive,HAR)模型预测股票对价差的波动率,提出了LOU-RV-HAR配对交易策略.基于沪深300指数成分股的五分钟高频数据的实证分析表明,LOU-RV-HAR策略取得了较好的配对交易业绩,夏普比率为1.6072,远超过同时期沪深300指数的市场表现.对比在价差模型、选股方法和交易阈值等方面调整所建立四种对比交易策略,LOU-RV-HAR策略的年化收益率和夏普比率均优于对比策略.进一步分析表明,LOU-RV-HAR策略在不同市场行情或者不同交易阈值下均表现出稳健的结果,沪深300成分股整体的配对交易业绩高于分行业的业绩.
The paper proposes the LOU-RV-HAR pairs trading strategy by the Lévy process of the double-exponential compound Poisson distribution to capture the jumps of stock pairs spread,the Ornstein-Uhlenbeck(OU)process to characterize the mean-reversion,stock pairs selection with the mean-reversion spread and realized volatility(RV),and the heterogeneous autoregressive(HAR)model to predict the volatility of the spread.The empirical results of the five-minute high-frequency data of the constituent stocks of the CSI 300 index show the LOURV-HAR strategy achieves a better pairs trading performance and the Sharpe ratio is 1.6072,greatly exceeding performance of the CSI 300 index during the same period.Comparing with the four comparative trading strategies established by adjusting the spread model,stock selection method and trading threshold,the annualized return and Sharpe ratio of the LOU-RV-HAR strategy are better than those of the comparative strategies.Further analysis shows that the LOU-RV-HAR strategy displays robust results in different market conditions or different trading thresholds,and the overall pairs trading performance of CSI 300 component stocks is higher than the performance of sub-industries.
作者
赵华
罗攀
王思胤
ZHAO Hua;LUO Pan;WANG Siyin(School of Economics,Xiamen University,Xiamen 361005,China;Head Office,China Guangfa Bank Co.,Ltd,Guangzhou 510080,China)
出处
《系统工程理论与实践》
EI
CSSCI
CSCD
北大核心
2023年第8期2251-2265,共15页
Systems Engineering-Theory & Practice
基金
国家自然科学基金面上项目(71871194)。
关键词
Lévy-OU过程
高频配对交易
HAR模型
Lévy-OU process
high-frequency pairs trading
heterogeneous autoregressive model
