非参数可加协整模型下的配对交易  

Pairs trading strategies underlain by nonparametric additive cointegrations

作　　者：董朝华 赵哲伟 DONG Chaohua;ZHAO Zhewei(School of Statistics and Mathematics,Zhongnan University of Economics and Law,Wuhan 430073,China)

机构地区：[1]中南财经政法大学统计与数学学院,武汉430073

出　　处：《系统工程理论与实践》2022年第6期1694-1720,共27页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金面上项目(71671143,72073143);中南财经政法大学中央高校基本科研业务费专项资金(2722022EG001)。

摘　　要：金融时间序列往往是非平稳的,在使用配对交易策略时协整方程更能描述其本质特性.本文对比使用了参数线性协整模型和非参数可加协整模型,对2015-2019年沪深300成份股的配对交易表现进行研究.以一年时间作为配对期,在此期间内选择配对交易的股票和模型拟合;以之后每半年的时间作为一个测试期,用于配对交易并进行收益核算.结果显示在3‰的佣金水平下,非参数可加协整模型下的配对交易可以取得正向收益,但传统线性参数协整方法收益为负.当降低拥金水平,非参数方法收益迅速增加,参数协整方法收益则增长缓慢;在不同交易触发阈值下,非参数方法的交易频率,累积交易时间都高于参数协整方法;交易比例对参数协整模型的影响大于非参数模型.使用更精细的模型,通过技术分析手段能够获得超额收益,说明中国股市尚且不是一个弱有效市场.Financial time series are often non-stationary,and hence cointegration equations can better describe their essential characteristics that facilitate the use of pairs trading strategy.This paper studies extensively on the performance of the pairs trading strategy that is underpinned by either linear parametric cointegration or nonparametric additive cointegration model,using the data of CSI 300 constituent stocks from 2015 to 2019.We use the first year of the specified period as the pairing period,in which the stocks are selected by certain criterion for pairing;the next each half year is used as executive period for pairing trading and then the income of the trading is calculated.As a result,at the commission level of 3‰,the additive nonparametric method underlying the pairs trading yields positive returns,while the traditional linear cointegration model gets negative returns.As trading cost falls,the returns of the nonparametric method increases rapidly,whereas that of the cointegration method increases slowly.Under different trading thresholds,the transaction frequency and cumulative transaction time for nonparametric method are relatively higher than those of parametric linear cointegration method.In addition,the impact of transaction proportion on cointegration model is stronger than that on nonparametric model.Using a more accurate model,we can obtain excess returns through technical analysis,which shows that China ’s stock market is not a weak efficient market at present.

关 键 词：非参数模型 协整 配对交易 统计套利 量化投资 

分 类 号：F224[经济管理—国民经济]