摘要
为了研究股指期货的数据频率对统计套利的绩效影响,本文提出了动态预测区间的择时交易机制。在实证检验中分别运用OLS恒定波动策略,GARCH时变波动策略与之比较。研究结果表明,在样本区间内外,动态预测区间模型各项指标均是最好的。其次,在确保交易过程中发出的信号始终处于有效状态的情况下,随着使用的数据频率趋向高频,可套利次数增多,同时持有期的收益率升高,而单次套利的最大回撤比例也有增加,说明采用高频数据套利将会牺牲一部分收益能力的稳定性来换取绝对收益量的增加。
To study the performance effect of stock index futures' data frequency, the author pro- pose for the select time trading system about dynamic foresting intervals. Compared to the OLS strategy implying constant variance and GARCH strategy implying time-varying variance in empiri- cal test, the result shows that all indicators are optimal when using the model of dynamic foresting interval in the internal and external sample intervals. Otherwise, with data frequency is rising, the arbitrage opportunities increase when the signals is valid in the transaction process. Meanwhile, the yield was promoted but the maximum withdrawal ratio of single arbitrage also boosted, which denotes statistical arbitrage substitute an increase in absolute yield for stability of making a profit. Key words:
出处
《科学决策》
CSSCI
2017年第2期61-75,共15页
Scientific Decision Making
基金
国家自然科学基金面上项目(项目编号:71471117)
关键词
协整模型:GARCH模型
均值回复
中高频
动态预测区间
Cointegration Model
GARCH-model
Mean Reversion
Intermediate or High Frequen- cy
dynamic forecasting intervals