基于符号收益和跳跃变差的高频波动率模型

Forecasting the realized volatility based on the signed return and signed jump variation

基于符号收益率的视角,对现有的HAR-RV类及其跳跃扩展模型进行相应分解,构建新型的HAR-RV类波动率模型.进一步,结合符号收益和不同的跳跃识别检验方法,提出了包含符号跳跃变差的HAR-RV类模型,并利用样本外滚动窗预测技术和"模型信度设定"(MCS)检验法评价了各种新旧HAR-RV模型对我国沪深300股价指数波动的预测能力.结果表明:基于C_TZ跳跃识别检验的符号跳跃变差能显著改善波动率模型的短期预测能力,但在中长期波动预测时,符号跳跃变差未能明显提升HAR-RV类模型的预测精度;新提出的HAR-S-RVTJ-TSJV模型和HAR-S-RV-TJ模型分别在对短期(未来1天)和中长期(未来5天和20天)的波动预测检验中,展现出了最高的预测精度.

This paper decomposes the HAR-RV model and its various extensions based on the signed return and signed jump variation. Furthermore, new HAR-type models including the signed jump variation are pro- posed, which are constructed by the different jump tests. The forecasting accuracies of the existing and new HAR-type models are evaluated according to the rolling window method and MCS test. The empirical results show that the new signed jump variation including the C_TZ test is a significant improvement in the model＇ s short horizon forecasting performance. However, in forecasting the medium and long horizons, the signed jump variation shows no substantial improvement in forecasting accuracies. Finally, our newly proposed models, HAR-S-RV-TJ-TSJV and HAR-S-RV-TJ, are the best forecasting models in short and medium and long hori- zons than others models discussed in this paper.