In this paper, we consider the effects of trading volume, positive and negative jumps and construct HAR-RV-PJ-NJ-TV model to estimate and forecast high-frequency volatility. We use Hu-Shen300 index high-frequency data to estimate model and compare with other high frequency volatility models. We find that the HAR-RV-PJ-NJ-TV model is much better than HAR-RV model both in out-of-sample forecasts and in-sample fitness. Our HAR-RV-PJ-NJ-TV model is a great progress in modeling high-frequency volatility.
机构:
Graduate School of Systems and Information Engineering, University of TsukubaGraduate School of Systems and Information Engineering, University of Tsukuba
Kohda S.
Yoshida K.
论文数: 0引用数: 0
h-index: 0
机构:
Graduate School of Business Science, University of TsukubaGraduate School of Systems and Information Engineering, University of Tsukuba