中文摘要：

以单一阈值日内跳跃识别方法为理论出发,首次提出通过分析日内高频收益波动的模式,设计具有更强适应性的可变阈值日内跳跃识别方法,并据此区分日跳跃波动与连续波动,用于已实现波动HAR-RV-CJ模型的估计。使用塑料期货1分钟高频价格数据的实证表明,结合日内收益波动L型折线模式的可变阈值日内跳跃识别方法,在应用于HAR-RV-CJ模型的标准差形式与对数形式时,可以显著提高对短期、中期波动的拟合及预测能力。

英文摘要：

With the constant threshold intraday jump identification method as a starting point,we propose a more adaptive variable threshold intraday jump identification method based on the analysis of returns＇ intraday patterns,which can then be used to distinguish daily jump volatility from continuous-time volatility for the estimation of HAR-RV-CJ models.Empirical results using plastic futures＇ 1-minute high-frequency prices indicate that,our proposed variable threshold intraday jump identification method can significantly improve the fit and forecast performance of short-term as well as middle-term volatilities when applied in the HAR-RV-CJ model＇s standard deviation form and logarithm form.