赋权已实现波动率模型的改进及实证研究

Improvement and empirical research on the weighted realized volatility model

对比分析三类已实现波动率异质自回归模型,基于异质市场理论、赋权已实现波动率以及连续-跳跃成分的分离,构建杠杆效用下带修正跳的隔夜赋权已实现波动率异质自回归模型。通过上证综指5min数据,将带有隔夜波动率、杠杆效用、带修正跳的隔夜赋权已实现波动率异质自回归模型与已有的三类异质自回归模型进行比较。研究结果表明:杠杆效用下带修正跳的隔夜赋权已实现波动率异质自回归模型,由于同时考虑了杠杆效应、分离连续-跳跃成分以及隔夜波动率,因而具有相对较好的样本外预测能力,并且模型的样本内拟合效果也得到显著提升。

Based on the theory of heterogeneous market,the weighted realized volatility and the separation of continuous-jump components,the leverage effect heterogeneous autoregressive model of weighted realized volatility with continuous-jump(LHAR-WAR-CJ-M)is constructed.With data of Shanghai Composite Index available in 5 minutes,the LHAR-WRV-CJ-M model with overnight fluctuation rate is compared with the three types of HAR model.The results show that the LHAR-WRV-CJ-M model with overnight volatility also considers the leverage effect and separates the continuous-jump components has the best extra sample prediction ability.In addition,and the insample fitting effect of the model has also been significantly improved.