高频时间序列的改进"已实现"波动特性与建模

高频时间序列的分析与建模是金融计量学的一个全新的研究领域，“已实现”波动是针对高频金融时间序列的一种全新的波动率度量方法．证明了“已实现”波动的极限性质，对“已实现”波动的计算方法进行了修正，得到了更有效的改进“已实现”波动．在综合考虑微观结构误差和测量误差的基础上，定义了最优抽样频率．通过蒙特卡罗模拟，说明改进“已实现”波动是比“已实现”波动更有效的估计方法．研究了上海股市的“已实现”波动和改进“已实现”波动的特性，并针对改进“已实现”波动的长记忆性和“杠杆”效应建立了ARFIMAX模型．

Characteristics and modeling of adjusted realized volatility for high frequency time series

High-frequency time series analysis and modeling are a new research field in financial econometrics, and realized volatility is a new measure approach of volatility in high_frequency data field. The realized volatility's limit characteristics is proved, and a more efficient approach is put forward, which is the adjusted realized volatility. Optimal frequency based on microstructure error and measure error is defined. Through Monte Carlo simulation, the paper shows the adjusted realized volatility is more efficient approach than the realized volatility. The characteristics of the realized volatility and adjusted realized volatility of Shanghai Stock Market are studied, and a ARFIMAX model for long memory of adjusted realiged volatility and leavery effect is building.