准确刻画和预测股指期货市场的波动率, 有利于实现股指期货的价格发现、套期保值和市场调节等功能.文章首先在已实现极差异质性自回归模型(the Heterogeneous Auto-Regressive with Realized Range, HAR-RRV model)的基础上, 构建HAR-HLT和LHAR-HLT模型; 接着, 以中国股市中沪深300股指期货的5分钟高频交易数据为样本, 通过运用HAR-HLT和LHAR-HLT模型分析股指期货市场波动率的特征以及预测股指期货市场未来的波动率.研究发现:在HAR-HLT和LHAR-HLT模型中, 高频已实现极差、低频已实现极差和趋势已实现极差都包含较多对未来波动率的预测信息; 股指期货市场的波动率存在短期的``动量效应''和中长期的``反转效应'', 其杠杆效应并不明显; HAR-HLT和LHAR-HLT模型对未来1 日、1周和1月波动率样本外预测能力都明显强于目前常用的HAR-RRV类模型.

It is beneficial to achieve the functions of price discovery, hedging and market regulation, if the volatility of future market could be described and predicted accurately. Firstly, this paper constructed HAR-HLT and LHAR-HLT model based on the the Heterogeneous Auto-Regressive with Realized Range model; then, Using CSI 300 Index's five minutes high-frequency trading data as a sample, we analyze the characteristic of stock index futures market volatility and predict stock index futures market's future volatility with HAR-HLT and LHAR-HLT model. The result of this paper demonstrates that all the realized ranges of high-frequency、low- frequency and trend contain more information to forecast future volatility in the HAR-HLT and LHAR-HLT model. In addition, there are short-term ``momentum effect" and long-term ``reverse effect" in the stock index futures market volatility, but the leverage effect is not obvious. In terms of predicting volatility sample in one day, one week or one month, we found that HAR-HLT and LHAR-HLT model are significantly stronger than the current used HAR-RRV model.