引入隔夜信息的已实现波动率

发布时间：2018-02-26 20:26

本文关键词： 隔夜信息已实现波动率 GARCH类模型模型置信集检验　出处：《系统工程》2017年04期 　论文类型：期刊论文

【摘要】：中国股市每天交易时间只有4小时,而可能影响股票收益的信息却在24小时内连续累积。隔夜信息即指非交易时段产生的影响股市的信息,既可能源自非交易时段的政策颁布、公司公告,也可能源自海外市场的股价变动等。在估计日已实现波动率时,对隔夜信息的处理方式尚无定论。以沪深300指数5分钟高频价格为实证数据,比较多种引入隔夜信息的方法,发现本文提出的以SP500指数交易时段已实现波动率代理隔夜波动率,与沪深300指数交易时段已实现波动率相加而构建的日已实现波动率估计量,在均方误差指标和模型置信集检验下具有最高的样本内精确度。并且以之为真实波动率代理变量时,常用GARCH类模型样本外预测性能的比较结果在各种估计窗/预测窗组合下最为一致。
[Abstract]:Every time the stock market trading China only 4 hours, and may affect the stock returns of the information is continuously accumulated in 24 hours. The overnight information refers to the stock market effect of non trading generated information can be from non trading period energy policy issued, the company announcement, may stem from overseas market price changes in. Estimated daily realized volatility, handling of overnight information is inconclusive. In Shanghai and Shenzhen 300 index 5 minutes high price of empirical data, the comparison of many methods into the overnight information, found that using SP500 index trading segment realized volatility proxy overnight volatility, and the Shanghai and Shenzhen 300 index trading has been achieved the volatility of the sum on the realized volatility estimator in the mean squared error and confidence index model has the highest accuracy in sample test set. Based on the real volatility generation The comparison results of the external prediction performance of the commonly used GARCH model are the most consistent in the combination of the various estimation windows / prediction windows.

【作者单位】：南京大学工程管理学院;
【基金】：国家自然科学基金资助项目(71201075;71671084) 高等学校博士学科点专项科研基金资助项目(20120091120003)
【分类号】：F832.51

【相似文献】