摘要
随着高频金融数据的获取,已有很多基于高频数据的研究,包括已实现波动率的估计及其分布特征分析等.尝试结合日内高频数据和日收益率数据,基于Copula方法分析了日收益率与"已实现"波动率以及日内价差之间的相依结构.通过分象限对数据进行了Copula拟合,给出了一类特殊数据的联合分布估计方法,进而给出了已实现波动率和日内价差条件下的CVaR的估计方法.最后基于中国股市上证综指和深证成指的高频收益率数据进行了实证分析,并对两种条件下的CVaR方法进行了预测效果的比较,实证结果表明已实现波动率条件下的CVaR预测效果更好.
There are lots of studies based on high-frequency data, including the estimating method and distri- bution character of realized volatility and so on. In this paper, the dependence structure between daily return and realized volatility and that between daily return and price range are analyzed based on Copula method from intraday high-frequency data and daily returns. Joint distribution of one special type of data is estimated based on distributions of different quadrants, and CVaR with consideration of realized volatility and price range is al- so calculated. An empirical analysis of high-frequency data from Shanghai and Shenzhen Stock Markets is presented. By comparing the forecasting results, the conclusion is obtained that the CVaR with consideration of realized volatility forecasts market risks better.
出处
《管理科学学报》
CSSCI
北大核心
2012年第8期60-71,共12页
Journal of Management Sciences in China
基金
国家自然科学青年科学基金资助项目(71001095
70901067)
高等学校博士学科点专项科研基金资助项目(20103402120010)