基于超高频数据的风险价值测度  

Measurement of Value at Risk Based on Ultra High Frequency Data

作　　者：陈勇[1] 姜梦诗 CHEN Yong;JIANG Meng-shi(School of Finance and Statistic,Hunan University,Changsha 410006,China)

机构地区：[1]湖南大学金融与统计学院,湖南长沙410006

出　　处：《数学的实践与认识》2020年第22期21-30,共10页Mathematics in Practice and Theory

摘　　要：投资者主要应用风险价值模型度量资产组合的市场风险.本文采集了股票市场的逐笔交易数据,应用傅里叶方法估计股票价格的波动率,提出了估计股票价格波动率的GARCH-UHFV模型,并与GARCH模型和GARCH-RV模型对比研究.结果表明,与GARCH模型相比,GARCH-RV模型能够提高风险价值的预测准确性,但容易高估风险,而应用傅里叶方法根据逐笔交易数据计算股票价格的日波动率,再结合基于日间数据的GARCH模型可以提高风险价值的预测精确度.Investors mainly adopt Value-at-risk model to measure market risk.In this paper,we forecast volatility and Value-at-risk based on ultra high frequency data using modified GARCH model.The modified model combines the ultra high frequency volatility estimated by Fourier method with GARCH model,named GARCH-UHFV model.Compared with GARCH model and GARCH-RV model,the results show that to some extent GARCHRV model can improve VaR prediction,but it tends to overestimate market risk.Ultra high frequency volatility which is calculated by means of Fourier method coupled with the GARCH model based on the daily data can improve VaR prediction.

关 键 词：超高频数据 傅里叶 风险价值 GARCH-UHFV模型 

分 类 号：O211.67[理学—概率论与数理统计] F832.51[理学—数学]