高频数据下投资组合风险预测模型比较  

Comparison and Analysis of the Models of Forecasting Portfolio Risk Based on High-frequency Exchange Data

作　　者：王春峰[1] 张蕊[1] 房振明[1] 李晔[1] 

机构地区：[1]天津大学金融工程研究中心,天津300072

出　　处：《系统工程》2007年第3期23-28,共6页Systems Engineering

基　　金：国家杰出青年科学基金资助项目(70225002);教育部优秀青年教师教学科研奖励基金资助项目

摘　　要：“已实现”协方差矩阵是对投资组合波动性及相关性的一种全新的度量方法。系统介绍基于高频交易数据的“已实现”波动率及由它拓展而来的“已实现”协方差矩阵。利用样本数据对模型进行检验,并比较分析该方法与DCC-GARCH方法的优劣。对比结果说明,这种基于高频交易数据的多元RV估计方法在估计精度和计算简便程度上明显优于DCC-GARCH方法。Realized covariance matrix is a new measurement of the volatility and correlation of portfolio. Based on the highfrequency data, the realized volatility and realized covariance matrix was introduced systematically. With sample data, the multivariate RV model was verified, and also compared with the DCC-GARCH model. The result indicated that the RV model has the better performances on the estimation accuracy and computation simplicity. mltivariate

关 键 词：多元波动性 "已实现"波动率 "已实现"协方差 DCC-GARCH 

分 类 号：F830[经济管理—金融学]