我国股票市场高频波动预测研究
2017-12-02刘光强
刘光强



关键词:金融资产;证券市场;资产定价;风险管理;收益率;波动率;股票市场
摘要:利用日内高频交易数据对金融资产收益率的波动率进行建模分析是近年来理论界和实务界共同关注的热点问题。以2005年初到2016年底我国日间HS300指数5分钟高频数据为样本,实证分析以渐进理論为基础的ARQ及HARQ模型对我国股票市场高频波动的预测效果,并与跳跃、跳跃变差及正负向跳跃变差为基础的HARRV、HARJC、CHAR及SHAR等多种高频波动率预测模型进行比较,发现:ARQ及HARQ模型对我国股票市场具有更高的预测精度。
中图分类号:F830.91
文献标志码:A文章编号:1009-4474(2017)04-0082-06
The Study of High Frequency Fluctuations Forecasting for Chinese Stock Market:
The Empirical Research with ARQ and HARQ Modes
LIU Guangqiang
(School of Economics and Management, Southwest Jiaotong University, Chengdu 610031, China)
Key words: financial assets; securities market; asset pricing; risk management; rate of return; fluctuation rate; stock market
Abstract: It is a hot topic in academics and industry that modeling the volatility of the yield rate of financial asset with the intraday high frequency trading data in recent years. This paper studies the realized volatility of HS300 index from January 8, 2005 to December 2016 with the ARQ and HARQ models which exploit the asymptotic theory for highfrequency realized volatility estimation to improve the accuracy of the forecasts. And different test results show that ARQ and HARQ model have higher prediction accuracy while compared the forecast results of HARQ with HARRV, HAR JC, CHAR and SHAR models which rely on a decomposition of the total variation into a continuous and a discontinuous (jump) part.
一、引言
对波动率的建模分析在资产定价、风险管理等方面具有重要的现实意义,利用日内高频交易数据对金融资产收益率的波动率建模分析是近年来理论界和实务界讨论的热点问题,其中具有代表性的是基于已实现波动率(Realized Volatility,RV)模型〔1〕,通过考虑金融市场日内交易的具体信息对其收益率建模一方面克服了传统GARCH模型、SV模型的不足,另一方面计算简便、无复杂模型,而且具有无偏性和较好的稳健性等优点。目前,越来越多的实证结果表明,RV测度方法对金融资产波动率刻画较传统的GARCH方法、SV方法具有更加优异的预测效果。
Andersen等提出运用分整模型(ARFIMA)刻画日内波动,Corsi在此基础上提出的近似长记忆性的HAR(Heterogeneous AR)模型弥补了ARFIMA模型计算的复杂性,而且有助于刻画金融资产的典型特征〔2〕。在此基础上,考虑了跳跃等特征的波动率模型得以逐渐完善和发展。……
