多分形波动率测度的VaR计算模型

VaR model based on multifractal volatility measurement

以上证综指长达6年时间的5分钟高频数据为实证样本,首先提出了一种基于多分形谱(Multifractal spectrum)分析的市场波动率测度方法(Volatility measurement),并进一步探讨了其在市场风险价值(VaR)计算中的模型设计和应用.实证结果表明:我国新兴资本市场的价格波动确实具有显著的多分形特性,且与各类线性和非线性GARCH族模型相比,在高风险水平上,基于多分形波动率测度的VaR模型具有更高的风险测度精度.

Recently research on econophysics discovers that, no matter in developed capital markets or in emerging markets, the price volatility presents chaos, fractal and more complicated multifractal nonlinear characteristics. Based on 5-min high-frequency data of SSEC during 6 years, this paper proposes a volatility measurement based on multifractal spectrum analysis. To verify its validity, we discuss its models in VaR calculation. The empirical results show that price volatility in Chinese stock market does present multifractal feature and on high-risk levels, VaR model based on multifractal volatility performs better than many linear and nonlinear GARCH models.