摘要
极值理论可以更精确地处理金融数据的厚尾特性。选取香港上市的汇丰控股数据,运用POT方法做实证分析,对时间序列取门限值(Threshold),对超过门限的样本数据建模,极限点渐进分布服从GPD分布,估计模型参数,检验模型的合理性,并给出置信度为95%下的VaR。
Extreme value theory can describe the heavy tail characteristics of finance data more exactly.HSBC Holdings in Hong Kong as an example are took,gaven the data analysis by the method of POT as follows: chose the threshold of the data series,modelinged the sample data excess the threshold whose limit converges to the distribution of GPD,estimated parameters,test the reasonability and gave the VAR under 95% confidence level.
出处
《科学技术与工程》
2011年第16期3858-3860,共3页
Science Technology and Engineering
关键词
GPD分布
阈值模型
POT模型
门限
泊松过程
GPD distribution peaks ouer threshold POT model threshold poisson process