稳健残差控制图的构建及在金融时序中的应用

发布时间：2018-04-13 09:49

本文选题：统计过程控制 + 稳健残差控制图　；参考：《数理统计与管理》2017年05期

【摘要】：对于呈现自相关和波动族聚性并存的受控过程,通常采用残差控制图对其进行监控。但异常点的存在会对自相关或波动族聚性模型的拟合产生重要影响,使得基于该模型的残差并非独立同分布导致常规残差控制图监控失效。为解决这类问题,本文提出稳健残差控制图。即建立稳健的ARMA模型解决自相关问题从而得到无自相关的残差序列,用稳健的GARCH模型来构建控制图的上下限。模拟和实证研究表明,本文提出的稳健残差控制图具有很好的抗异常点能力并能更好的对金融时间序列的异常现象进行监控。
[Abstract]:The residual control chart is usually used to monitor the controlled processes in which autocorrelation and volatility clustering coexist.However, the existence of outliers will have an important effect on the fitting of autocorrelation or volatility clustering model, which makes the residual based on the model not independent and same distribution, which leads to the failure of the conventional residual control chart monitoring.In order to solve this problem, a robust residuals control chart is presented in this paper.The robust ARMA model is established to solve the autocorrelation problem and the non-autocorrelation residuals are obtained. The upper and lower bounds of the control chart are constructed by the robust GARCH model.Simulation and empirical studies show that the robust residual control chart proposed in this paper has a good ability to resist outliers and can better monitor the abnormal phenomena of financial time series.
【作者单位】：广东财经大学统计与数学学院;
【基金】：国家社科基金一般项目(16BTJ035) 广东省自然科学基金项目(2016A030313108)资助
【分类号】：F830;O213.1

【相似文献】