关键词: 线性模型, Ｈｕｂｅｒ损失函数, 厚尾误差, 序列相关性检验

Abstract:  Heavy tails or asymmetric characteristics data are often found in financial data analysis, which brings many troubles to model correlation testing. In this paper, we address the problem of serial correlation testing for heavy tails and asymmetric distributions by combining the rank correlation coefficient and Huber loss, proposing the HubT and CCT test statistics, and obtaining the asymptotic distribution of the test statistics under the null hypothesis. The numerical simulation shows that the newly constructed statistic can have good performance in heavy tails and asymmetric distributions.