报告题目：Statistical inference of spot quantities under infinite variation jumps

报告人：刘志 副教授（澳门大学）

报告时间：2021年12月31日（星期五）9:00-10:00

报告地点：腾讯会议（ID：560 903 156） 

校内联系人：牛一 副教授   联系电话：84708351-8081

报告摘要: Empirical evidence has revealed that the jumps in financial markets appear to be very frequent. We consider the statistical inference of the spot quantities of a two-dimension semi-martingale by using high-frequency data, in a setting where both the co-jumps and the individual jumps in the underlying driving processes are allowed to be of infinite variation. We firstly propose a kernel estimator of the spot variance, and then the idea is extended to the bivariate case. The proposed procedure is based on the empirical characteristic function of the increments of the processes and as well the application of the polarization identity in two-dimensional case. An important finding is that when the jump is frequent enough, the asymptotic normality suffers from a bias. We employ a two-scale approach to remove the bias iteratively. The finite sample performances of the proposed estimators and other existing estimators are assessed and compared by extensive simulation studies. We illustrate the method with some real high-frequency financial datasets. I will also introduce some recent development of the noise-robust estimators. This talk is based some joint works with Qiang LIU and Qiqi LIU.

报告人简介：刘志, 2011年博士毕业于香港科技大学，2011年8月---2012年8月任厦门大学王亚南经济研究院与经济学院双聘助理教授，2012年8月起先后任澳门大学数学系助理教授，副教授。主要研究方向为金融高频数据分析、金融风险管理、随机过程统计推断，生物信息等。在AoS，Jasa，JoE，Jbes，Bioinformatics，ET等相关研究方向的权威期刊发表论文50多篇，主持澳门政府科技基金5项和国家自然科学基金2项。