主 题:Bootstrap and Large Deviation for Realized Laplace Transform of Volatility
内容简介:We develop and implement bootstrap methods for realized Laplace transform of volatility-based statistics. We show that a naive wild bootstrap method fails to work for realized Laplace transform of volatility. Next, we consider a modified wild bootstrap and the local Gaussian bootstrap methods and prove their first-order asymptotic validity. Motivated by the good performance of the local Gaussian bootstrap method in finite samples, we use Edgeworth expansions to compare its accuracy with the existing first-order feasible asymptotic theory. Our cumulants expansions show that the local Gaussian bootstrap able to mimic the higher-order bias of the studentized statistic and for which second-order asymptotic are obtained. Our Monte Carlo simulations studies show that the local Gaussian bootstrap outperforms the finite sample properties of the modified wild bootstrap and the existing first-order asymptotic theory. Finally, we will discuss some large deviation principles for the realized Laplace transform of volatility. This is joint works with Ulrich Hounyo and Rasmus T. Varneskov, and with Lidan He and Xinwei Feng.
报告人:刘志 副教授
时 间:2018-12-26 10:00
地 点:竟慧东楼302
举办单位:统计与数学学院 统计科学与大数据研究院 科研部
