报告人: 朱丹 (莫纳什大学)

日  期: 2019年7月10日

时  间: 10:00

地  点: 理科楼 LD202

摘  要: This paper provides three generic improvements to least square Monte Carlo approaches for solving optimal stopping problems. First, we modify the regression specification to target the stopping boundary directly. Second, we employ Bayesian model averaging to select the set of basis functions used in the least square regression. Third, we formulate a single index regression to reduce the dimensionality without substantial compromise in results. Without loss of generality, we demonstrate the implementation of each improvement on a different early exercisable product and compare the results with established methods. We emphasise that the methods introduced are not only generic but are easily transplanted to any scenario where least square Monte Carlo is viable without any signi_cant increase in computational cost.

报告人简介：朱丹，莫纳什大学高级讲师，主要研究金融数学，随机模拟等问题,在JoE，EJOR等杂志上发表过多篇论文。

公司联系人:张志民

欢迎广大师生积极参与！