报告地点：腾讯会议 478994734
邀请人：孙海琳教授
摘要：In this paper, we propose a joint input modeling and computation approach in risk measurement and portfolio optimization. We propose to use the Gaussian mixture model (GMM) to learn the randomness of the asset returns, and illustrate the flexibility of GMM with alternative mix modes for input modeling. We show that under GMM, the conditional value-at-risk (CVaR) and the entropic risk measure (ERM) have analytical structures. Based on the structures, we study how to reduce the stochastic problems to deterministic ones for risk measurement and portfolio optimization. We also relate the results to the well known CVaR approximation and Bernstein approximation of chance constrained programs. We illustrate our approach via numerical examples in financial risk management. This is a joint work with Jing Li at Tongji University and Shushang Zhu at Sun Yat-Sen University.
报告人简介：胡照林，同济大学经济与管理*女王调教-女王调教视频-女王 调教小说教授、同济大学青年百人计划入选者。他分别于浙江大学数学系和香港科技大学工业工程及物流管理系获得学士和博士学位。他的研究兴趣包括随机优化，仿真理论和实践，机器学习，金融风险管理，智能决策等。在MS, OR, JOC, IIE, NRL等主流期刊发表论文。目前担任Journal of Management Science and Engineering的Associate Editor。研究获得国家自然科学基金优青、面上、青年基金的资助。