Academy of Mathematics and Systems Science, CAS
Colloquia & Seminars

Speaker:	黄鑫, 南方科技大学
Inviter:	娄有成
Title:	Discrete-time Adaptive Mean-variance Portfolio Selection with Ambiguous Mean and Covariance
Language:	Chinese
Time & Venue:	2023.09.06 10:00-11:00 思源楼S315
Abstract:	We consider in this research a discrete-time mean-variance portfolio selection problem with both mean and covariance being ambiguous (but constant). At the very start, an investor takes a prior joint density on the unknown mean and covariance. During the periods of investment, the investor could update her knowledge on the ambiguity through Bayesian learning at every time point after she observes the newly realized assets’ prices. Taking the advantage of the setting of a conjugate family, we are able to derive the analytical optimal policy which possesses the feature of active learning. In the language of reinforcement learning or dual control, our algorithm achieves the best balance between the exploitation and exploration in the decision making under uncertainty. We also discuss the computational aspects of our proposed learning policy.