Title: Dynamic Portfolio Optimization using Mean-Semivariance

Advisors: Dr. Hayriye Ayhan and Dr. Shijie Deng

Committee Members:

Dr. Sebastian Pokutta

Dr. Chelsea White

Dr. Jun Xu (School of Computer Science)

Date and time: Thursday, November 2nd, 3:00 PM.

Location: Groseclose 403

Abstract:

This dissertation studies the mean-semivariance portfolio optimization problem. We describe the relationship of this kind of optimization in the context of other types of portfolio optimization. We construct a novel analysis of mean-semivariance in the context of piecewise quadratic optimization. The unique structure of mean-semivariance is leveraged to provide insight into properties of the optimal portfolio as a function of its key input parameters. This characterization allows us to introduce a new approach to solving a multi-period dynamic mean-semivariance portfolio problem. The proposed methodology provides significant improvements over naive approaches not leveraging the unique structure of the mean-semivariance value function. Finally, we develop a novel, distributionally robust piecewise quadratic formulation using semidefinite programming. We apply the robust formulation to the mean-semivariance portfolio problem to construct a distributionally robust mean-semivariance portfolio. We prove that the robust mean-semivariance portfolio is actually equivalent to the classical mean-variance portfolio.