Scenario Specification for Robust Portfolio Analysis

Abstract

Alternative approaches to scenario generation for robust optimal portfolio problems are considered. The underlying strategy is based on worst-case analysis, or min-max. Robustness is ensured by considering the the optimal strategy in view of multiple scenarios generated and evaluating the portfolio corresponding to the best performance, simultaneously with the worst-case scenario. The robust property follows from the fact that the resulting strategy has the best lower bound performance which can only improve if any scenario, other than the worst-case, is realized. Min-max optimization is performed over various single-period scenarios of risk and return, relative to various benchmarks, and incorporating scalable (not fixed) transaction costs.

This work was supported by EPSRC Grant # GR/M41124.