Robust Portfolio Optimization

Sammanfattning

The problem of portfolio optimization involves designing a portfolio that achieves an optimal return-to-risk ratio. The subfield of robust optimization seeks to address the uncertainties in the standard portfolio optimization formulations. Many different types of optimization models exist and they all have their robust formulation. This thesis focuses on the Mean-Variance Markowitz model and the factor model as well as their robust counterparts. It studies the differences and similarities in formulation and in performance of them, based on key results such as efficient frontiers, asset weight distribution and the Sharpe ratio. The underlying data used by the models are the stocks that make up the OMX 30 index as of 2026-01-01. The results of the thesis show that the differences between standard and robust models exist in theory. The advantages of the robust model are not entirely apparent in practice, as the results of the robust models are a linear shift of the regular models. However, the robust models offer the practitioner insight into the uncertainties of the data and how to optimally safeguard for the worst-case scenario. The results also show the outcome of the models onto both downward and upward trending markets.