Robust optimization approaches for portfolio selection: a comparative analysis

Area of Science:

• Finance
• Operations Research
• Econometrics

Background:

• Robust optimization (RO) offers a formal approach to incorporate uncertainty in portfolio selection.
• Existing RO models lack comprehensive empirical performance assessments.
• This study addresses the need for empirical validation of RO in portfolio management.

Purpose of the Study:

• To empirically evaluate and compare the out-of-sample performance of various robust optimization models for portfolio selection.
• To assess the effectiveness of RO models against traditional models under market uncertainty.
• To provide a comprehensive analysis using real-world US market data.

Main Methods:

• Comparative analysis of different RO models, including robust mean-variance, conditional value-at-risk, and Omega ratio models.
• Evaluation using US market data from 2005-2020.
• Focus on out-of-sample performance metrics to assess real-world applicability.

Main Results:

• Robust optimization models demonstrate superior out-of-sample performance compared to their nominal counterparts.
• The empirical analysis provides quantitative evidence supporting the benefits of RO in portfolio selection.
• Different RO models exhibit varying degrees of performance enhancement.

Conclusions:

• Robust optimization models provide a valuable enhancement to traditional portfolio selection frameworks.
• Empirical evidence supports the practical utility of RO in managing portfolio uncertainty.
• Further research can explore advanced RO techniques and diverse market conditions.