Enhancing Markowitz's Portfolio Selection Paradigm with Machine Learning

ADIA Lab Research Paper Series

Authors: Marcos López de Prado, Joseph Simonian, Francesco A. Fabozzi & Frank J. Fabozzi

Date Published: October 2024

This paper examines how machine learning (ML) techniques can be incorporated into Markowitz's portfolio selection framework, demonstrating their value in strengthening the robust mathematical strategies required by today's financial markets. By pairing traditional econometric approaches with modern ML methods, we illustrate improvements to key portfolio management functions, including alpha generation, risk management, and the optimization of risk measures such as conditional value at risk. Because ML can process large, complex datasets, it enables more adaptive and data-driven decision-making in constructing portfolios. We further explore practical, real-world applications of these methods in portfolio management, addressing both the benefits they offer and the obstacles portfolio managers encounter when putting ML-based strategies into practice.