Unveiling Diversification Strategies: The Most Diversified Portfolio vs. Traditional Optimization: An Empirical Factor Investing Analysis

Jacob Hartmann Henrichsen & Christoffer Felix Kjær

Student thesis: Master thesis


In a field of investment undergoing rapid advancement, it is crucial to exploit diversi-fication opportunities to enhance portfolio performance and mitigate portfolio risks. Traditional mean-variance optimization should provide significant benefits, but numer-ous papers documents the poor out-of-sample results. As a response, alternative meth-ods have emerged. In this thesis, we investigate whether the framework of the most Diversified Portfolio developed by Choueifaty and Coignard (2008), which utilizes the risk-based framework provides a more stable solution. Motivated by the long-term correlation structure of factors with traditional markets and the favorable cross-corre-lation between factors, we examine the most diversified portfolio relative to traditional portfolio optimization strategies. We find the most diversified portfolio appears to deliver a higher out-of-sample Sharpe ratio relative to the conventional portfolio opti-mization methods in the U.S. and developed markets. The most diversified portfolio also seems to better mitigate the tail-risk properties of factor investing. Furthermore, we explore whether engaging in timing the diversification level by dynamically scaling the most diversified portfolio can enhance the risk-adjusted returns. Although, we find there appears to be no value in timing the diversification. Thus, our findings provide the existing literature with supporting evidence on the relative performance of the most diversified portfolio and traditional portfolio optimization methods on factor-based strategies. However, the minor differences and slight inconsistency in terms of risk-adjusted returns across the regions make it difficult to draw a meaningful conclu-sion on whether the most diversified portfolio proves to be a more stable out-of-sample solution.

EducationsMSc in Finance and Investments, (Graduate Programme) Final Thesis
Publication date2023
Number of pages147
SupervisorsMarc Hartung Eskildsen