Finding Near-Optimal Portfolios With Quality-Diversity

Bruno Gašperov,Marko Đurasević,Domagoj Jakobovic

2024-02-25

Abstract:The majority of standard approaches to financial portfolio optimization (PO) are based on the mean-variance (MV) framework. Given a risk aversion coefficient, the MV procedure yields a single portfolio that represents the optimal trade-off between risk and return. However, the resulting optimal portfolio is known to be highly sensitive to the input parameters, i.e., the estimates of the return covariance matrix and the mean return vector. It has been shown that a more robust and flexible alternative lies in determining the entire region of near-optimal portfolios. In this paper, we present a novel approach for finding a diverse set of such portfolios based on quality-diversity (QD) optimization. More specifically, we employ the CVT-MAP-Elites algorithm, which is scalable to high-dimensional settings with potentially hundreds of behavioral descriptors and/or assets. The results highlight the promising features of QD as a novel tool in PO.

Portfolio Management,Computational Finance

What problem does this paper attempt to address?

The paper discusses a problem in Portfolio Optimization (PO) in financial investment, which is how to find an approximate optimal investment portfolio. Although the traditional Mean-Variance (MV) framework can achieve an optimal balance between risk and return, it is highly sensitive to input parameter estimation. Therefore, research has shifted towards finding a set of approximately optimal investment portfolios that are more robust.

The paper proposes a new approach based on Quality-Diversity (QD) optimization, using the CVT-MAP-Elites algorithm to find a diverse set of approximately optimal investment portfolios in high-dimensional settings. This approach allows investors to consider personal preferences and market friction factors, such as transaction costs, in decision-making, rather than relying solely on a single optimal portfolio.

The paper reviews previous work, including methods that construct convex hulls to find approximate optimal investment portfolios, which may become impractical or inefficient when dealing with a large number of assets. Then, the paper introduces QD optimization, which is a method that combines convergent and divergent search strategies to generate diverse and well-performing investment portfolios.

Experimental results show that QD optimization performs competitively in both toy models and high-dimensional settings, especially in providing diverse investment portfolios, which provides a new tool for modern portfolio optimization. The paper concludes with a discussion on the robustness of the method and its adaptability to changing estimated parameters.

In summary, the problem addressed in this paper is how to find a set of investment portfolios that are close to optimal in risk-adjusted return while being diverse in the investment weight space or behavioral feature space through quality-diversity optimization.