Jeremy BELLINA

Composition of the Jury

Abstract

The contemporary financial landscape offers investors a wide range of options for allocating their capital, including commodities, real estate, stocks, and bonds. Among these options, mutual funds and exchange-traded funds (ETFs) have grown in popularity due to their ability to provide cost-effective diversification and professional management. However, the growing complexity of these instruments, combined with increased requirements for transparency and sustainability, highlights the need to develop innovative methodologies to address the challenges associated with portfolio valuation and index tracking.

This dissertation lies at the intersection of these major issues, contributing to both academic research and practical applications. This work focuses on two fundamental problems: the daily estimation of mutual fund assets using publicly available data, and the development of high-performance algorithms capable of addressing the challenges of parsimonious index tracking and portfolio estimation. Drawing on approaches that combine genetic algorithms, least-squares regression, dynamic programming, and Lasso-based methods, this dissertation proposes robust solutions for accurately inferring fund holdings, validated using both simulated and real-world data. In particular, the Lasso-based approach stands out for its superior performance in handling sparse data, offering faster and more accurate estimates than methods based on genetic algorithms. Furthermore, this thesis explores the integration of ESG (Environmental, Social, and Governance) considerations into frugal index tracking. An innovative framework is introduced, combining the minimization of tracking error with the maximization of the ESG score, supported by a time-weighted optimization model.

This approach highlights a key trade-off between sustainability and tracking accuracy, demonstrating that it is possible to achieve significant improvements in ESG scores with minimal impact on tracking error. Through these contributions, this research illustrates the potential of advanced optimization techniques to enhance transparency in the mutual fund industry and integrate sustainability into index tracking. Future research directions include applying these methodologies to various market contexts, incorporating practical constraints such as transaction costs, and collaborating with fund managers to further validate and refine the proposed models.