Files

PHD_THESIS_VANDERVEKEN.pdf
  • Open Access
  • Adobe PDF
  • 3.71 MB
Download

Details

Authors
Supervisors
Vrins, Frédéric
 ;
Lassance, Nathan
Abstract
Portfolio selection lies at the heart of financial decision-making for a wide range of economic agents. Building upon the classical mean-variance framework, this thesis addresses the challenge of optimal portfolio selection under parameter uncertainty, i.e., when the required inputs, e.g., expected returns and covariances, are unknown and must be estimated from historical data. This estimation process inevitably introduces estimation errors which, if not accounted for properly, lead to poor portfolio performance. To address this issue, this thesis proposes several complementary approaches to deal with parameter uncertainty in the portfolio selection context, all grounded in the expected out-of-sample utility (EU) framework. Specifically, we develop methods that improve performance by combining portfolios more effectively, determining the optimal portfolio size before optimization, and designing shrinkage estimators tailored to the portfolio selection problem. Collectively, these approaches mitigate estimation risk, provide novel analytical insights into the portfolio selection problem, and deliver robust out-of-sample portfolio performance.
Affiliations
  • Institution iconUCLouvainSSH/LIDAM/LFIN - Louvain Finance

Citations

  • APA
  • Chicago
  • FWB

Vanderveken, R. (2025). Optimal portfolio selection under parameter uncertainty. https://hdl.handle.net/2078.5/260593