Loading...Loading...

Robust portfolio optimization: past performance versus future performance ; Do portfolio combinations improve the performance compared to individual portfolio strategies?

Martin, Jasmine

(2024)

Details

Supervisors

Lassance, Nathan

Faculty

Louvain School of Management

Degree label

Master [120] : ingénieur de gestion, à finalité spécialisée

Abstract

The selection of optimal portfolio strategies is crucial in active asset management, particularly when balancing the trade-off between risk and return. This thesis investigates whether combining different portfolio optimization strategies can enhance performance. Both the sample covariance matrix and linear shrinkage covariance matrix were employed under a rolling window of 120 months and 180 months to compare various individual portfolio strategies (mean-variance, equally weighted, minimum-variance) and their combinations. This analysis spans an out-of-sample period from January 1970 to December 2023. In theory, portfolio combinations maximizing the expected out-of-sample utility should outperform individual strategies. The findings confirm this theory when applying the Monte Carlo simulations approach. However, when applying the rolling-window method on real-world data, the individual global minimum-variance and equally weighted portfolio strategies outperformed all portfolio combinations when the sample covariance matrix is computed. It is only when a large rolling window and the linear shrinkage covariance matrix are used that portfolio combinations outperform for all levels of risk aversion. This comprehensive evaluation highlights the importance of strategy selection based on data sample length and shrinkage methods, providing valuable insights for constructing diversified and robust portfolios.