Weight Bound Constraints in Mean-Variance Models: A Re-examination Based on Machine Learning

37 Pages Posted: 9 Feb 2022

See all articles by Gilles Boevi Koumou

Gilles Boevi Koumou

Université Mohammed VI Polytechnique

Date Written: October 15, 2021

Abstract

Using a novel form of the weight zero or strictly negative lower and positive upper bound constraints mean-variance model in terms of the support vector data description --- a machine learning algorithm --- we bolster the theoretical foundation of weight bound constraints by offering a new interpretation in terms of robust control theory. Additionally, we enhance the understanding of the source of their performance by providing new insights on their mechanism in terms of a tool of asset classification errors reduction. Finally, from these new insights, we demonstrate --- by simulation and empirical analysis --- how to improve the performance of the naive portfolio.

Keywords: Mean-Variance Models, Weights Constraints, Estimation Error, Machine Learning, One-Class Classification, Support Vector Data Description, Robust Control Theory.

JEL Classification: G11, D81

Suggested Citation

Koumou, Gilles, Weight Bound Constraints in Mean-Variance Models: A Re-examination Based on Machine Learning (October 15, 2021). Available at SSRN: https://ssrn.com/abstract=4027548 or http://dx.doi.org/10.2139/ssrn.4027548

Gilles Koumou (Contact Author)

Université Mohammed VI Polytechnique ( email )

Rabat
Morocco

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
26
Abstract Views
102
PlumX Metrics