Lest we forget: learn from out-of-sample forecast errors when optimizing portfolios

Review of Financial Studies (RFS), forthcoming

62 Pages Posted: 29 Apr 2016 Last revised: 27 Apr 2021

See all articles by Pedro Barroso

Pedro Barroso

CATÓLICA-LISBON School of Business & Economics

Konark Saxena

University of New South Wales

Date Written: January 4, 2019

Abstract

Portfolio optimization often struggles in realistic out-of-sample contexts. We de-construct this stylized fact, comparing historical forecasts of portfolio optimization inputs with subsequent out of sample values. We confirm that historical forecasts are imprecise guides of subsequent values but also find the resulting forecast errors are not entirely random. They have predictable patterns and can be partially reduced using their own history. Learning from past forecast errors to calibrate inputs (akin to empirical Bayesian learning) results in portfolio performance that reinforces the case for optimization. Furthermore, the portfolios achieve performance that meets expectations, a desirable yet elusive feature of optimization methods.

Keywords: Portfolio Optimization, Estimation Error, Covariance Matrix, Risk Management

JEL Classification: G11, G12, G17

Suggested Citation

Barroso, Pedro and Saxena, Konark, Lest we forget: learn from out-of-sample forecast errors when optimizing portfolios (January 4, 2019). Review of Financial Studies (RFS), forthcoming, Available at SSRN: https://ssrn.com/abstract=2771664 or http://dx.doi.org/10.2139/ssrn.2771664

Pedro Barroso (Contact Author)

CATÓLICA-LISBON School of Business & Economics ( email )

Palma de Cima
Lisbon, Lisboa 1649-023
Portugal

HOME PAGE: http://https://clsbe.lisboa.ucp.pt/person/pedro-monteiro-e-silva-barroso

Konark Saxena

University of New South Wales ( email )

School of Banking and Finance
Australian School of Business
Sydney, NSW 2052
Australia

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

Paper statistics

Downloads
614
Abstract Views
2,384
rank
54,756
PlumX Metrics