Portfolio Optimization Using Factor Scores as Constraints—Factor Constrained Portfolio Optimization Approach

Abstract

In this article, the authors construct portfolios by integrating quantitative factors in the portfolio optimization problem using their normalized score values in a constraint. The resulting optimized portfolios are tilted by the factor exposures controlled by the scoring constraint. The authors test the performance of the mean–variance optimization problem using scores of three value factors (book-to-price, earnings-to-price, cash flow-to-enterprise value) and the momentum factor. Optimized portfolios constrained with the book-to-price and momentum factor deliver the best results in terms of mean and risk-adjusted return. The combination of book-to-price and momentum factor portfolios further improves the risk-adjusted return. The results of this method are comparable with those of the portfolio optimization based on the estimation of the standard stock selection models.

TOPICS Analysis of individual factors/risk premia, factor-based models, portfolio construction

Key Findings

• ▪ The authors assess the usefulness of a factor score as a proxy for the stock expected return in the construction of factor-tilted portfolios.

• ▪ The approach is tested on well-known cross-sectional factor strategies based on the momentum factor and three value factors.

• ▪ The authors consider a two-step approach in which individual value and momentum factor-tilted optimized portfolios are combined.