Portfolio selection in a data-rich environment Articles uri icon

authors

  • Bouaddi, Mohammed
  • TAAMOUTI, ABDERRAHIM

publication date

  • December 2013

start page

  • 2943

end page

  • 2962

issue

  • 12

volume

  • 37

International Standard Serial Number (ISSN)

  • 0165-1889

Electronic International Standard Serial Number (EISSN)

  • 1879-1743

abstract

  • We model portfolio weights as a function of latent factors that summarize the information in a large number of economic variables. This approach (hereafter diffusion index approach) offers the opportunity to exploit a much richer information base to improve portfolio selection. We use factor analysis to estimate the space spanned by the factors. This provides consistent estimates for the optimal weights as the number of economic variables and sample size go to infinity. We consider an empirical application to illustrate the practical usefulness of our approach. The results indicate that the diffusion index approach helps to improve the portfolio performance. (C) 2013 Elsevier B.V. All rights reserved.

keywords

  • portfolio's weights modeling; factor analysis; principal components; portfolio performance; stock returns; fama–french factors; economic factors; vix