Medida de performance de carteira por média-variância e a medida Ômega: uma análise empírica dos modelos CAPM e OCAPM para o Ibovespa e Dow Jones

Abstract

The objective of this work is to investigate the efficiency of the traditional model of performance analysis and portfolio allocation based on the mean and variance structure of Sharpe and Treynor, known as Capital Asset Pricing Model (CAPM), as opposed to the model Omega Capital Asset Pricing Model (OCAPM), a model based on the Omega measure with less restrictive assumptions, in which utility function and return distributions of the specific assets are not required. In addition, all moments of return distributions are considered indirectly, that is, there is no need to be calculated and observed by investors. During the study the optimization of the series of stock returns that compose the Dow Jones and Ibovespa indexes was performed, following the methodology of each model to find the market portfolio. From this, and following the objective function of the CAPM and OCAPM, it was verified which model best explains the cross-sectional returns of the shares of the two indices for the period from 1999 to 2016. For that, the GRS tests were used (in-sample) And Diebold-Mariano (out-sample). The result of the first test showed the non-rejection of the efficiency in mean-variance and betas with high explanatory grade for the returns for all the portfolios, being the portfolio formed by the CAPM model with better results for the Dow Jones index, while the one formed by the OCAPM obtained better results for the Ibovespa. In the second test, it was verified that the OCAPM showed greater predictive power for any time horizon tested.