Mauricio Arcos Mora

Julián Benavides Franco

Luis Berggrun Preciado


This article uses four methods to derive optimal portfolios comprising investments in the seven most representative stock exchanges in Latin America from 2001 to 2006 and it studies their composition and stability through time. The first method uses a historical variance - covariance matrix and the second one employs a semi-variance - semi-covariance matrix. The third method consists of an exponentially weighted moving average and the fourth and last method applies resampling. The first three methods suggest similar weights for the tangency portfolio whereas the last method suggests a different composition characterized by more stable, diversified weights. From a practical point of view, this result is significant because less rebalancing can mean greater potential savings. The article further analyzes the performance of optimal portfolios as compared to equally weighted portfolios. Initially the results of applying the Sharpe ratio in the out-of-sample period provided no evidence of statistically significant differences between optimal portfolios and equally weighted portfolios. However, some evidence is provided in favor of resampling as the returns obtained in the out-of-sample period showed stochastic dominance over the returns of the portfolios estimated using more traditional methodologies.



Optimal portfolios, portfolio resampling, stochastic dominance, Latin America

Cómo citar
Arcos Mora, M., Benavides Franco, J., & Berggrun Preciado, L. (2010). Optimal Portfolio Allocation for Latin American Stock Indices. Cuadernos De Administración, 23(40). https://doi.org/10.11144/Javeriana.cao23-40.opaf
Artículos más leídos del mismo autor/a