A Comparison of Pre and Post Modern Portfolio Theory Using Resampling

This article introduces the Resampling approach to Portfolio modeling, targeted at reducing the effect of estimation error present in any practical implementation of a Portfolio Model. Resampling is a method used in portfolio modeling to try to obtain better out of sample performance for given input model parameters. In the real world, where the possibility of estimating errors for future model forecasts certainly exist, it is necessary to consider the error component in building portfolios. Resampling does this by recombining the input parameters required for a portfolio model. In this paper an application of Resampling is performed using a sample of equities from different stock markets. The results are presented for Tracking Error Minimization, Mean Absolute Deviation Minimization (MADM) and Shortfall Probability Minimization Models. The innovation in this study lies in the comparison made with different portfolio models. Unlike previous studies, the evidence shows that Resampling applied to the Markowitz model does not generate better out of sample performance. However, the benefits of Resampling applied to the Post Modern Theory model are remarkable