Portfolio construction using short-run & long-run measures of association.

Abstract

In this paper, we present an analysis of the effectiveness of various portfolio optimization strategies applied to the stocks included in the S&P 1500 index, for a period of 10 years, from 2004 until 2014. Employing short-run and non-parametric long run measures of association such as stock beta relative to the index we construct the global minimum variance portfolio which is one of the minimum variance of all feasible portfolios in the efficient set. The non-parametric long-run analysis has been followed to construct the similar global minimum variance portfolios and provided that it leads to a superior alternative. In a long-run impact we construct portfolios based on the cointegration relationship with the index distributing stocks in each portfolio based on how strongly cointegration evidences there are. The study was extended taking into consideration the Lehmann Brothers Collapse in United States on September 2008. In the last section, we proceed a sector analysis of each portfolio of each strategy before and during crisis and supply us with useful information about how stock sectors are spread before and during a shock.