Abstract

GOMEZ R, Juan Manuel  and  JIMENEZ M., José Alfredo. Optimal portfolio selection based on first and second order Markov chains. Lect. Econ. [online]. 2020, n.92, pp.33-66. ISSN 0120-2596.  https://doi.org/10.17533/udea.le.n92a02.

Searching for create investment strategies in pursuit of maximizing the expected return on investment and minimizing the risk two models of selection of optimal portfolios are studied. The first portfolio composition model is adjusted using logarithmic returns, and the other uses principal component analysis (PCA) at these returns. Then, for each of them its weighted performance is established and measures are created to establish the states of the first and second order Markov chains, this allows to predict whether the shaped portfolios will have bullish or bearish behaviors given the probabilities of the states of the Markov chains. An application is made using the daily closing price returns of 21 COLCAP shares for the period from January 2014 to October 2017. Concluding that in the Colombian Market a portfolio formed by PCA of the returns has a higher expected profitability and less risk in the long term, having an accuracy of model’s forecast according with the stationary vectors of the Markov chains.

Keywords : portfolio selection; Markov chain; principal component analysis; risk aversion; stock index..

        · abstract in Spanish | French     · text in Spanish     · Spanish ( pdf )