Portfolio optimization with asset preselection using data envelopment analysis

Abstract This paper uses data envelopment analysis (DEA) approach as a nonparametric efficiency tool to preselect efficient assets in large-scale portfolio problems. Thus, we reduce the dimensionality of problems, considering multiple asset performance criteria linear DEA model. We first introduce several reward/risk that are typically used literature identify features financial returns. Secondly, suggest some input/output sets for preselecting framework. Then, evaluate impact preselected different optimization strategies. In particular, propose an ex-post empirical based on two alternative datasets: components S &P500 and Fama French 100 formed size book market. According this observe better performances preselection than classic PCA factor models large scale selection Moreover, proposed model outperform index strategy fully diversified portfolio.