Backard/Forward Optimal Combination of Performance Measures for Equity Screening

Abstract

We introduce a novel criterion for performance measure combination designed to be used as an equity screening algorithm. The proposed approach follows the general idea of linearly combining existing performance measures with positive weights and the combination weights are determined by means of an optimization problem. The underlying criterion function takes into account the risk-return trade-off potentially associated with the equity screens, evaluated on a historical and rolling basis. By construction, performance combination weights can vary over time, allowing for changes in preferences across performance measures. An empirical example shows the benefits or our approach compared to naive screening rules based on the Sharpe ratio.

Keywords: performance measures, combining performance measures, portfolio allocation, equity screening, differential evolution

JEL Classification: C44, C58, C61, G11, G17