Description:
We study the empirical performance of alternative risk and reward specifications in portfolio selection. In particular, we look at models that take into account asymmetry of returns, and treat losses and gains differently. In tests on a dataset of German equities we find that portfolios constructed with the help of such models generally outperform the market index and in many cases also the risk-based benchmark (minimum variance). In part, higher returns can be explained by exposure to factors such as momentum and value. Nevertheless, a substantial part of the performance cannot be explained by standard asset-pricing models.