A rank-based approach in portfolio asset allocation.

We propose a novel rank-based approach (RBA) that can be applied in portfolio allocation tasks. The proposed approach penalizes downside deviations below zero by using the sign, magnitude, and distributional rank of assets. Performance of RBA is compared with the equally weighted, traditional mean-variance and more recently proposed portfolio strategies including improved estimators for covariance misspecification. Overall, applied various performance metrics, in particular, return-per-unit of risk measures provide convincing evidence of enhanced performance in favour of RBA portfolios. [ABSTRACT FROM AUTHOR]