On efficient optimisation of the CVaR and related LP computable risk measures for portfolio selection

The portfolio optimisation problem is modelled as a mean-risk bicriteria optimisation problem where the expected return is maximised and some (scalar) risk measure is minimised. In the original Markowitz model the risk is measured by the variance while several polyhedral risk measures have been introduced leading to Linear Programming (LP) computable portfolio optimisation models in the case of discrete random variables represented by their realisations under specified scenarios. Recently, the second order quantile risk measures have been introduced and become popular in finance and banking. The simplest such measure, now commonly called the Conditional Value at Risk (CVaR) or Tail VaR, represents the mean shortfall at a specified confidence level. The corresponding portfolio optimisation models can be solved with general purpose LP solvers. However, in the case of more advanced simulation models employed for scenario generation one may get several thousands of scenarios. This may lead to the LP model with a huge number of variables and constraints, thus decreasing the computational efficiency of the model. We show that the computational efficiency can be then dramatically improved with an alternative model taking advantages of the LP duality. Moreover, similar reformulation can be applied to more complex quantile risk measures like Gini's mean difference as well as to the mean absolute deviation.