Conditional value at risk optimization using vines - The effect of complex modelling on input error in portfolio optimization

Abstract

In this paper, a complex modelling technique, vine modelling, is compared with the Monte Carlo method to test whether it reduces the sensitivity to errors in inputs for CVaR portfolio optimization. The sensitivity to errors in input is measured by processing the returns of the assets and compare it to the correctly estimated returns when used in CVaR portfolio optimization. The comparison showed that the vine modelling does not decrease sensitivity to errors in input, but the modelling technique does yield a higher expected return with lower risk than the Monte Carlo method.