The aim of this diploma thesis is to solve the problem of handling the parameters in portfolio optimization. In contrast to the Markowitz model one only assumes that the expected return is within an uncertainty set. As elliptically distributed returns are considered, at first elliptical distributions and their properties are studied. Afterwards unbiasedness, asymptotic normality and consistency are shown for the estimators which are necessary for the portfolio optimization. Subsequently the Markowitz model and the robust model are introduced. After that the classical portfolio optimization is compared numerically to the robust one with historical data. It turns out that the robust portfolio, although having a better out-of-sample performance, behaves more riskavers. Another advantage of the robust method is the smaller maximum drawdown and the smaller turnover. At last the effects of different elliptical uncertainty sets in the robust portfolio optimization are investigated.     «

The aim of this diploma thesis is to solve the problem of handling the parameters in portfolio optimization. In contrast to the Markowitz model one only assumes that the expected return is within an uncertainty set. As elliptically distributed returns are considered, at first elliptical distributions and their properties are studied. Afterwards unbiasedness, asymptotic normality and consistency are shown for the estimators which are necessary for the portfolio optimization. Subsequently the Mark...     »