In this paper we develop a portfolio optimization strategy based on the extraction of option-implied distributions and the application of robust asset allocation. We compute the option-implied probability density functions of the constituents of the Euro Stoxx 50 Index. To obtain the corresponding risk-adjusted densities, we estimate the risk aversion coefficient through a Berkowitz likelihood test. The correlation structure among the stocks is computed via an ad hoc technique, which provides a correction term for the historical correlations. We implement a robust portfolio construction, in order to incorporate the uncertainty about the estimation error for the expected returns in the optimization procedure.