Bayesian portfolio optimization from a static and dynamic perspective

This work presents new approaches to portfolio optimization taking account of both estimation risk an the stylized facts of empirical finance. A general approach to static portfolio optimization is presented, taking account of both estimation risk and the stylized facts of financial data. Furthermore, some desired expert knowledge can easily be included and thus supports the decision process. Nevertheless, the framework is still based on the well-known Markowitz principle and is therefore easy to understand and interpret. Since the static character of the classical Markowitz approach is unrealistic when it comes to practice, the influence of estimation risk on optimal dynamic portfolio strategies is studied. It is shown that the negative effects cannot be compensated by portfolio rebalancings, but are instead even reinforced. A Bayesian approach using subjective prior information is developed, which reduces these negative effects. A simulation study is performed approving the theoretical results and further evaluating the performance of the Bayesian approach.