Mean-Variance Analysis in a Multiperiod Setting

Similar to the classical Markowitz approach it is possible to apply a mean-variance criterion to a multiperiod setting to obtain efficient portfolios. To represent the stochastic dynamic characteristics necessary for modelling returns a process of asset returns is discretized with respect to time and space and summarized in a scenario tree. The resulting optimization problem is solved by means of stochastic multistage programming. The optimal solutions show equivalent structural properties as the classical approach, however, by taking rebalancing activities into consideration a different efficient frontier is obtained.