Insurance companies and other institutional investors maintain portfolios with very large capital volumes. The security of investments is preeminent, but on the other hand one wishes to achieve a reasonable return. A direct generalization of the classical Markowitz approach, the Dynamic Mean-Variance-Analysis (DEVA, Frauendorfer 1995), models the tradeoff between risk and return but allows a much more detailled representation of expected market developments over several periods when making the current decision.

Goals of the project are the theoretical analysis of the resulting quadratic multistage stochastic programs and the development and implementation of highly efficient solution algorithms.