We study worst-case utility maximization in a financial market with drift uncertainty. By a dual approach we derive a minimax theorem and show that the optimal strategy converges to generalized uniform diversification as uncertainty increases. Reasonable uncertainty sets can be found with filtering techniques.
Investors estimate the drift based on returns and expert opinions. We investigate the effect of large numbers of expert opinions on the estimation. Lastly, we show how the worst-case approach can be combined with filtering for a financial market with stochastic drift.