Multi-asset worst-case optimal portfolios

We generalize the worst-case portfolio approach of Korn & Wilmott (2002) to a multi-asset setting. The nonuniqueness of indifference strategies results in a much more complicated portfolio optimization problem as in the single risky asset framework. To determine the worst-case optimal portfolio processes we develop two new approaches, a Lagrangian multiplier approach in the log-utility case and a combined constrained HJB equation and indifference strategy approach for dealing with power-utility functions. Various examples illustrate remarkable effects and differences compared to the single risky asset setting, in particular the possibility for using some stocks for crash hedging and thereby allowing stock investment possibilities that are not present in the single-stock case.