In this paper we consider the robust portfolio selection problem
involving two types of uncertainties; the uncertainty in the
distribution of exit time and the uncertainty in the distribution
of portfolio return conditional on exit time. To deal with these
uncertainties, we propose a tractable approach by applying
worst-case VaR strategy to the case where partial information on
the exit time distribution and on the conditional distribution of
portfolio return is available, and formulate the corresponding
problems as semidefinite programs which can be efficiently solved.
Moreover, we present some numerical results with real market data.