We define a class of dynamic Markovian games that we call directional dynamic games (DDG) in which directionality is represented by a partial order on the state space. We propose a fast and robust state recursion algorithm that can find a Markov perfect equilibrium (MPE) via backward induction...