General-Purpose Planning Algorithms in the Card Game Duelyst II.

Duelyst II is an online collectible card game (CCG) that features a 9x5 grid board, making it a cross between the popular CCG Hearthstone and chess. It is a partially-observable stochastic game (POSG) with a large branching factor and the ability to take several actions in a time-limited turn, making it a challenging domain for AI. The existing "starter AI" in the game is an expert-rule-based player that is limited to using certain decks and is weak against humans. We develop simple general-purpose planning algorithms that are able to consistently beat the starter AI using little domain knowledge and no learning. The most complex of these is a variant of Monte Carlo tree search (MCTS), for which we show that a novel action factoring method is helpful under certain conditions.