DeepStack is an algorithm designed to play heads-up no-limit Texas hold'em poker, a game with imperfect information. It combines recursive reasoning to handle information asymmetry, decomposition to focus computation on relevant decisions, and a learned intuition to estimate the value of holding different cards in various situations. DeepStack uses deep learning to train its intuition, which is then used to approximate a Nash equilibrium strategy during play. In a study involving 44,000 hands, DeepStack defeated professional poker players with statistical significance, outperforming previous approaches that relied on abstraction and offline computation. The algorithm's performance is theoretically sound and it produces strategies that are more difficult to exploit than those of prior methods. DeepStack represents a significant advancement in the field of artificial intelligence, particularly in handling complex, sequential games with imperfect information.DeepStack is an algorithm designed to play heads-up no-limit Texas hold'em poker, a game with imperfect information. It combines recursive reasoning to handle information asymmetry, decomposition to focus computation on relevant decisions, and a learned intuition to estimate the value of holding different cards in various situations. DeepStack uses deep learning to train its intuition, which is then used to approximate a Nash equilibrium strategy during play. In a study involving 44,000 hands, DeepStack defeated professional poker players with statistical significance, outperforming previous approaches that relied on abstraction and offline computation. The algorithm's performance is theoretically sound and it produces strategies that are more difficult to exploit than those of prior methods. DeepStack represents a significant advancement in the field of artificial intelligence, particularly in handling complex, sequential games with imperfect information.