Each player then has to play a card from the same suit if they have one, otherwise any card from their hand is permissible. Interestingly, predicting correctly gives you extra points or ‘tricks.’ So, if many games are being played, a good prediction strategy is needed.Īfter bidding, the round starts with the player to the left of the dealer playing a card from their hand. We played a variation where each player is given 7 cards and then the players take it in turn to predict how many rounds they will win. The idea for this project started when I was introduced to the game, Oh Hell, by one of the contributors to the project. The main author of the project and the blog-post is Albert Nyarko-Agyei. Special credits to Henry Charlesworth for providing valuable insights. The rules and an online version of the game can be found here. This post is the corresponding write-up for a WDSS project in which Albert Nyarko-Agyei, Alexandru Pascru and Ron Cvek applied reinforcement learning to better understand the mechanics of a British card game- Oh Hell.
