Deck Building in Collectible Card Games using Genetic Algorithms: A Case Study of Legends of Code and Magic

Collectible card games (CCGs) are a kind of game in which players use a variety of cards to achieve the goal of game, for example, defeating the opponent. To play a game, players need to build a card deck, which consists of specified number of cards; only cards in the deck can be used in the game. Deck building is an important part in playing CCGs. This paper addresses deck building in a recently released CCG, Legends of Code and Magic. Our strategy is to evaluate cards by some scoring functions and then select cards into the deck accordingly. We adopt three types of scoring functions and use genetic algorithms (GA) to set proper values of parameters in the scoring functions. We checked the performance of the three versions of deck-building GAs against existing agents and found that scoring by card attributes is effective. We also analyzed the decks built by these algorithms and discussed their pros and cons.