Adaptivity of Card Recommendation Systems for Legends of Code and Magic

The complexity of online games asks for support of new players, e.g., by recommending deck improvements in case of Collectible Card Games. But many online games regularly change the properties of existing game elements to either balance dynamics or to keep experienced players engaged. To successfully offer a recommendation service, it is inevitable that they adapt to such changes. We study four card recommendation systems (RS) based on logistic regression models trained from game log data. The RS are compared based on their models' ability to adapt to different scenarios of change (game rules and opponent behavior). Using a genetic algorithm, a simulation of a player community was implemented and RS accuracy and adaptivity were explored in a competitive environment. We extend earlier work by including different player strategies (agents) and observe that they strongly affect the RS' model parameters and their spread in the player population. Complex interactions were observed, but systems with high accuracy have proven more robust against change, but it took them more data (and thus time) to adapt. As time is a crucial factor in competitive environments, versatile RS may nevertheless be displaced by specialized RS that adapt to a specific population faster.