Play Next in Steam: A Game Recommendation System! We built a video game recommendation system based on a matrix factorization algorithm and trained the model on a subset of real Steam Users and their playtimes in the games in their libraries. The model predicts a Steam User's potential playtime in a game they have not played before. With this prediction, the recommender can recommend games to a Steam User based on what games the model predicts the user will put the most hours into. This also provides helpful information to the Steam User by seeing which games can potentially give them the most playtime per game price.