“Can Mumbai Indians Chase the Target?”: Predict the Win Probability in IPL T20-20

Indian Premier League(IPL) is a popular T20-20 over cricket tournament, played with the participation of players from different cricket playing nations. The speedy nature of the game has caught the attention of many sport enthusiasts and data scientists around the world. Predicting the match stats and winner has become a highly discussed topic nowadays. However, due to unavailability of a annotated dataset and limited IPL matches per team, creating a real-time win probability model is a challenging task. In our work, we present a IPL annotated dataset which can be used by research community in the future. We also present a model that use many key features from player stats to determine the win probability of a given match. We analyse different types of attributes from playing teams and also the venue. Using these various attributes we created a model that is capable of predicting the win probability in real time (i.e., winning probability is predicted in every cricket over).