Abstract: A system maps a user's interaction with a game controller to predefined hand positions that can be used in gaming and multimedia applications. A game controller with sensors receives user input with capacitive sensors and generates sensor outputs. A controller modeling module receives the sensor outputs and trains a model based on the outputs. The trained kinematic model maps the sensor outputs to one or more of a plurality of predefined hand positions. When an application requests the status of a user's hand position, e.g., through an API, the mapped hand position is sent to the application.