Video game controllers have a high influence factor on players, as they are responsible for the fun, motivation, and personality of a game. The organization and arrangement of the buttons are one of the relevant factors when developing new controllers since they are responsible for serving as an input of actions within the games. This work presents the construction of a generative design model to support game designers finding different and innovative layouts of virtual controllers for their games. The generative design produces many valid designs or solutions instead of one optimized version of a known solution. This solution was developed by linking genetic algorithms to generate a large number of layouts and machine learning techniques (SVN) to classify individuals between valid and invalid, seeking to facilitate the exploration of the design space by the designer. The tests performed sought to measure the variability of the results generated by the proposed model, showing that several solutions of different controllers with different configurations can be developed for a game.