Summary: This paper evaluates quarterback performance in the National Football League. With the availability of player tracking data, there exists the capability to assess various options that are available to quarterbacks and the expected points gained resulting from each option. The options available to a quarterback are based on considering multiple frames during a play such that a current option may evolve into new options over time. Our approach also considers the possibility of quarterback running options. With tracking data, the location of receivers on the field and the openness of receivers are measurable quantities which are important considerations in the assessment of quarterback options. Machine learning techniques are then used to estimate the probabilities of success of the passing options and the estimated expected points gained from the options. The estimation procedure also takes into account what may happen after a reception. The quarterback’s observed execution is then measured against the optimal available option.