The National Football League and Amazon Web Services have announced a multi-year extension of their technology partnership. At the center of this collaboration is the league-wide rollout of a new predictive machine learning model called Tackle Probability. The tool expands the existing Next Gen Stats platform, which has gathered detailed tracking metrics for several years. With this algorithmic framework, the league and the cloud provider aim to elevate the quantitative evaluation of defensive gameplay to a new level.
The technical foundation of the system originated from data science approaches developed during the league's Big Data Bowl competition and was trained using AWS SageMaker. The algorithm continuously ingests optical tracking data provided by Next Gen Stats throughout each play. For all eleven defensive players on the field, the system calculates approximately 20 different parameters every 0.1 seconds. These variables include the direct distance to the ball-carrier, pursuit angles, intervening blockers, and instantaneous acceleration rates.
By fusing these metrics, the model evaluates in real time the probability that a specific defender will successfully tackle the ball-carrier. The system also systematically logs missed tackle attempts whenever a defender fails to finish a play despite possessing a high statistical likelihood of success. Conversely, this mathematical framework enables an objective assessment of running backs and wide receivers by quantifying their elusiveness when breaking away from imminent contact.
The generated metrics are integrated directly into live NFL broadcast graphics to provide viewers with deeper context on how close a defensive stop was. Beyond fan engagement, the model offers substantial utility for coaching staffs and scouting departments across the league. Teams are leveraging the data to perform objective evaluations of defensive pursuit angles, moving beyond traditional box-score metrics that merely record completed tackles.
With the rollout of Tackle Probability, the league reinforces an ongoing shift toward real-time predictive analytics in professional football. Aspects of defensive positioning that previously relied on scout intuition are now translated into standardized probability curves. AWS and the NFL intend to continue refining these tracking models to provide deeper evaluations of blocking assignments and field spacing in future seasons.