For the 2021-2022 NFL season, the NFL and AWS are partnering to add a few new statistics to their Next Gen Stats (NGS) tool. The league will now track fourth down and two-point conversion analytics with Next Gen Stats Decision Guide, powered by AWS.
By leveraging AWS’s broad range of cloud-based machine-learning capabilities, the NFL is taking how it can Stat That on game day to the next level, so that fans, broadcasters, coaches, and teams can benefit from deeper insights.
Training data from traditional box score statistics, as well as data collected from the stadium, will run through hundreds of processes within seconds, with the output fed into Amazon Sagemaker. From there, machine-learning models built by the NGS team ingest the data, which continually train and refine the models. The machine-learning models are then used or inferenced in real-time during games to generate outputs such as formations, routes, and events.
Next Gen Stats uses a flexible machine-learning algorithm that frames pass-completion probability as a linear-regression problem. From contextual inputs, the algorithm calculates a probability that any given play would result in a completed pass or not.
Using the business intelligence tool Amazon QuickSight, the NFL is able to gain greater insight while also opening a window for fans, broadcasters, and editorial teams to engage with data.
Some of the stats translate well into real-time graphics during broadcasts. Others provide announcers with valuable points of discussion. The stats can reveal an understanding of fan engagement, game presentation, and can help model the potential impact of adjusting the rules of play or how the game is called. It can even improve player performance and safety.
In complex NFL scenarios, business-logic-specific rule-based systems can start to unravel as there are simply too many variables to process. Machine learning helps calculate complex stats, like completion probability, in real-time when there are too many variables to rely on a simple decision tree.
Not all passes are created equal. But with machine learning, the NFL can enhance traditional box score statistics and contextualize events on a per play basis. Teams are using analytics as a complementary tool, to enhance decisions typically drawn from instincts and experience.
The league has built several new stats on AWS, each of which relies on different data points. The sports industry's use of AI technologies is driving shifts in real-time analysis, automation, prediction, and dynamic information.