AWS’ 10-Year Partnership with the NFL Next Gen Stats Powering Stats for Fans

Amazon Web Services (AWS) has had a long-standing partnership with the NFL and its Next Gen Stats (NGS) platform. For die-hard fans or even casual viewers tuning in to Superbowl LX, this partnership results in the most important stats and metrics being relayed to coaches, analysts and viewers.

Next Gen Stats originated in 2015 in the form of RFID chips, which were placed in every set of shoulder pads worn by players. Leveraging 20 ultrawideband receivers mounted in the stadium, the NFL was able to capture coordinates and data of all 22 players and the ball. Over time, this technology has evolved, thanks to an infrastructure built on AWS’ SageMaker platform.

Amazon’s SageMaker hosts an XGBoost machine learning model that can integrate everything from the shape of a throw to pressure depth and insanely in-depth metrics. By 2018, the NFL and AWS were able to expand their partnership to the Big Data Bowl, the league’s first large-scale effort to open player-tracking data. This included a months-long hackathon, which resulted in the training of ML models on historical tracking data to generalize unseen plays.

Currently, the infrastructure of NGS has evolved quite a bit. The system now uses 4K cameras to capture the full 3D position of players. This includes key joints such as shoulders, elbows, knees, hips, and hands. With this, a massive amount of data is able to be fed to coaches and even analysts covering a game. Additionally, broadcasters will have meaningful bits of data to highlight to fans watching from home.

Ari Entin, Head of Sports Marketing at AWS, spoke to iPhone in Canada about how these stats may be leveraged during Superbowl LIX. “Let’s say during the Super Bowl, for example, [NGS] can be communicating with the producers from NBC for the game and saying this play that just happened and this is a statistical anomaly. Or Kenneth Walker got 24 rushing yards over expected on this play. And this hasn’t happened in this kind of setting in five years. So they’re not just giving producers or broadcasters the raw stat. They’re also able to give them context on why it’s interesting not just in the context of the game, but why fans might care about it too.”

NGS is able to capture and relay data in nearly real time. The infrastructure use AWS’ servers to process data within around 7000 milliseconds. Once the data is processed, it’s sent to the cloud where ML models run in under 100 milliseconds and returned to the production team. The full capture-to-analysis workflow is said to be under a second. With broadcasts like Thursday Night Football operating on a two-second delay, NGS data is effectively delivered in real time as plays develop on screen.

“The nice part is that whether fans realize it or not, NGS is impacting some of the viewing experience,” Entin says. “Whether it’s the specific stat that is going to show up in the 15 seconds right after a play happens, or some other aspect of the game that you know AWS is helping power underneath the hood.”

As Superbowl LX is being held on Sunday, February 6th, fans can look forward to the most amount of intregal data being shown on screen than ever before. For long-time fans, this solidifies further connection with the game. For newcomers or returning fans, NGS offers insights that help viewers better understand or further appreciate the athletes and what they’re putting out while on the field.