Case Study: NFL achieves real-time tackle probability insights with AWS Elemental

NFL delivers real-time tackle predictions in under 100 ms with AWS Elemental

The National Football League (NFL) sought to quantify the complex factors that lead to a successful or missed tackle, a challenging task that required analyzing massive datasets in near real-time. To achieve this, the NFL's Next Gen Stats team partnered with AWS Elemental, leveraging the Amazon SageMaker suite of machine learning tools to build a new AI-powered model.

AWS Elemental implemented a solution using Amazon SageMaker to process a full season's worth of data, developing approximately 15 new predictive features through advanced feature engineering. The final model showed improvements across all key performance metrics and could generate over 15 million predictions in about an hour, a process that previously took weeks. This allowed for a real-time tackle probability metric with sub-100 millisecond response times, enabling its use during live broadcasts to provide fans with instant insights.

NFL