Where does the data come from? Short answer: we build it.
Every pressure, route, coverage, and outcome you see in SūmerLive is generated in real time by a deep learning model trained on raw tracking data and validated by Sūmer experts.
How It Works
The model processes player and ball movement during live games. Its job is to identify what's happening on the field as plays unfold: formations, coverages, pressure types, route concepts, targets, and outcomes.
It's trained on licensed Genius Sports tracking data and validated by our internal football experts—more than 500 years of combined NFL experience across scouting, coaching, and front offices.
On every snap, the model charts over 20 distinct data points and this number only increases as our model improves.
Built for Speed and Accuracy
Live charting is different from film analysis. There's no replay, no second look, no time to review. The model has to be right the first time.
To validate the model's accuracy, our experts chart plays independently, without seeing the model's output. Only when our experts agree with the model at least as often as they agree with each other is the data signed off and available for public usage.
That process ensures what you see in SūmerLive is both fast and reliable.
Why It Matters
When you check SūmerLive for pressure rates, route efficiency, or coverage tendencies during a game, the answer doesn't come from manual tagging or crowd-sourced stats.
It comes from a model trained to see football the way evaluators do. Now applied in real time.
Check it out here: sumersports.com/live
