Introduction
In professional football, every play contains dozens of micro-decisions that can be the difference between success and failure. Today, we’re diving deep into a fascinating play through the lens of our FootballAI model, which tracks player movements and model estimates in real-time.
Spatial Analysis
The model predicts a gain of +26.9 yards on the play at the time of the second handoff to Jameson Williams, showcasing how our AI system processes spatial relationships and player velocities to project play outcomes. FootballAI produces so much more than just expected yards, though. We can diagnose decisions at every step of the play. As the play progresses, FootballAI predicts tackle probabilities and so much more.
Model Insights
The FootballAI model’s strength lies in its ability to process multiple variables simultaneously:
- Player positions, velocities, and accelerations
- 120 Real Time model prediction results per frame.
The Key Moment: A Tale of Two Players
What makes this play particularly interesting is the intersection of two critical player trajectories. Our probability analysis shows two notable spikes in tackle likelihood:
- Quan Martin (#20) shows an earlier but significant spike around frame 38 (3.8 seconds after snap), reaching about 0.4 on our tackle probability scale.
- Percy Butler (#35) reaches a peak tackle probability of approximately 0.725 at frame 84 (8.4 seconds after the snap).
Conclusion
This single play, lasting mere seconds in real-time, reveals the incredible complexity of modern football when viewed through our AI lens. The model helps us understand not just what happened, but why it happened and what factors contributed to the play’s outcome. By breaking down these moments in such detail, we gain a deeper appreciation for the split-second decisions that shape every football game.
In the coming months, there are other outputs that will become available to our clients with the addition of film for side-by-side comparisons down to the frame level. In addition, we are dedicated to bringing radical transparency to football analytics and our platform. Users will be able to click on a metric and see example plays demonstrating why a certain grade was earned.
Our FootballAI model continues to evolve, offering increasingly sophisticated insights into the complexity of football. This analysis demonstrates how Sumer can illuminate aspects and moments of a play helping scouts evaluate players, coaches prepare for games, and players improve their performance.
