HMMotion: Predicting coverage schemes in the NFL

Using pre-snap tracking data and HMM-model-generated features to predict man vs. zone coverage accurately.

Authors

Robert Bajons

Joint work with Rouven Michels and Jan-Ole Koslik

Published

January 1, 2025

Hit Counter by Digits

Table 1
Resource Date Link
Rouven’s talk presented at the 2025 New England Symposium on Statistics in Sports (NESSIS) 2025-09-27 Talk Rouven

Note

Rouven, Ole, and I participated again in this year’s NFL Big Data Bowl (for last year’s submission, which obtained an honorable mention, see here). Although this year we did not earn any mention from the NFL, we believe that the project is interesting and are currently working on refining the details. More info on the project will follow on this page in due time. For now, you can have a look at our Kaggle submission or the code on github.