Determining actual player contact from 2D tracking data

Determine whether physical contact between the ball-carrier and defenders can be identified using only the two-dimensional NFL Next Gen Stats tracking data provided in the NFL Big Data Bowl 2024 dataset, in order to validate the contact window definition employed in the fractional tackles framework.

Background

The paper introduces a model-free framework for valuing defensive tackling contributions using player tracking data by defining “contact windows,” time intervals when a defender is within a 1.5-yard threshold of the ball-carrier. This threshold is calibrated using annotated first-contact and tackle events in the NFL Next Gen Stats data, but the tracking data provide only two-dimensional positions and derived movement attributes.

Because the dataset does not explicitly encode physical contact, the authors note a key limitation: they do not know whether players are actually making contact based on 2D tracking alone. Resolving whether contact can be inferred from the available 2D tracking signals is essential for validating and potentially refining the contact window identification that underpins the fractional tackles metric.