Introduction

Pre-snap motion can force a defense to adjust before the ball is snapped. This analysis asks: Which three NFL offenses get the most defenders to reposition through a single player’s pre-snap motion? Using the 2025 Big Data Bowl dataset, I define a defensive repositioning metric, rank offenses, investigate how motion distance and duration relate to defensive reactions, and explore whether those reactions are associated with yards gained.

Data and Setup

My analysis uses the games, plays, players, player_play, and each of the nine weeks of player tracking data provided.

Identifying Single-Player Motion

I first identify players flagged for motion in the player_play data. I retain plays with exactly one flagged motion player so that the defensive response can be attributed to one identifiable offensive motion event rather than multiple simultaneous motions.

Measuring Defensive Repositioning

For each tracking week, I identify the period between the line set and the snap, locate the start of motion, and measure the motion player’s lateral displacement. A play qualifies when the flagged offensive player moves at least 2 yards laterally.

I count a defender as repositioned when they move less than 0.5 yards cumulatively in the five frames before motion, have all five pre-motion frames available, and then move at least 1 yard in net displacement from motion start through the snap. Tracking frames occur at 10 Hz, so five frames represent about half a second.

Rendering note: Because the processing of 9 tracking weeks takes a long time (and a lot of space) the first run processes all nine tracking weeks and saves them to a compact .rds file. Later renders load that file instead of repeatedly loading them.

Ranking NFL Offenses

The primary team metric is the average number of defenders repositioned per qualifying single-player motion play. To reduce the influence of very small samples, the main ranking requires at least 50 qualifying plays. I also compare the results using a stricter 75-play threshold to analyze the difference produced by a larger sample size.

Original Team Rankings

The chart ranks eligible offenses by average defensive reactions. The top three are highlighted in turquoise, and the team name and qualifying-play count can be seen on hover.

How Motion Distance Relates to Defensive Reactions

Next, I compare the motion player’s lateral displacement with the number of defenders repositioned. The boxplots show the distribution of motion distance for each defensive-reaction count, with the number of qualifying plays printed below each category. I recognize that this is not proof that motion distance caused a defender to move, but felt it was still useful to showcase.

Team-Level Motion Distance

I also compare each team’s average lateral motion distance with its average defensive reactions, and fit a simple linear model to describe the play-level association.

Accounting for Motion Distance and Duration

Longer and more sustained motions may naturally be associated with more defensive movement. I therefore compared three descriptive models: motion distance, motion duration, and a combination of both predictors. The combined model is used to estimate an expected number of defensive reactions for each play.

Adjusted Team Rankings

I calculate each team’s average reactions above expected by subtracting model-predicted reactions from observed reactions. This provides a comparison of offenses after accounting for motion distance and duration, but is not a definitive measure of motion quality.

Animation of a Sample Play: Houston’s Pre-Snap Motion

To make the metric more tangible, I animated a qualifying Houston play against Indianapolis. The motion player appears in yellow, while defenders who meet the repositioning criteria (appear to respond to the offensive motion) are shown in purple. I also implemented trails to better illustrate their movement as the play approaches the snap.

Does Defensive Repositioning Translate to Offensive Production?

Last, I examine whether more defensive reactions are associated with a greater number of offensive yards. I look at both team averages and individual plays, because a team-level relationship can differ from a play-level relationship.

Team-Level Comparison

The bubble chart below plots average defensive reactions against average yards gained. Each bubble represents an offensive team, bubble size indicates their number of qualifying motion plays, and color represents the average yards gained. Hovering over a bubble will allow you see the team and its statistics. This is especially helpful for the few teams with less qualifying plays that are also situated close to one another on the chart.

Play-Level Comparison

I also calculate the play-level correlation and compare average yards gained across plays with different numbers of defenders repositioned. Sample-size labels help contextualize categories with relatively few plays.

Conclusion

The original rankings identify the offenses that generated the most qualifying defensive repositioning on average. The adjusted rankings show how those comparisons change after accounting for motion distance and duration. Across my variety of analyses, I found that defensive repositioning does not show a clear, consistent positive relationship with yards gained. This metric therefore describes a particular type of pre-snap defensive response, rather than being a direct predictor of offensive success.

The original rankings place Houston (HOU) first, with an average of approximately 3.14 defenders repositioned per qualifying motion play across 59 plays. The New York Jets (NYJ) followed at approximately 2.90 defenders across 108 plays, and Arizona (ARI) ranked third at approximately 2.68 defenders across 76 plays. These results answer my original research question using the 50-play minimum. However, Houston’s smaller sample of qualifying plays is worth considering: when I raised the eligibility requirement to 75 plays, Houston no longer qualified, illustrating how a team’s position can depend on the minimum sample size selected.

I also found that motion distance and duration are important factors to be considered when interpreting the rankings. Greater lateral motion distance was moderately associated with more defensive repositioning (a play-level correlation of approximately 0.47). The distance-only and duration-only regression models each explained around one-fifth of the variation in reactions, while the combined model explained approximately 29%. After comparing actual reactions with those predicted by the combined model, Houston and the Jets remained the top two offenses, but Jacksonville moved ahead of Arizona into third. This suggests that rankings based only on raw defender counts may favor offenses that use longer or more sustained motions. The adjusted results provide another useful perspective, but they do not establish which offense is most effective at causing defensive movement.

Finally, I wanted to see whether the metric was connected to what happened after the snap. The team-level relationship between average defensive reactions and average yards gained was slightly negative (approximately -0.19), while the play-level relationship was essentially zero (approximately 0.04). The bar chart also showed that average yards gained did not increase consistently as the number of repositioned defenders increased. This was an important distinction for my interpretation: motion may change a defense’s alignment or reveal information to the offense without necessarily producing an immediate gain in yards. The analysis does not measure every possible benefit of motion, such as influencing coverage assignments or helping a quarterback read the defense.

Limitations: These results depend on the chosen motion and movement thresholds, the nine weeks of tracking data available to us, and the minimum-play eligibility rule. A useful additional validation would be to compare defenders’ movement before and after motion begins more directly, while recognizing that observational tracking cannot establish that motion alone caused the response.