Visualizing Wide Receiver Routes with Player Tracking Data

oza 1Image from

This May, NFL teams will finally be granted access to league-owned player tracking data from the 2015 season. The league had previously reached an agreement with Zebra Technologies to install chips in players’ shoulder pads that use radio frequency to broadcast each player’s location and speed 25 times per second, effectively generating data similar to SportVU in the NBA and Statcast in the MLB. This data has already been used in TV broadcasts for simple purposes such as speed and distance run (see above for an example). While these simple metrics can make for good supplemental TV broadcast content, they aren’t exactly all that relevant to the player personnel or game management decisions that teams face to try to gain a competitive advantage on one another. So are there any relevant uses of player tracking data from a football operations standpoint?

Luckily, ESPN has been developing a similar player tracking technology which was put to the test during the second half of the 2016 Under Armour High School All-American Game. And the participants of the 2016 Sloan Sports Analytics Conference Hackathon, presented by ESPN, received access to that player tracking data. Submissions ranged from quantifying and visualizing offensive line performance to the shape/stretch of a defense to the routes run by receivers and the coverages employed against them. I focused on visualizing wide receiver routes and defensive back coverages. Although only a few hours of data analysis and visualization of one half of a high school all-star game will barely scratch the surface of what is possible with football player tracking data, I nevertheless produced some interesting results.

With the ggplot2 package of the statistical software, R, I generated two sets of visualizations for any given play that plot each player’s starting position on the field, and the routes of the offensive skill players and defensive backs. Distance to the nearest defender is used as a color scale for one set of visualizations, running from red (open) to white (covered). The player’s instantaneous speed is used as a color scale for the other set of visualizations, running from red (fast) to white (slow).

Pressure, Scrambling and Getting Open

Team Armour has the ball. It’s 1st and 10 with 14:53 left in the 3rd quarter (right after the second-half kickoff). They line up in a standard shotgun formation, as they did the entire game. Team Highlight lines up in a 4-3 formation playing Cover 1, as they did the entire game (high school all-star games make for simplified playbooks).

oza new

The split end to the far left makes a good post cut to get open as the QB scrambles away from pressure and out of bounds

After the ball is snapped, the split end runs a deep post, accelerating off the line of scrimmage. The cornerback lined up across from him backpedals, then opens his hips to turn and run to keep the receiver in front of him. Roughly 20 yards out, the receiver sharply decelerates as he makes a post cut. After being shadowed closely by the cornerback, suddenly the receiver is open. After faking a handoff to the running back, the QB had looked left, but quick pressure had caused him to scramble right and tuck and run. The play ends as he runs out of bounds for a loss of five yards.

oza 1

The split end on the far left decelerates sharply as he makes his post cut, creating separation between himself and the cornerback

Off-Man Coverage, Sloppy Routes and the “Coverage Sack”

Two plays later, Team Armour is facing 3rd and 15 with 14:06 left in the 3rd quarter. Team Highlight’s defense, still in 4-3 Cover 1, realizes this and sags back to prevent a first down completion down the field.

oza 2

The Team Highlight defense sags back on 3rd and 15

After the ball is snapped, the QB looks left, but the left side route combination is very slow to develop – the slot receiver, in particular, runs a very circuitous route. Because the defense is sagging, while the receivers have space around them near the line of scrimmage, none are open past the first down marker. After about 4 to 4.5 seconds, the QB takes a sack.

oza 3

Since the defense sags back and the receivers on the left side take inefficient routes, they are open near the line of scrimmage, but not near the 1st down marker

Applying this to Football Operations

The above visualizations demonstrate a plethora of potential analysis opportunities afforded to teams by player tracking data. And I only illustrated two plays! This was just scratching the surface of player tracking data’s potential in football. Nevertheless, I think I have demonstrated a few concrete ways for a smart team with analytical capabilities (admittedly, not exactly a given among NFL teams) to gain a competitive advantage with respect to player personnel decisions:

  • Acceleration and deceleration’s relationship with getting open – the 1st and 10 play showed a receiver using his deceleration to get open on a post route. But should all receivers put on the brakes right before all of their cuts, or do certain receivers running certain routes against certain coverages get open more frequently without hard deceleration? With data from a full season, coaches and front office personnel should be able to better understand this relationship and possibly tweak the instructions they give to their receivers in order to get them open more frequently.
  • Open vs. Covered – as we can see from the 1st and 10 play, just because a receiver is open doesn’t necessarily mean he will have the receptions and yards to show for it. Who runs the best routes? In other words, who gets open the most? How does that vary on short routes vs. deep routes? Which defensive backs are the best at preventing their man from getting open? Which receivers are overrated/underrated because they don’t get the ball even when they’re open?
  • Route efficiency – we have already seen some of this with Statcast in the MLB, allowing us to get closer to figuring out which outfielders get the best jump and take the most direct route to fly balls hit their way. As we can see from the 3rd and 15 play, this is readily applicable to wide receivers – who runs the most efficient routes? Who might be costing their QB precious tenths of a second with inefficient routes on 3rd and long?
  • Coverage Sacks – as we can see from the 3rd and 15 play, there are many times when a sack is not the result of quick pressure but rather the result of nobody open far enough down the field. Was a given sack the result of a stellar pass rush, a QB holding the ball for too long or tight coverage down the field?

Further Considerations

There are many nuances to take into consideration when conducting the suggested analysis on league-wide data. Most obviously, we must adjust for zone coverage and probably should also adjust for inside/outside techniques employed by defensive backs. Additionally, not all defender distance is created equal. As an extreme example, a receiver running a vertical route 3 yards ahead of a defensive back is a much better for the offense than a receiver running a vertical route with the defensive back 3 yards ahead of him. It also probably doesn’t make any difference whether a receiver is open by 8 yards or by 12 yards on a short route – if he’s that wide open, he’s wide open. Though I did straight defender distance, we really should be measuring open vs. covered or something along the lines of “probability that a pass to this player will be completed.”

And of course much of this analysis is already conducted by scouting departments via film study and player grading. However, as there are some things that are much easier to capture on film than with data, there are some things that are much easier to capture with data than with film. Route efficiency, in particular, seems to be one of those things. Moreover, with player tracking data, teams have the potential to conduct large amounts of analysis and potentially grade every player on certain dimensions (such as average defender distance on certain routes) much more quickly, more efficiently, and more bias-free than video analysis. That’s not to suggest that the tried and true tradition of watching tape will die – far from it. Film study and data analysis have different strengths and weaknesses and a smart team would do well to learn from both.


It will be interesting to see to what extent NFL teams incorporate the insight gained (or lack thereof) from player tracking data into their player personnel and game management decisions. Even after SportVU was made available to all 30 NBA teams, many lacked the front office personnel to get anything useful out of the data. The same can be said for Statcast (and, going back even farther, for PITCHf/x and HITf/x) with MLB teams, despite having had quite a head start over other sports thanks to the popular success of the book, Moneyball. NFL teams have been notoriously conservative when it comes to implementing analytics to gain a competitive advantage on the football side of the organization (and, to some extent, on the business side as well) and very few teams have the analytical capability to get useful information out of the data. But that only makes the potential advantages gained from player tracking data even greater.

Thanks to ESPN and the MIT Sloan Sports Analytics Conference for putting on the hackathon and thanks to ESPN for access to the data

Nik Oza

Georgetown Class of 2016


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s