Examining Wide Receiver Metrics

by Eric Eager|May 23, 2023


As the NFL calendar transitions to summer, we are all searching for an edge when preparing for our fantasy football leagues, best ball drafts, and fall prop bets. One area that has grown in interest over the past few years is trying to map what wide receivers do when they are not targeted.

For most of football history, we judged pass catchers by what they did when they caught the ball. Total catches, total yards, yards per catch, and touchdowns were how we ranked receivers year in and year out. Pro Football Reference increased our understanding of the game when they went back and added targets to player statistics going back to 1992 (be careful when studying receivers from that era since if they played before 1992, they might have more listed catches than targets!).

My former employer, Pro Football Focus (PFF), went even further when they started charting games almost two decades ago. They charted not only the number of snaps a player played but how many routes the player ran. That gave us yards per route run (YPRR), which is probably the best publicly-available efficiency metric for pass catchers that we currently have (be aware of the biases that YPRR have as one of my former interns and current Eagles analyst Zach Drapkin laid out here).

However, these metrics – as with many others – leave out some important information. To obtain a target, there are things that have to happen that a pass catcher does not fully control. Before the snap, the receiver does not choose where in the quarterback’s progression they would be. The offensive line has to do a good enough job to give the quarterback the ability to throw the ball. The quarterback has to then throw an on-target pass to a receiver, and that receiver must be the receiver we are talking about. While the receiver can sometimes help by getting open quickly, being the most open player, and/or garnering the quarterback’s trust over time, they will never be in complete control of the situation.

What the receiver does have more of a say in, though, is whether or not they beat the defender guarding them. For years, football analytics companies shied away from measuring such things due in large part to the immense cost of doing so by hand. About five years ago, though, the NFL put RFID chips in the shoulder pads of every player, making the impetus to collect such information stronger. Automatic measures of pass catcher openness are not trivial but can be done as ESPN demonstrated last year with their new suite of metrics for receivers. PFF has also charted every receiver/coverage player matchup since 2019.

Teams undoubtedly believe in buying players low when they get open in low-production offenses. For example, Robert Woods never had over 700 yards during his time with the Bills causing some to wonder what value he brought to the Rams with his five year deal before the 2017 season. Woods, who was adept at getting open in Buffalo, caught the attention of Los Angeles and went on to be one of the league’s most-valuable receivers in helping the Rams go to two Super Bowls while with the team.

The data I want to use to make my point in this article, though, has been collected by football analyst Matt Harmon since 2014 as a part of his Reception Perception project (which, for my money, is very much worth the yearly subscription). Harmon goes through a sample of notable wide receiver games and charts whether or not the player “wins” their matchup, what type of route the player runs in that matchup, and whether that matchup was in man or zone. This data captures something fundamental about a player as evidenced by its significant year-to-year correlation:

Success rate, or the rate at which a player wins their repetition (minimum 235 sampled routes each year, n = 143 players) is correlated at a rate of r = 0.75, which is very high for a metric in American football. This consistency is similar to what my former colleague, Timo Riske, found when looking at PFF data and how it correlated with things like time to throw and play outcome in this article:

How PFF route-running grades teach us more about coverage matchups | NFL News, Rankings and Statistics | PFF

Versus Man vs. Versus Zone

As seen above, getting open is a trait for wide receivers in the NFL, and it can be measured. The next consideration I want to look at is similar to the orthogonal decomposition one would perform in linear algebra or the “signal versus noise” framework popularized in data science. If a metric is stable year-to-year, we can largely compose it into a stable (e.g., signal) and unstable (e.g., noise) term. Looking at how players perform in man coverage, you get about the same stability as how they perform overall but with a significantly smaller sample size of plays (minimum 112 sampled routes each year, n = 145 players):

Here, we get a correlation coefficient of r = 0.72, which again is quite high for a football metric. It appears that getting open is a trait, and getting open against man coverage is a trait. Let’s look at the same metric for zone coverage:

Here, the r value is equal to 0.42 (minimum routes at 93, n = 130 players). This is nowhere close to the value for man coverage but does not exactly qualify the metric as “noise” either as almost every running back metric you can find has a lower correlation year-to-year than this one does, and the best quarterback metrics are just a tad above these.

Thus, what this data tells us so far is that the ability to beat man coverage is more fundamental to a player than beating zone coverage is, but there is signal to beating zone coverage. If a player’s scouting report says something like “has a good feel for zone coverage,” that cannot be dismissed offhand as noise, but it must be checked with data.

One thing that is difficult to show with data is that ability versus man coverage translates to ability versus zone coverage. Here we looked at all players that met one of the two route thresholds above and looked at how success versus man translated into success versus zone the following year and vice versa. There were n = 187 such players. First, how does success beating man coverage translate to beating zone:

This is very noisy, and r is only 0.18. Now for success rate in zone predicting success rate in man:

This is a little better but still noise (r = 0.26). This leads us to a pretty definitive conclusion that there are wide receivers who excel against man coverage and there are wide receivers who excel against zone coverage, and we can be more confident that a player is good or bad against man coverage than we are that they are good or bad against zone coverage.


This article would not be complete without a mention of usage. At first blush, if we know that there are man-coverage beaters and zone-coverage beaters, and it is easier to predict who the man-coverage beaters are, defenses would be better off playing zone in an effort to muck things up statistically. That is way too simplistic of a view, though, as my former interns at PFF, Arjun Menon and Judah Fortgang found last year in the following article:

Should NFL teams be playing more man or zone coverage? | NFL News, Rankings and Statistics | PFF

Menon and Fortgang showed that man coverage occurs more frequently on high-leverage plays, and the distribution of outcomes is wider on man-coverage plays than zone coverage plays. They reasoned that if a defense has a mismatch against an offense in man coverage, they should press that edge even further and run a lot of man coverage thus exacerbating the expected outcome (a win for the defense). If not, they should run zone, which dampens the expected outcome (a loss for the defense).

This shows up in the Reception Perception data as the rate in which players face man and zone coverage (assumed complementary) is pretty stable from season to season:

Here, r = 0.66, rivalling the correlations seen in success rate overall and success rate in man coverage.


New data brings with it an opportunity to learn new things about the game. In some cases, new data helps us validate things we thought we already knew about the game.

Using Reception Perception data, we show that there is significant signal in how well a wide receiver gets open overall and especially against man coverage. While getting open against zone coverage is not nearly as stable season-to-season, it is not exactly noise, either. However, there is not a lot of crossover in ability versus man and zone, which means that we can classify players as man-coverage beaters and zone-coverage beaters with more confidence in the former than the latter.

This information is quite actionable, both on the team and the gaming side. To the degree that one’s scheme and players are versatile, a defensive coordinator can deploy schemes to enhance or dampen the delta in the ability between their players and that of their opponents. In fantasy football or gambling, a player can better predict the potential profile of a wide receiver’s opponent by classifying them as a man-coverage beater or a zone-coverage beater and their foe’s willingness to change their approach to match.


Related Posts

Using Text Analytics to Evaluate the 2024 NFL Draft Quarterback Class

April 17, 2024

The Tour of the Top Four: Caleb Williams

April 17, 2024

Need a Kicker? Don’t Draft One.

April 17, 2024

UFL Week 4: Updated Projections

April 16, 2024
Download our FREE 2024 NFL Rookie Guide with advanced analytics and future projections.
FREE 2024 NFL Rookie Guide