Writers of Pro Football Prospectus 2008

Most Recent FO Features

WayneReg01-1.jpg

» Audibles at the Line: Week 7

Week 7 features big comebacks for Buffalo and Detroit, big routs in Denver and Indianapolis, and big fat cojones for the St. Louis Rams special teams.

08 Nov 2012

Varsity Numbers: Let's Rate Some Receivers!

by Bill Connelly

If you have been reading Varsity Numbers for a while, you are probably (well, hopefully) familiar with the Adj. POE (Points Over Expected) measure, my attempt to quantify the point value a given ball-carrier brings to the table as compared to an average college back. It is not a perfect measure, but in adjusting for opponent and converting to a point value, I hope it is at least somewhat informative and relatable. We will revisit that measure (and, one day, figure out an easy way to share it from week to week) soon enough.

From the moment I came up with Adj. POE (it's been a while now), I wanted to come up with a way to create an equivalent measure for receivers. It is trickier, of course, since a) you need full targets data (i.e. both catches and the incomplete passes targeting you) and b) pass-catchers need somebody throwing them the ball. It is rather complicated, but it's time to take a step forward in that regard.

What I'm sharing below is very much a rough draft of whatever a receiver measure might end up becoming. Honestly, the end goal is probably an "Adj. POE" figure that equates to points the same as the running backs measure does, but I think I want some feedback first.

Below, you will find a measure that attempts to answer the following questions about a given pass-catcher:

1) How much do you produce?
2) How important are you to your team's passing game?
3) How good is the passing game to which you are important?
4) And how much is the forward pass featured in your team's offense?

All four of these questions are important. To begin attempting to answer them, I simply multiplied the following four measures together:

  • Yards Per Target. This is exactly what you think it is: total receiving yards divided by total targets. This attempts to answer question No. 1 above, telling you how much a receiver has produced with his given opportunities.
  • Target Rate. This is the percentage of a team's targets targeting a given player. This addresses Question No. 2. It is, in theory, easier to produce a large per-target average if you are a small piece of your team's passing game and an opponent isn't putting its No. 1 cornerback on you (or, really, paying you much mind).
  • Passing S&P+. This is the Passing S&P+ rating for a given player's team. Question No. 3 is tricky. The quality of the quarterback throwing you the ball matters, as does the quality of the line protecting the quarterback throwing you the ball. With infinite time and resources, we could come up with a "happy place" measure that splits credit between the line, quarterback, and receiving corps. We do not have that yet. This is a decent place-holder for such a measure.
  • Team Pass Rate. The ratio of a team's passes to its overall plays. Let's put it this way: Air Force's No. 1 receiver almost always has an incredible per-target average. This year, Falcons receiver Ty MacArthur has caught 20-of-28 passes for 372 yards; that's a 71 percent catch rate, and 13.3 yards per target. Air Force's No. 2 threat, Drew Coleman, has caught 12-of-17 balls for 305 yards, a 71 percent catch rate, and 17.9 yards per target. Those are fantastic averages, but ... of course they are. Air Force runs so much that opponents are probably playing the run even on third-and-long. Five Arizona receivers have been targeted more than MacArthur. Hell, six Michigan State receivers have been targeted at least 41 times, and the Spartans aren't even an incredibly pass-heavy team. The point, of course, is that if MacArthur and Coleman were facing the normal load of targets for a No. 1 or No. 2 receiver, on a team that actually passes a normal amount (or more), they probably wouldn't be averaging 13.3 and 17.9 yards per target, respectively. The inclusion of Team Pass Rate, then, is simply a safeguard against overrating big numbers from run-heavy schools. (Of course, this probably would have punished Georgia Tech's Demaryius Thomas in 2009 to a certain degree, but his numbers were probably absurd enough to overcome it.)

Below you will see the nation's top-50 receivers according to the product of these four measures, something I am tentatively calling "RYPR," which is simply (Target Rate x Yards Per Target x Passing S&P+ x Pass Rate). We'll worry about a fancy name later.

This is not intended to be the final version of a receiver rating tool, but I must say, the list it produces below is pretty damn good for a rough draft. My first requirement for any ranking of receiver probably would have been "By God, Marqise Lee better be first," and he is. But the entire list is strong. In need of improvement? Always. But it's a good first step.

Team Player Targets Catches Yards Catch
Rate
Yds Per
Target
Target
Rate
Target
No.
Passing
S&P+
Pass
Rate
RYPR Rk
USC Marqise Lee 121 88 1286 72.7% 10.6 39.9% 1 144.6 56.7% 348.0 1
Baylor Terrance Williams 102 71 1340 69.6% 13.1 33.3% 1 124.2 53.5% 291.0 2
Arkansas Cobi Hamilton 114 67 1063 58.8% 9.3 36.5% 1 134.1 55.9% 255.6 3
Clemson DeAndre Hopkins 85 62 1037 72.9% 12.2 27.8% 1 149.4 47.7% 241.4 4
Oregon State Brandin Cooks 67 50 906 74.6% 13.5 23.6% 2 122.8 56.1% 219.7 5
Tennessee Justin Hunter 98 56 838 57.1% 8.6 29.8% 1 156.6 53.8% 214.7 6
Boston College Alex Amidon 112 67 1073 59.8% 9.6 34.8% 1 103.9 61.1% 211.6 7
Arizona Austin Hill 83 59 994 71.1% 12.0 21.8% 2 138.9 55.6% 201.3 8
West Virginia Tavon Austin 109 84 896 77.1% 8.2 30.9% 1 128.2 60.5% 196.9 9
Oregon State Markus Wheaton 86 54 787 62.8% 9.2 30.3% 1 122.8 56.1% 190.8 10
West Virginia Stedman Bailey 77 61 830 79.2% 10.8 21.8% 2 128.2 60.5% 182.4 11
USC Robert Woods 86 59 661 68.6% 7.7 28.4% 2 144.6 56.7% 178.9 12
San Jose State Noel Grigsby 75 55 852 73.3% 11.4 23.5% 1 119.0 54.6% 173.5 13
Middle Tennessee Anthony Amos 84 58 846 69.0% 10.1 34.9% 1 116.2 42.4% 172.8 14
Vanderbilt Jordan Matthews 88 61 850 69.3% 9.7 40.0% 1 107.9 41.0% 171.0 15
Wisconsin Jared Abbrederis 51 37 675 72.5% 13.2 26.2% 1 123.1 38.8% 165.2 16
Oklahoma Kenny Stills 72 51 649 70.8% 9.0 23.5% 1 134.3 57.3% 162.7 17
Louisiana Tech Quinton Patton 109 74 991 67.9% 9.1 30.4% 1 115.7 50.3% 160.9 18
Central Michigan Titus Davis 71 40 821 56.3% 11.6 25.3% 2 99.4 55.3% 160.5 19
Texas A&M Mike Evans 86 56 802 65.1% 9.3 25.8% 1 130.9 50.4% 158.8 20
Team Player Targets Catches Yards Catch
Rate
Yds Per
Target
Target
Rate
Target
No.
Passing
S&P+
Pass
Rate
RYPR Rk
Texas Mike Davis 57 38 724 66.7% 12.7 23.6% 1 117.2 45.2% 158.5 21
Tulane Ryan Grant 75 42 834 56.0% 11.1 21.2% 1 101.3 66.2% 158.3 22
Pittsburgh Devin Street 68 53 730 77.9% 10.7 26.4% 1 115.6 48.0% 157.0 23
Alabama Amari Cooper 41 32 472 78.0% 11.5 20.4% 1 159.1 41.1% 153.6 24
Georgia Tavarres King 42 27 551 64.3% 13.1 17.2% 1 149.6 45.3% 153.0 25
Fresno State Davante Adams 100 71 907 71.0% 9.1 25.4% 1 117.2 55.5% 150.2 26
Tennessee Cordarrelle Patterson 66 36 585 54.5% 8.9 20.1% 2 156.6 53.8% 149.9 27
Kansas State Chris Harper 57 35 542 61.4% 9.5 29.4% 1 145.5 36.7% 149.1 28
Baylor Tevin Reese 59 37 682 62.7% 11.6 19.3% 2 124.2 53.5% 148.1 29
Kansas State Tyler Lockett 46 34 528 73.9% 11.5 23.7% 2 145.5 36.7% 145.3 30
Nebraska Kenny Bell 49 33 622 67.3% 12.7 20.9% 1 130.8 41.8% 144.7 31
New Mexico State Austin Franklin 118 64 1119 54.2% 9.5 38.8% 1 68.3 57.4% 144.3 32
Pittsburgh Mike Shanahan 58 41 669 70.7% 11.5 22.5% 2 115.6 48.0% 143.9 33
Syracuse Marcus Sales 78 49 710 62.8% 9.1 23.2% 1 128.1 52.8% 142.9 34
Oklahoma State Josh Stewart 75 56 670 74.7% 8.9 25.0% 1 124.0 51.2% 141.7 35
BYU Cody Hoffman 78 56 689 71.8% 8.8 28.0% 1 115.3 49.4% 140.7 36
Arizona Dan Buckner 86 54 678 62.8% 7.9 22.6% 1 138.9 55.6% 137.3 37
Ball State Willie Snead 105 68 901 64.8% 8.6 29.7% 1 104.2 50.5% 134.4 38
Duke Jamison Crowder 89 60 812 67.4% 9.1 23.3% 2 113.1 55.1% 132.5 39
Washington State Marquess Wilson 88 52 813 59.1% 9.2 20.0% 1 91.9 77.7% 132.1 40
Team Player Targets Catches Yards Catch
Rate
Yds Per
Target
Target
Rate
Target
No.
Passing
S&P+
Pass
Rate
RYPR Rk
Duke Conner Vernon 94 59 808 62.8% 8.6 24.6% 1 113.1 55.1% 131.8 41
Texas Tech Eric Ward 72 49 643 68.1% 8.9 18.6% 2 128.5 60.7% 129.7 42
Texas A&M Ryan Swope 60 45 655 75.0% 10.9 18.0% 2 130.9 50.4% 129.7 43
Ole Miss Donte Moncrief 59 39 540 66.1% 9.2 24.4% 1 123.7 47.0% 129.7 44
Washington Austin Seferian-Jenkins 67 48 632 71.6% 9.4 25.2% 2 103.6 52.5% 129.1 45
Miami (Ohio) Dawan Scott 60 39 630 65.0% 10.5 17.4% 3 113.7 61.7% 128.5 46
Georgia Marlon Brown 40 26 461 65.0% 11.5 16.4% 2 149.6 45.3% 128.0 47
East Carolina Justin Hardy 89 62 820 69.7% 9.2 25.9% 1 98.1 54.0% 126.3 48
Central Michigan Cody Wilson 72 55 644 76.4% 8.9 25.6% 1 99.4 55.3% 125.9 49
Vanderbilt Chris Boyd 56 35 622 62.5% 11.1 25.5% 2 107.9 41.0% 125.1 50


You can find the entire list of receivers with at least one catch here in a Google Doc.

Anyway, for a rough draft, I like this. But I'm very, very much open to suggestions and feedback. Thoughts?

National Averages

While we're playing with this data, I thought it would be interesting to take a look at some averages. All but three teams have completed passes to at least 10 targets, so below are some general averages for a given team's top-10 targets.

Target No. Avg.
RYPR
Yards Per
Target
Target
Rate
1 107.0 8.59 24.0%
2 76.4 8.17 17.9%
3 51.7 7.67 13.1%
4 42.0 7.84 10.4%
5 29.0 6.89 8.3%
6 24.2 7.26 6.5%
7 18.3 7.13 5.1%
8 14.5 7.18 4.0%
9 11.4 6.95 3.2%
10 8.5 6.99 2.5%

As you pore through the full data in the Google Doc provided above, keep some of this in mind. A typical No. 1 target sees about one-fourth of a team's passes, but the range is quite large, from Vanderbilt's Jordan Matthews (40.0 percent) and USC's Marqise Lee (39.9 percent) to Oregon's De'Anthony Thomas (14.1 percent) and UTSA's Earon Holmes (12.1).

Meanwhile, the average RYPR (can't wait to come up with a better name than that) for a No. 1 is around 107.0, but the range spans from Lee (348.0) and Terrance Williams (291.0) to New Mexico's Lamaar Thomas (13.0) and South Alabama's Bryant Lavender (27.1). Worst No. 1 target for a BCS conference team: Kansas' Kale Pick (45.1).

(In this case, "No. 1 target" literally just means "most-frequently targeted player," and at this point it in no way takes injuries and missed games into account. That is too difficult at this stage in the game. In the lengthy offseason To Do List, though, is a way to track injuries at least relatively well.)

This Week at SB Nation

It's been a fun week at the ol' SBN. It's featured one of my favorite Numericals and one of my favorite previews (Alabama-A&M).

Posted by: Bill Connelly on 08 Nov 2012

9 comments, Last at 29 Dec 2012, 3:35am by sox12

Comments

1
by Anon E. Mouse (not verified) :: Thu, 11/08/2012 - 7:40pm

How well do Adj. POE scores correlate with NFL performance?

2
by Anon E. Mouse (not verified) :: Thu, 11/08/2012 - 7:41pm

Or RYPR...

3
by Thok :: Thu, 11/08/2012 - 8:35pm

Or anything?

4
by Kal :: Thu, 11/08/2012 - 10:55pm

Neat!

My concern is simply multiplying four factors without doing any actual weighing. For instance, is a receiver 'better' because he's targeted a lot, or is he better for the number of actual targets that they have and how much that they do with them?

It also feels like it should have some kind of YAC measure and some combination of yards the pass travels.

5
by Bill Connelly :: Thu, 11/08/2012 - 11:59pm

Regarding the point about YAC ... that would be in here in a HEARTBEAT if it were possible to get that data for everybody. Alas, it's not.

6
by Michael (not verified) :: Fri, 11/09/2012 - 12:43pm

Ultimately, you are trying to measure a WRs ability to get open and catch the ball.
I reserve the right to be wrong, but I think you are over thinking this. The only real statistics that I think matter are the catch rate and yards per catch, which ultimately combine to an S&P like measure for a WR.

Obviously, there has to be some minimum thresholds so that a guy with a small number of targets and catches. on the other hand, it penalizes some very good WRs that happen to play in WR groups with a lot of depth. Quite frankly, if I had to figure out the group of WRs ball to throw to get a first down, I want the guys who had very high catch rates and very high ypc all RELATIVE the average yards per attempt and completion percentage allowed by the defenses they faced.

Marquis Lee is an insanely great player, but a big portion of his "numbers" are a function of the defenses he faces and the fact that USC has no real commitment to run the ball.

For all intents and purposes, he can't control the coverage he runs into, how often his QB chooses to target him verses other WRs nor can one control. Additionally, he can't control the overall passing effectiveness of the offense.

As for the running or passing rate, I could make strong arguments as to why the mesure is relevant or not. I have no idea if any of the following "logical" assertions are true, but...
1)the more a team passes the more likely it is because they don't have a good running game.
2)the more a team passes the more likely they are to have a good QB
3)the more a team runs, the more likely a WR and QB are likely to benefit from such ((i.e., look how crappy patterson and hunter come out - partly because UT is not a good running team and partly because they both drop passes or don't get open)

in an ideal world, one would have the s&P for times they are targeted and compare that to the other WRs on their team.

7
by Anonymous77 (not verified) :: Fri, 11/09/2012 - 1:53pm

When comparing this to the benchmark for RB's, what is considered a 'successful' +1 Adj. POE for WR's?

8
by mm(old) (not verified) :: Fri, 11/09/2012 - 5:04pm

I read 'RYPR' as 'Reaper', which sounds good for a defensive stat, not so good for an offensive stat.

A name like 'SCAMPER' would be better for a receiver stat ("That Marqise Lee, he's got a lot of SCAMPER!"). You'd need to come up with what the initials stand for.

9
by sox12 (not verified) :: Sat, 12/29/2012 - 3:35am

I'm sure you noticed this stat lines up very closely with the NFL Draft prediction sites, which is impressive given that there's no explicit strength of schedule component (although Pass S&P helps keep WAC WRs where they belong).

From a quick lazy small sample regression I ran, if your only goal were to maximize your correlation with the rankings on the draft sites, decreasing the weight of yards per target slightly (~^.9) would help. Cordarelle Patterson and Robert Woods are the most notable players they rank higher than you do, and both have underwhelming yards per target. They also aren't nearly as high on Brandin Cooks and his yards per target is very high.