WR Ratings

[ivanvamp]

All the Welker talk got me thinking about how to rate the quality of WR. I am 100% bonafide *not* a statistician, but I like to fiddle around with numbers. I have come up with four formulas for determining WR ratings. I won't get too into the formula, but here's the basic idea.

1. Productivity. This measures how, um, productive a receiver is. It takes into account receptions, yards, TD, first downs, and fumbles. Each item is weighted (thus a TD doesn't count the same as a reception, etc.). To give you an idea, here are a few guys and their Productivity Ratings. Note: there were 451 players that caught passes in the NFL this past year. The "AVG" is the average of all 451 players (this includes one reception by Joe Flacco).

AVG: 7.84
Calvin Johnson: 49.11
Wes Welker: 44.59
Deion Branch: 20.82
Mike Wallace: 31.53
Reggie Wayne: 26.10

2. Efficiency. This measures how efficient a player was in turning a reception into a yards, first downs, and touchdowns. So a player that has 1 rec, 20 yds, 1 td is more efficient than a player that has 5 rec, 20 yds, and 1 td. Here are those same players and their Efficiency Ratings:

AVG: 24.52
Calvin Johnson: 33.26
Wes Welker: 27.04
Deion Branch: 27.70
Mike Wallace: 31.58
Reggie Wayne: 25.73

3. Danger. This measures how dangerous a receiver was in turning a reception into a big play, so it takes into account TD/rec, ypc, YAC, etc. Again, the same players and their Danger Rating:

AVG: 19.62
Calvin Johnson: 27.96
Wes Welker: 21.36
Deion Branch: 23.25
Mike Wallace: 27.06
Reggie Wayne: 18.63

4. Total Receiver Rating. This simply adds up the three ratings, with productivity rating being weighted most heavily, followed by efficiency rating, and lastly, danger rating. The reason I weighted them this way is because a player could have just a handful of catches all year but they all go for big plays. Is that player a "better" receiver than a guy with a TON more receptions, yards, and TD, even if his "danger rating" is a little bit lower? Not in my book. But the Danger Rating is a factor. Again, the same guys:

AVG: 72.09
Calvin Johnson: 176.07
Wes Welker: 151.10
Deion Branch: 106.44
Mike Wallace: 137.49
Reggie Wayne: 109.41

I'm sure there are huge flaws in my metrics, but it at least gives me something to go on. Based on these admittedly flawed (but very fun) stats, here's the top ten guys in each category (min 48 rec):

Productivity
1. C. Johnson, Det - 49.11
2. Gronkowski, NE - 46.17
3. Welker, NE - 44.59
4. Graham, NO - 40.40
5. Nelson, GB - 39.13
6. White, Atl - 38.66
7. Cruz, NYG - 37.46
8. Fitzgerald, Ari - 36.21
9. Colston, NO - 31.93
10. Nicks, NYG - 31.92

Efficiency
1. Nelson, GB - 34.80
2. Jackson, SD - 34.12
3. C. Johnson, Det - 33.26
4. Cruz, NYG - 33.25
5. Wallace, Pit - 31.58
6. Gronkowski, NE - 31.51
7. Fitzgerald, Ari - 31.45
8. Smith, Car - 31.32
9. Robinson, Dal - 31.27
10. Jones, Atl - 31.20

Danger
1. Cruz, NYG - 30.10
2. Jones, Atl - 29.97
3. Nelson, GB - 29.91
4. Smith, Car - 28.16
5. Jackson, SD - 27.12
6. Wallace, Pit - 27.06
7. Gronkowski, NE - 26.32
8. Baldwin, Sea - 26.11
9. Foster, Hou - 25.66
10. Smith, Bal - 25.56

TOTAL RATING
1. C. Johnson, Det - 190.65
2. Gronkowski, NE - 176.07
3. Nelson, GB - 165.93
4. Cruz, NYG - 154.90
5. Welker, NE - 151.10
6. Fitzgerald, Ari - 147.49
7. Graham, NO - 144.31
8. Jackson, SD - 141.21
9. Wallace, Pit - 137.49
10. Smith, Car - 137.42

Here's the Pats' key receivers (Welker, Gronk, Hernandez, and Branch) and where they ranked in each category (again, minimum of 48 receptions to qualify):

Productivity
Welker - 44.59, #3
Gronkowski - 46.17, #2
Hernandez - 27.60, #23
Branch - 20.82, #49

Efficiency
Welker - 27.04, #35
Gronkowski - 31.51, #6
Hernandez - 25.20, #51
Branch - 27.70, #28

Danger
Welker - 21.36, #29
Gronkowski - 26.32, #7
Hernandez - 20.47, #35
Branch - 23.25, #19

Total Rating
Welker - 151.10, #5
Gronkowski - 176.07, #2
Hernandez - 113.47, #27
Branch - 106.44, #41

Again, very flawed number-crunching, but when I look at the lists and how the rankings came out, I think they're pretty much in line with what you'd think. The only interesting part was that I figured Hernandez would be more "dangerous" than Welker or Branch, but in reality, his numbers in those key areas weren't as impressive as I thought they would be, and Branch's were better than I thought they would be.

But as a whole group, man, four of the top 41 in the league (again, min 48 receptions)? Well, I guess it kind of makes sense given how prolific the Pats' passing offense was this season.

 

[ivanvamp]

Just for fun, I plugged in Randy Moss' 2007 season. Here's where he would be compared with this year's guys:

Productivity
1. C. Johnson, Det - 49.11
2. Gronkowski, NE - 46.17
3. Welker, NE - 44.59
4. Graham, NO - 40.40
5. Nelson, GB - 39.13
6. White, Atl - 38.66
7. Cruz, NYG - 37.46
8. Fitzgerald, Ari - 36.21
9. Colston, NO - 31.93
10. Nicks, NYG - 31.92
2007 Moss - 55.13

Efficiency
1. Nelson, GB - 34.80
2. Jackson, SD - 34.12
3. C. Johnson, Det - 33.26
4. Cruz, NYG - 33.25
5. Wallace, Pit - 31.58
6. Gronkowski, NE - 31.51
7. Fitzgerald, Ari - 31.45
8. Smith, Car - 31.32
9. Robinson, Dal - 31.27
10. Jones, Atl - 31.20
2007 Moss - 31.26

Danger
1. Cruz, NYG - 30.10
2. Jones, Atl - 29.97
3. Nelson, GB - 29.91
4. Smith, Car - 28.16
5. Jackson, SD - 27.12
6. Wallace, Pit - 27.06
7. Gronkowski, NE - 26.32
8. Baldwin, Sea - 26.11
9. Foster, Hou - 25.66
10. Smith, Bal - 25.56
2007 Moss - 22.28

TOTAL RATING
1. C. Johnson, Det - 190.65
2. Gronkowski, NE - 176.07
3. Nelson, GB - 165.93
4. Cruz, NYG - 154.90
5. Welker, NE - 151.10
6. Fitzgerald, Ari - 147.49
7. Graham, NO - 144.31
8. Jackson, SD - 141.21
9. Wallace, Pit - 137.49
10. Smith, Car - 137.42
2007 Moss - 179.42

Here's the Pats' key receivers (Welker, Gronk, Hernandez, and Branch) and where they ranked in each category (again, minimum of 48 receptions to qualify):

Productivity
Welker - 44.59, #3
Gronkowski - 46.17, #2
Hernandez - 27.60, #23
Branch - 20.82, #49
2007 Moss - 55.13

Efficiency
Welker - 27.04, #35
Gronkowski - 31.51, #6
Hernandez - 25.20, #51
Branch - 27.70, #28
2007 Moss - 31.26

Danger
Welker - 21.36, #29
Gronkowski - 26.32, #7
Hernandez - 20.47, #35
Branch - 23.25, #19
2007 Moss - 22.28

Total Rating
Welker - 151.10, #5
Gronkowski - 176.07, #2
Hernandez - 113.47, #27
Branch - 106.44, #41
2007 Moss - 179.42

Very interesting. I'm obviously surprised that Moss' Danger Rating is lower than 2011 Deion Branch's. The things that hurt Moss were his very low YAC per reception number. In 2007, Moss' average YAC per reception was just 2.9. In 2011, the *average* YAC per reception in the NFL was 5.4. Moreover, Moss had just 18 plays of 20+ yards in 2007, and just a 0.18 20+ plays per reception number, which is just slightly above the 2011 NFL average (0.16). So Moss, though he had numerous deep plays, wasn't as productive deep down the field as I had remembered him being. He made a lot of mid-range receptions where he was tackled almost immediately, and a lot of his TDs were in the end zone, which obviously limited his YAC and YAC/Rec numbers.

 

[jmt57]

Nice work. I'm glad you included both tight ends and wide receivers, as the distinction between the two becomes more and more blurred with each passing (no pun intended) year. I'm curious: would you consider sharing the complete, specific formulas you used with the rest of us?

 

[ivanvamp]

Nice work. I'm glad you included both tight ends and wide receivers, as the distinction between the two becomes more and more blurred with each passing (no pun intended) year. I'm curious: would you consider sharing the complete, specific formulas you used with the rest of us?

Sure, but I'm kind of embarrassed by the simplicity and utter capriciousness of them. Meaning that yes, I think they tended to produce results that look like they make sense (i.e., the top 10 list seems like it would fit what a reasonable evaluation of receivers in 2011 would look like), but they're hardly robust and based on anything other than what seemed reasonable to me.

TOTAL RATING
1. C. Johnson, Det - 190.65
2. Gronkowski, NE - 176.07
3. Nelson, GB - 165.93
4. Cruz, NYG - 154.90
5. Welker, NE - 151.10
6. Fitzgerald, Ari - 147.49
7. Graham, NO - 144.31
8. Jackson, SD - 141.21
9. Wallace, Pit - 137.49
10. Smith, Car - 137.42

I mean, yeah, that makes sense. Welker was a top-5 receiver this past year, but he wasn't as "good" as Johnson or Gronk or even Victor Cruz, who had just an incredible season.

But sure, here are the formulas, such as they are.....

PR (production rating) = ((10*rec)+yds+(100*td)+(10*1dn)+(-100*fum))/100

ER (efficiency rating) = 10*(rec%+1dn/rec)+ypc+(10*(td/rec))

DR (danger rating) = ypc+yac/rec+(10*20+/rec)+(10*(td/rec))

TR (total rating) = (2*PR)+(1.5*ER)+DR

rec = receptions
yds = yards
td = td
1dn = first downs gained
fum = fumbles
rec% = receptions / targets
1dn/rec = first downs / receptions
ypc = yards per catch
yac = yards after catch

Please, if anyone would like to refine these or make them worlds better, I would love it. This was just my way of trying to have a ratings system that seemed to reflect reality and didn't produce very strange numbers.

I'm sure the real numbers guys and statisticians here could improve this a whole lot. Go easy on me, guys!

 

[jmt57]

I like the formula as it gives a nice balance between quantity and quality, and takes into account pretty much any worthwhile metric I can think of. The names on both the overall rankings as well as the sub-rankings bear out that it does seem to be a pretty good way of looking at a receiver's performance and productivity.

 

[patfanken]

This was awesome, Ivan. Since a lot of people are taking about him, I'd be interested to seeing how Brandon Lloyd's numbers would look like. As a matter of fact, I'd be interested in seeing where the top ten FA WR's would fit into this formula you've developed.

Again, thanks a lot for the effort,.

 

[VrabelJr]

If you won't reveal your formula then this is useless. Which is a very supporting fact to your claim that you are not a statistician.

 

[PatsFanSince74]

...

...

But sure, here are the formulas, such as they are.....

PR (production rating) = ((10*rec)+yds+(100*td)+(10*1dn)+(-100*fum))/100

ER (efficiency rating) = 10*(rec%+1dn/rec)+ypc+(10*(td/rec))

DR (danger rating) = ypc+yac/rec+(10*20+/rec)+(10*(td/rec))

TR (total rating) = (2*PR)+(1.5*ER)+DR

rec = receptions
yds = yards
td = td
1dn = first downs gained
fum = fumbles
rec% = receptions / targets
1dn/rec = first downs / receptions
ypc = yards per catch
yac = yards after catch

...
I'm sure the real numbers guys and statisticians here could improve this a whole lot. Go easy on me, guys!

A lot of work and thought went into that. Thank you.

My quibble is that you are mixing yardage indicators ("yds", "ypc", "yac") with percentage indicators, which latter you then have to adjust by multiplying by a factor of 10 to make them comparable to yardage. That seems artificial to me.

Also, I could figure out what you mean in all of your formulas except for "(10*20+/rec)." How does that read out?

Otherwise, I'll have to think about it a bit. You must have a very good database on the Receivers. Could you give us the source for your raw data so we can play around with it a bit?

Thanks again.

 

[PatsFanSince74]

If you won't reveal your formula then this is useless. Which is a very supporting fact to your claim that you are not a statistician.

Well, "useless" is a little harsh dontcha think? Why not just ask him nicely for it like JMT did?

 

[jmt57]

If you won't reveal your formula then this is useless. Which is a very supporting fact to your claim that you are not a statistician.

The guy with post #7 may want to read post #4

 

[patsfan13]

Nice approach. A couple of things this approach would indicate. Branch is a bargain as the #4 option in the passing game. and would be very valuable as the #5 if we sign someone like Lloyd. Looking at B Lloyd's season when playing in this system (2010) would be interesting.

 

[ivanvamp]

A lot of work and thought went into that. Thank you.

My quibble is that you are mixing yardage indicators ("yds", "ypc", "yac") with percentage indicators, which latter you then have to adjust by multiplying by a factor of 10 to make them comparable to yardage. That seems artificial to me.

Also, I could figure out what you mean in all of your formulas except for "(10*20+/rec)." How does that read out?

Otherwise, I'll have to think about it a bit. You must have a very good database on the Receivers. Could you give us the source for your raw data so we can play around with it a bit?

Thanks again.

There are plenty of things to quibble with - and some things that would merit a term stronger than "quibble". I won't even for a minute try to "defend" these formulas as if they are really objectively accurate. It was, as I've said, just my way of trying to capture what receivers have done, and broken it up into different categories. That said, as I've said, I do think, when the numbers shake out, that it *looks* like what you'd suspect a really accurate listing would probably look like. So I'm probably not light years away from something useful.

To your "what is (10*20+/rec)" question, it's this... 20+ is the number of 20+ yardage pass plays. I divided that by the number of receptions. So take these three guys:

C.Johnson: 96 rec, 32 20+ passing plays, 0.33 20+/rec
Gronkowski: 90 rec, 22 20+ passing plays, 0.24 20+/rec
Welker: 122 rec, 21 20+ passing plays, 0.17 20+/rec

Make sense?

I can definitely send the group the excel spreadsheet I put together. I don't know how to upload it, though. If someone can give me the instructions for that I'd be happy to get it out there.

 

[ivanvamp]

If you won't reveal your formula then this is useless. Which is a very supporting fact to your claim that you are not a statistician.

You might want to look upthread just a few posts.

I don't always read every post in a thread before I throw in my two cents, but generally that's on threads that are already pretty long. You posted what I quoted above in the 7th post of the thread and I had given my formulas in post #4. So.........might be worth a look-see.

But yes, it's very true that I'm not a statistician.

 

[ivanvamp]

This was awesome, Ivan. Since a lot of people are taking about him, I'd be interested to seeing how Brandon Lloyd's numbers would look like. As a matter of fact, I'd be interested in seeing where the top ten FA WR's would fit into this formula you've developed.

Again, thanks a lot for the effort,.

I decided to update the lists with 32 being the minimum number of receptions instead of 48. The thought being that 2 rec a game still makes a guy a receiver that QBs throw to semi-regularly. Anyway, another arbitrary number, but hey, it's what it is.

Here's Brandon Lloyd, 2010 and 2011 (to satisfy both of these questions):

2011:
PR: 25.26 (#33 in NFL) - 2010: 40.38
ER: 25.75 (#68 in NFL) - 2010: 34.62
DR: 18.77 (#84 in NFL) - 2010: 25.79
TR: 107.92 (#44 in NFL) - 2010: 158.48

So Lloyd's 2010 season would have had him ranked here for 2011:

PR: #5
ER: #3
DR: #10
TR: #4

No matter how you slice it, that's an elite receiver.

Later today I'll do this for all of the top free agents and give you where they stack up.

 

[MoLewisrocks]

Cursory as it may be, it sure does seem to indicate that Branch isn't yet road kill as some here seem determined to make him out to be and Welker is a top 5 WR overall, so while factoring in his impending AARP status as a soon to be 31 year old you might not really want to pay him as such for another 5 years. But top ten payment should be impossible to quibble with. FWIW that is presently calculated to have been $8.5M (this year's transition tag #) per over the last 5 years...and that # will undoubtedly increase over the next 3-4 seasons as new deals are done and the impending cap increase approaches. And that's before you even apply some formula that weights value to this team based on proven production in a complex system many otherwise accomplished and highly drafted WR's can't manage to grasp.

 

[vyrago]

If you won't reveal your formula then this is useless. Which is a very supporting fact to your claim that you are not a statistician.

How about a little respect for the work? Look at an earlier post and I think you'll find part of the formula.

 

[PatsFanSince74]

You might want to look upthread just a few posts.

I don't always read every post in a thread before I throw in my two cents, but generally that's on threads that are already pretty long. You posted what I quoted above in the 7th post of the thread and I had given my formulas in post #4. So.........might be worth a look-see.

But yes, it's very true that I'm not a statistician.

You were a lot more polite to him than he deserved.

 

[patsfan13]

Here's Brandon Lloyd, 2010 and 2011 (to satisfy both of these questions):

2011:
PR: 25.26 (#33 in NFL) - 2010: 40.38
ER: 25.75 (#68 in NFL) - 2010: 34.62
DR: 18.77 (#84 in NFL) - 2010: 25.79
TR: 107.92 (#44 in NFL) - 2010: 158.48

So Lloyd's 2010 season would have had him ranked here for 2011:

PR: #5
ER: #3
DR: #10
TR: #4

No matter how you slice it, that's an elite receiver.

Later today I'll do this for all of the top free agents and give you where they stack up.

I am sure the method is not perfect, but I appreciate the effort put into this, and look forward to the next installment.

Lots of off season fun.

Another observation, the composite for the top receivers will reflect of course the quality of the QB, IOW Brady has a lot of top ranked targets, one suspects that Rogers and Breeze do also.

I wonder if you can identify an elite target who works with an JAG QB by how much higher they rank than their peers. I have seen pitchers who had very good W/L record when their team was lousy, like Steve Carlton for example. I would guess Welker was ranked far higher than other Dolphin Targets when he played there.

My guess is a fair amount of separation between Lloyd and Royal on the 2010 Bronco's whereas Wallace has Brown, Miller, Sanders and Ward as targets.

 

[MaineMan]

There are plenty of things to quibble with - and some things that would merit a term stronger than "quibble". I won't even for a minute try to "defend" these formulas as if they are really objectively accurate. It was, as I've said, just my way of trying to capture what receivers have done, and broken it up into different categories. That said, as I've said, I do think, when the numbers shake out, that it *looks* like what you'd suspect a really accurate listing would probably look like. So I'm probably not light years away from something useful.

To your "what is (10*20+/rec)" question, it's this... 20+ is the number of 20+ yardage pass plays. I divided that by the number of receptions. So take these three guys:

C.Johnson: 96 rec, 32 20+ passing plays, 0.33 20+/rec
Gronkowski: 90 rec, 22 20+ passing plays, 0.24 20+/rec
Welker: 122 rec, 21 20+ passing plays, 0.17 20+/rec

Make sense?

I can definitely send the group the excel spreadsheet I put together. I don't know how to upload it, though. If someone can give me the instructions for that I'd be happy to get it out there.

Excellent work!

I kinda like the separate performance categories and especially like the concept of evaluating things like 20-yard-plays and 1st-downs as a percentage of plays, rather than as a raw number.

Inspired by the pro-football-reference.com Average Value (AV) system, I've been trying to develop something like you have - not so much for ranking purposes as for trying to get a feel for the average career value of players compared to what they "cost" in terms of draft pick value (as per the Standard Trade Value chart). IOW, trying to answer the question, "What has been the typical difference in ROI between guy taken in the late first (around 640 ATV) versus a guy taken in the late second (around 280 STV) or a guy taken in the late 3rd (around 120 STV). In the long run, I was also trying to rate the relative strength (in terms of ROI) of each draft class by position.

I started with RBs, which is instantly way more complicated when trying to factor in receiving as well as rushing (and I haven't even begun to address KR/PR contributions). So, my system seems a bit more elaborate than yours - which is not to say that it's any more informative.

Anyway, some of the issues I've been wrestling with:
- fumbles should be a performance deduction, it seems, and a lost fumble should maybe be a bigger deduction - but how much bigger?
- catch rate (CR) should probably figure in somewhere, but perhaps needs to be balance against YPC since guys with higher YPC typically have lower CR than guys (like RBs, TEs and slot receivers) who are most often much closer to the LoS when targeted. HOWEVER, this seems even further complicated (in my always-over-thinking-it brain) by the fact that the YPC of some players is supported by relatively high YAC.

So, in addition to being a compliment, I guess this is sort of a warning about the danger of falling down the rabbit hole.

 

[patsfan13]

Excellent work!

I kinda like the separate performance categories and especially like the concept of evaluating things like 20-yard-plays and 1st-downs as a percentage of plays, rather than as a raw number.

Inspired by the pro-football-reference.com Average Value (AV) system, I've been trying to develop something like you have - not so much for ranking purposes as for trying to get a feel for the average career value of players compared to what they "cost" in terms of draft pick value (as per the Standard Trade Value chart). IOW, trying to answer the question, "What has been the typical difference in ROI between guy taken in the late first (around 640 ATV) versus a guy taken in the late second (around 280 STV) or a guy taken in the late 3rd (around 120 STV). In the long run, I was also trying to rate the relative strength (in terms of ROI) of each draft class by position.

I started with RBs, which is instantly way more complicated when trying to factor in receiving as well as rushing (and I haven't even begun to address KR/PR contributions). So, my system seems a bit more elaborate than yours - which is not to say that it's any more informative.

Anyway, some of the issues I've been wrestling with:
- fumbles should be a performance deduction, it seems, and a lost fumble should maybe be a bigger deduction - but how much bigger?
- catch rate (CR) should probably figure in somewhere, but perhaps needs to be balance against YPC since guys with higher YPC typically have lower CR than guys (like RBs, TEs and slot receivers) who are most often much closer to the LoS when targeted. HOWEVER, this seems even further complicated (in my always-over-thinking-it brain) by the fact that the YPC of some players is supported by relatively high YAC.

So, in addition to being a compliment, I guess this is sort of a warning about the danger of falling down the rabbit hole.

Good article on turnovers:

Coaching, the Super Bowl, and the wealth of turnovers | Cold Hard Football Facts

Turnovers more often than not are an express connection to the gain or loss of points. Given that NFL kickers are now very reliable inside 50 yards, a team that provides a turnover once it has attained a first-down at its opponent’s 40-yard line can expect to lose at least three points because that team could expect to move into reasonable field goal range without attaining any more first downs.

Likewise, a team that receives a turnover inside its opponent’s 40-yard line can expect to gain three points without achieving even one first down.

Typically, one only thinks of 20 percent of the field—the “Red Zone”—as scoring territory. But turnovers transform 80 percent of the field into scoring territory in that when a turnover occurs a team is likely to either lose points or gain points without any further meaningful progress (i.e., a first down) unless the turnover occurs in the middle 20 percent of the field between the 40-yards lines.