Sunday, December 19, 2010

Are short-passing QBs the secret to NFL success?

By Steven Buzzard

After watching the Pats dismantle yet another team (as well as moving to #1 on advancednflstats offense ranking) I was curious if more teams should use a dink and dunk passing attack like the Pats have used almost exclusively this year. Could other teams pull off a similar feat? Does the Pats offense go against the idea that deep passes aren't being used enough despite their higher EPA? Is an accurate QB like Brady the most important ingredient? Or are quick WR's like Welker and Woodhead required so they can get a lot of YAC?

First of all I wanted to verify that the Pats were in fact throwing as short as it seemed which they were ranking 3rd lowest in deep%. Next I wanted to figure out what was really making the offense go. But there weren't really stats out there that took into consideration depth of pass. So I tried to create a few new stats that would adjust for scheme.

For a QB their goal is to throw the ball to a specific spot on the field and while it's the receiver’s job to catch it several studies have shown comp% travels with a QB pretty well between teams if they move to a new team. So to test a QB's accuracy I used a simple regression on Deep% to try to determine an expected completion %. I subtracted this amount from the player’s actual completion % to get what I would call comp% added (similar to EPA). See a complete listing of all QB's at the end of the article.
The top ten looks like a pretty good who's who of QB's and I think correlates a lot more with talent the a top 10 of simply using comp%. Brady rates pretty high in this stat but he does take a hit from his standard completion % ranking. Completion % correlates highly with success rate which we know is an important variable in why teams win. But additionally we know AYPA is a key factor in why teams win.

To come up with an adjusted AYPA I started by researching how INT% correlated to passing depth and there didn't seem to be any. So instead I simply focused on an adjusted YPA and any INT adjustments could be made after the fact. Again, I ran a regression using Comp% and deep% to determine an expected YPA. I will call this QB YPA since it is measuring how far and how accurately they are throwing the ball. This stat would essentially be similar to air yards/att+league avg YAC/att (adjusted for pass depth).

1 9-T.Romo 6.77
2 9-D.Brees 6.53
3 17-P.Rivers 5.84
4 18-P.Manning 5.60
5 12-A.Rodgers 5.18
6 10-E.Manning 4.76
7 9-D.Garrard 4.15
8 3-J.Kitna 3.79
9 12-T.Brady 3.67
10 7-B.Roethlisberger 3.64
11 5-J.Flacco 3.43
12 7-M.Vick 3.16
13 10-V.Young 3.12
14 12-C.McCoy 2.31
15 2-M.Ryan 1.96
16 4-K.Kolb 1.94
17 6-J.Cutler 1.24
18 8-M.Schaub 0.42
19 17-J.Delhomme (0.11)
20 4-B.Favre (0.19)
21 14-Sh.Hill (0.34)
22 7-M.Cassel (0.47)
23 7-C.Henne (0.71)
24 8-M.Hasselbeck (1.00)
25 8-K.Orton (1.02)
26 9-C.Palmer (1.46)
27 5-J.Freeman (1.94)
28 8-J.Campbell (2.08)
29 14-R.Fitzpatrick (2.16)
30 5-K.Collins (2.40)
31 5-D.McNabb (3.19)
32 11-A.Smith (3.45)
33 8-S.Bradford (3.70)
34 3-M.Moore (3.75)
35 1-T.Smith (5.86)
36 6-M.Sanchez (7.00)
37 3-D.Anderson (7.22)
38 5-B.Gradkowski (7.58)
39 2-J.Clausen (11.86)

Interestingly, Tom Brady takes a big hit in this stat dropping from 7th in standard YPA to 21st. The thing I like about this formula is you can also use the residual between QB YPA and actual YPA to see how much the WR's are contributing to YPA above and beyond a league average WR in regards to YAC given the same passing depth. You could really break this down for individual WR's as well which would be interesting. I'm not sure what I would call this stat but you can find a list of how much WR's are contributing to each teams YPA in terms of YAC at the end of the article. Some might argue that the QB hitting a WR in stride greatly contributes to YAC but the QB's at the top of the list don't seem to correlate to QB talent at all. It seems like a random grouping of very good QB's (Vick, Brady, and Rivers) and poor ones that can't keep jobs and rate low on the Comp%added stat (Troy Smith, McNabb, Alex Smith).

Finally, I used some of these thoughts to try to allocate some of the credit for the Pats stellar 0.28EPA/P. In general for every increase in 1 of AYPA it increases EPA/P increased by .102. Additionally every 1% increase in Int% decreases AYPA by .68. Using these ratios we could estimate that the WR YAC above league average accounts for approximately (.89*.102)=.09 (1/3rd) of the Pats passing EPA/P and their positive INT% of 2% (0.9% vs. 2.9% league average) accounts for about (2*.68*.102)=0.14 (about half) of the total. Since the Pats QB YPA rating is exactly league average they aren't creating any of their advantage from this measure. This leaves approximately .05 (about 1/6th) for sack rates (which they are above average at equal to about .1 YPA) and the general passing premium since the average pass attempt has a positive EPA.

1 10-V.Young 8.24
2 7-B.Roethlisberger 7.98
3 17-P.Rivers 7.87
4 10-E.Manning 7.86
5 12-A.Rodgers 7.80
6 5-J.Flacco 7.79
7 18-P.Manning 7.76
8 9-D.Brees 7.55
9 9-T.Romo 7.53
10 7-M.Vick 7.53
11 9-D.Garrard 7.48
12 3-J.Kitna 7.30
13 6-J.Cutler 7.27
14 2-M.Ryan 7.25
15 12-C.McCoy 7.24
16 4-K.Kolb 7.20
17 8-K.Orton 7.16
18 8-J.Campbell 7.14
19 7-M.Cassel 7.12
20 4-B.Favre 7.12
21 12-T.Brady 7.10
22 3-M.Moore 7.00
23 8-M.Hasselbeck 6.98
24 8-M.Schaub 6.97
25 5-J.Freeman 6.94
26 7-C.Henne 6.92
27 14-Sh.Hill 6.91
28 1-T.Smith 6.88
29 14-R.Fitzpatrick 6.88
30 5-K.Collins 6.82
31 17-J.Delhomme 6.78
32 9-C.Palmer 6.74
33 3-D.Anderson 6.61
34 5-D.McNabb 6.61
35 6-M.Sanchez 6.39
36 5-B.Gradkowski 6.36
37 11-A.Smith 6.31
38 8-S.Bradford 6.09
39 2-J.Clausen 5.31

So while it is difficult to say for sure how likely it would be for a team to replicate the NE passing game using short passing attacks here are some take a ways. First of all, I’m not sure given how heavily their INT% is bumping their EPA/P that even the Pats can keep it going at this rate since it is accounting for about half their +EPA/P. Brian and several others have shown that almost all QB's tend to regress to an average of about 3% over time and that INT's seem to be due more to situation than anything else. Although maybe quick WR’s coupled with a good OL and an accurate and quick release QB is a good way to keep INT% down. However, with the WR YAC seemingly accounting for about 1/3rd of the total and most of the WR's on the team such as Branch and Woodhead coming for very cheap this may be the best route to go to replicate the success. I'm not sure how many WR's in the league are as quick as those guys but I would guess they are easier to find than a QB like Brady. It will be interesting to see if other teams try to mimic this type of passing game given how much of a copy cat league this is or if they will look at the success of deep passing and try to attack more that way.

1 1-T.Smith 1.23
2 7-M.Vick 1.02
3 17-P.Rivers 0.94
4 12-T.Brady 0.89
5 6-J.Cutler 0.57
6 5-D.McNabb 0.55
7 8-M.Schaub 0.52
8 12-C.McCoy 0.44
9 11-A.Smith 0.41
10 5-B.Gradkowski 0.34
11 8-J.Campbell 0.33
12 12-A.Rodgers 0.22
13 8-K.Orton 0.17
14 3-J.Kitna 0.15
15 6-M.Sanchez 0.08
16 7-B.Roethlisberger 0.05
17 9-T.Romo 0.01
18 5-J.Freeman 0
19 7-C.Henne 0
20 8-S.Bradford -0.01
21 7-M.Cassel -0.05
22 9-D.Garrard -0.1
23 2-J.Clausen -0.11
24 8-M.Hasselbeck -0.12
25 4-B.Favre -0.15
26 10-V.Young -0.2
27 9-D.Brees -0.2
28 14-R.Fitzpatrick -0.22
29 3-D.Anderson -0.3
30 5-J.Flacco -0.31
31 9-C.Palmer -0.33
32 4-K.Kolb -0.44
33 2-M.Ryan -0.62
34 10-E.Manning -0.62
35 14-Sh.Hill -0.62
36 18-P.Manning -0.7
37 5-K.Collins -0.92
38 17-J.Delhomme -0.92
39 3-M.Moore -1.01


James said...

I remember reading somewhere that YAC correlates better with wide receivers than with QBs as it appears to follow them between teams (similar to QBs and completion %). I don't remember where I read that or the details of how they came to that conclusion, but I thought it was worth sharing.

Buzz said...

Yes after re-reading up on YAC it does correlate better with WR's. It is what I thought but couldn't 100% remember at the time. Most famously they go through this correlation in pro football prospectus. Cassel was a poster child for this in his move to KC.

Post a Comment

Note: Only a member of this blog may post a comment.