Writers of Pro Football Prospectus 2008

Most Recent FO Features

JohnsonKer18.jpg

» Seventh Day Adventure: Week 13

The biggest game this week is the Iron Bowl, where the playoff hopes of Alabama, Auburn, and Georgia hang in the balance.

19 Dec 2016

Week 15 Quick Reads

by Vincent Verhei

The average quarterback in the NFL in 2016 is completing 63.1 percent of his passes, while throwing an interception just 2.2 percent of the time. Both would be the best marks for quarterbacks in league history. Our average 2016 passer is also averaging 7.20 yards per pass -- down a bit from 2014 (7.21) and 2015 (7.25), but otherwise better than every season since 1960*. He is throwing a touchdown on 4.2 percent of his passes, which is lower than the 5-plus percent rates of the '50s and '60s, but higher than the sub-4 percent rates of the early '90s. By any objective measure, the NFL has never had better overall passing than it has in recent seasons.

* NOTE: This sentence was originally published with slightly incorrect data. It has been fixed.

Why is it, then, that so many critics of the NFL say the biggest problem in the league is the proliferation of bad quarterbacks? CBS analyst Bart Scott, a former linebacker with the Ravens and Jets, has said as much publicly: "It's either the haves or have-nots, and listen, there's never been so many bad quarterbacks in the NFL -- ever." Scott actually said that in November of 2015, but it still applies this season. Scott specifically cited Blaine Gabbert as an example of a terrible starting quarterback, and though Gabbert has since been rightfully returned to the bench, his replacement Colin Kaepernick has hardly fared any better. The league still has enough Blake Bortles- and Ryan Fitzpatrick-types (to say nothing of the messes in Los Angeles and Houston) that any random NFL game seems likely to feature at least one quarterback who needs to be replaced.

So we have these two viewpoints, which seem to conflict with each other and yet both seem to be valid. So which is more accurate, the objective viewpoint, or the subjective?

Fortunately we have DVOA, a tool that is uniquely qualified to measure this sort of thing. One of the key features of DVOA is that it compares each quarterback to the average baselines of that specific season. For example, Eli Manning has completed 63.4 percent of his passes this year. That would have been the highest such rate in the NFL in 1990, but this year it is just 18th. As such, Manning is punished more for his incompletions this year than he would have been 26 years ago.

We have DVOA data for quarterbacks going back to 1989. Typically in that timespan, about 30 percent of all quarterbacks in a given season have had a DVOA of -10.0% or worse, and about 30 percent have had a DVOA of 10.0% or higher, with the remaining 40-ish percent falling in the middle. Looking at the data for each of those seasons, though, we find a changing trend that has been subtle, but very consistent:


Good, Bad, and Average QB Distribution, 1989-2016
Year Bad QBs
(DVOA Worse
Than -10%)
Average QBs
(DVOA between
-10% and 10%)
Good QBs
(DVOA Better
Than 10%)
Total
1989 8 15 8 31
1990 9 10 10 29
1991 8 14 11 33
1992 10 8 12 30
1993 9 14 11 34
1994 6 18 8 32
1995 10 8 13 31
1996 7 16 12 35
1997 7 17 9 33
1998 11 10 13 34
1999 8 20 11 39
2000 13 17 10 40
2001 6 18 7 31
Year Bad QBs
(DVOA Worse
Than -10%)
Average QBs
(DVOA between
-10% and 10%)
Good QBs
(DVOA Better
Than 10%)
Total
2002 10 15 11 36
2003 9 18 9 36
2004 13 11 12 36
2005 9 15 12 36
2006 9 15 11 35
2007 15 11 12 38
2008 7 16 11 34
2009 14 7 13 34
2010 8 15 11 34
2011 11 13 11 35
2012 13 13 12 38
2013 14 14 11 39
2014 8 20 9 37
2015 12 18 7 37
2016* 13 12 10 35
AVERAGE 9.9 14.2 10.6 34.7
* 175 passes needed to qualfy in 2016

The trend may not be obvious in that table, but if you read it very closely, you'll note that the numbers of quarterbacks in the "good" column have stayed about the same, while the numbers in the "bad" column have been generally growing. From 1989 to 2003, there were at least as many good quarterbacks as bad 14 times in 15 seasons. The one year in that span when bad quarterbacks outnumbered good was in 2000 -- a unique year, as we shall get to shortly.

Since that 2003 season, though, things have changed. There have been more bad quarterbacks than good in eight of the 13 seasons since then, and four of the last five (both of these statistics include 2016, the numbers for which obviously are not final). The worst differential was set last season, when there were 12 bad quarterbacks and only seven good ones. With 13 bad passers and 10 good, 2016 would tie 2000, 2007, and 2013 as next-worst seasons on record when it comes to lousy quarterback differential.

So no, your eyes have not been deceiving you. In recent seasons, there really have been more bad quarterbacks across the NFL, with fewer good quarterbacks to balance them out. It's not crystal-clear exactly when this trend began, but it's probably not coincidental that the league expanded from 28 teams to 30 in 1995, then again to 31 in 1999, and finally to 32 in 2002. As pro football has grown, the NFL (and NCAA) have failed to produce a talent pool to grow along with it.

It is interesting to note what has happened in the Super Bowl after those seasons heavily slanted towards bad quarterback play:

  • In 2000, the Ravens had an all-time great defense that was capable of carrying Trent Dilfer to a championship over another mediocre quarterback in Kerry Collins.
  • In 2007, a record-setting Patriots offense was upset by a Giants team known more for its defense than for its offense.
  • The Seahawks' win over Denver in 2013 was hardly an upset like the Giants' win, but it was similar in that a record-setting offense was clearly outplayed by the opposing defense.
  • Finally, in 2015, we had a Super Bowl that was somewhat similar to that 2000 game, with an all-defense, no-offense champion (Denver) beating a more balanced opponent (Carolina).

That's a pretty random and obscure pattern, but should it hold true, it might be good news for Pittsburgh, Seattle, Baltimore, Denver, and the Giants -- this year's remaining Super Bowl contenders who are strongest on defense.

Quarterbacks
Rk
Player
Team
CP/AT
Yds
TD
INT
Sacks
Total
DYAR
Pass
DYAR
Rush
DYAR
Opp
1.
Drew Brees NO
37/48
389
4
0
3
187
187
0
ARI
With three minutes to go in the third, Brees got the ball in a tie game. That didn't last long -- he completed eight passes in a row on the drive, for 65 total yards, and Tim Hightower finished the drive with a 3-yard touchdown to put New Orleans ahead. Arizona came back and tied the game on their next drive, but no worries -- Brees was perfect again, completing five passes in a row, each for a first down, for 68 more yards and another go-ahead score. Between the two drives, that's 13 straight completions for 133 yards and two go-ahead scores in less than 15 minutes of game time.
2.
Matt Ryan ATL
17/23
286
2
0
1
176
176
0
SF
Ryan completed 74 percent of his passes, which is outstanding, but what's even more impressive is that he made those completions count. About 55 percent of all completions have gone for a first down this year. Only of of Ryan's completions, though, came up short of the sticks -- and that was a 4-yard gain on second-and-7. Put another way: The all-time record for passing first downs in a game is 29, set by the Giants against Cincinnati in 1985. It took them 40 completions and 62 attempts to get there. Ryan had half as many first downs on barely 40 percent as many completions and barely one-third as many attempts.
3.
Andrew Luck IND
21/28
250
2
0
0
159
162
-3
MIN
Late in the first quarter, when the Colts led just 3-0, they had a first-and-goal at the 3 and a chance to open a big lead. Luck then threw incompletions on first, second, and third down, and the Colts kicked a field goal. However, a Minnesota penalty took that field goal off the board, and the Colts added a rushing touchdown shortly thereafter. That sequence seemed to get all the bad football out of Luck's system -- from that point forward, he went 15-of-17 for 188 yards and two touchdowns.
4.
Matt Moore MIA
13/18
236
4
1
1
122
122
0
NYJ
It was feast or famine for Moore on third downs, with four conversions in nine attempts. The four conversions went for 71 total yards and three touchdowns, plus a 17-yard gain on third-and-6. The failures, though, consisted of three incomplete passes, one sack, and one interception. Another note of weirdness: Moore had only two passes on second down, a 7-yard completion, and an incomplete pass.
5.
Dak Prescott DAL
32/36
279
0
0
3
114
99
14
TB
Prescott's lob to Dez Bryant downfield just before halftime was notable. Not for what it achieved -- it was incomplete -- but for what followed. That was the last incompletion Prescott threw all day. He completed each of his last 12 passes, for 90 total yards.
6.
Ben Roethlisberger PIT
21/35
286
1
0
1
112
112
0
CIN
Roethlisberger hit a barrage of deep balls against Cincinnati, going 6-of-11 for 172 yards, with a 12th deep throw resulting in a DPI for 16 more yards.
7.
Matt Barkley CHI
30/43
362
2
3
1
103
103
0
GB
One of Barkley's interceptions came on a Hail Mary at the end of the first half, so we count that as an incompletion for DVOA and DYAR purposes. That was not the only deep failure Barkley had on the day though -- he threw seven passes that traveled more than 20 yards past the line of scrimmage, with only one caught by the Bears (a 34-yard catch by Cameron Meredith), but three caught by the Packers.
8.
Carson Palmer ARI
28/40
320
2
0
1
101
101
0
NO
Palmer was nearly perfect in short yardage. With 5 yards or less to go for a first down, he went 6-of-7 for 51 yards and six conversions, including a touchdown.
9.
Tom Savage HOU
23/36
260
0
0
0
95
96
-1
JAC
10.
Cam Newton CAR
22/37
300
2
0
2
68
77
-10
WAS
11.
Russell Wilson SEA
19/26
229
3
1
2
57
60
-3
LARM
12.
Tom Brady NE
16/32
188
0
0
2
48
48
0
DEN
Brady's first seven dropbacks resulted in six incompletions and a fumble-sack. Somehow the game was still tied at 3 at that point, and from there Brady played much better. He was second-worst in first-quarter DYAR this week, but ninth-best after that.
Rk
Player
Team
CP/AT
Yds
TD
INT
Sacks
Total
DYAR
Pass
DYAR
Rush
DYAR
Opp
13.
Matthew Stafford DET
24/39
273
0
1
1
45
44
1
NYG
Stafford had almost no luck in scoring range. Inside the Giants' 30 he went 3-of-8 for 9 yards with an interception and a sack. His only first down came on a 6-yard DPI.
14.
Philip Rivers SD
17/30
206
2
1
3
42
42
0
OAK
Each of Rivers' first six dropbacks of the second half resulted in a completion for a first down, gaining 74 total yards in the process. But Rivers wouldn't pick up any first downs after that, going 2-of-6 for 9 yards with an interception and two sacks.
15.
Eli Manning NYG
20/28
201
2
0
2
40
40
0
DET
16.
Kirk Cousins WAS
32/47
315
0
1
1
22
15
7
CAR
17.
Tyrod Taylor BUF
17/24
174
1
0
1
20
30
-10
CLE
The Bills got the ball with about eight minutes left in the third quarter, up 17-10 in what was still anyone's game. From that point forward, Taylor went 7-of-7 for 69 yards, with five first downs.
18.
Aaron Rodgers GB
19/30
252
0
0
4
19
22
-3
CHI
19.
Andy Dalton CIN
16/27
157
0
1
1
15
7
8
PIT
The Bengals were ahead 20-9 at halftime in this game, then still led 20-12 after the Steelers kicked a field goal on the first drive of the second half. From that point forward, Dalton went 6-of-11 for just 41 yards and only two first downs, with a sack.
20.
Marcus Mariota TEN
19/32
241
0
1
1
14
14
0
KC
21.
Robert Griffin CLE
17/28
196
0
0
5
0
-8
8
BUF
Griffin and the Browns hardly even tried to test the Bills deep until this game was out of reach. Only one of his first-half passes traveled more than 7 yards past the line of scrimmage: a 59-yard bomb to Corey Coleman on the last play of the half. Then he threw only three deep passes in the second half, even though the Browns were down by at least 14 points the entire time.
22.
Joe Flacco BAL
16/29
206
2
1
3
-4
-4
0
PHI
Flacco was Griffin in reverse. He threw only four deep passes, all in the first half. Three were incomplete; the fourth was a 34-yard touchdown to Steve Smith.
Rk
Player
Team
CP/AT
Yds
TD
INT
Sacks
Total
DYAR
Pass
DYAR
Rush
DYAR
Opp
23.
Colin Kaepernick SF
20/33
183
2
0
2
-7
-5
-2
ATL
Kaepernick's last pass of the first half was a 5-yard touchdown to Rod Streater. He wouldn't pick up another first down until the middle of the fourth quarter. In between, he went 3-of-9 for 19 yards with a sack.
24.
Derek Carr OAK
19/30
213
1
1
2
-7
-13
6
SD
25.
Case Keenum LARM
5/9
32
0
0
0
-16
-16
0
SEA
All of Keenum's dropbacks came with Los Angeles down by 21 points in the fourth quarter. His only first down was a 16-yard completion to Brian Quick on second-and-5.
26.
Ryan Fitzpatrick NYJ
5/10
31
0
1
0
-22
-20
-2
MIA
All of Keenum's Fitzpatrick's dropbacks came with Los Angeles the Jets down by 21 20 or more points in the fourth quarter. His only first down was a 16-yard 13-yard completion to Brian Quick Bilal Powell on second-and-5 first-and-10.
27.
Jameis Winston TB
17/35
247
2
3
4
-50
-54
4
DAL
Winston's last pass of the third quarter was a 10-yard touchdown to Cameron Brate that put Tampa Bay ahead 20-17. In the fourth quarter, he went 3-of-12 for 19 yards with one first down, two interceptions, and three sacks, and the Bucs lost 26-20. Winston was sixth among all quarterbacks through three quarters this week, but last by a huge margin in the fourth.
28.
Blake Bortles JAC
12/28
92
0
1
2
-71
-67
-5
HOU
Apparently Bortles' magic garbage time powers do not activate when the Jaguars are actually ahead. When Bortles first got the ball in the fourth quarter, the Jaguars were up 20-11. Fifteen minutes later, he had gone 2-of-9 for 17 yards with as many first downs (one) as interceptions, and the Jaguars had lost 21-20. These are the kinds of performances that get coaches fired.
29.
Jared Goff LARM
13/25
135
0
0
4
-73
-73
0
SEA
The new kickoff rules this year mean that teams will start a lot of drives at their own 25. Within his own 25, Goff went 4-of-8 for 23 yards with one first down and three sacks.
30.
Alex Smith KC
15/28
163
0
1
1
-76
-66
-10
TEN
31.
Carson Wentz PHI
22/42
170
0
1
1
-77
-92
15
BAL
Red zone passing: 2-of-9 for 8 yards and no first downs. (He did pick up a first down on a 13-yard DPI.) So of course, on a potential game-winning two-pointer, they passed and failed.
32.
Sam Bradford MIN
32/42
291
0
1
5
-84
-84
0
IND
Third/fourth downs: 5-of-9 for 21 yards with three conversions and a sack.
33.
Brock Osweiler HOU
6/11
48
0
2
0
-91
-91
0
JAC
Osweiler only threw two passes that traveled more than 8 yards downfield. Both were intercepted.
34.
Bryce Petty NYJ
20/36
235
1
2
3
-100
-100
0
MIA
In the third quarter, Petty went 4-of-10 for 15 yards with no first downs and a sack.
35.
Trevor Siemian DEN
26/40
282
0
1
4
-104
-105
1
NE
Almost all of Siemian's negative value came on third and fourth downs, where he went 6-of-12 for 51 yards with one interception, one sack, and only two conversions.


Five Best Running Backs by DYAR (Total)
Rk
Player
Team
Runs
Rush
Yds
Rush
TD
Rec
Rec
Yds
Rec
TD
Total
DYAR
Rush
DYAR
Rec
DYAR
Opp
1.
Ryan Mathews PHI
20
128
1
1/1
5
0
80
79
1
BAL
One of the best games you'll ever see by a running back who had a fumble. But Mathews had six 10-yard runs and eight first downs against Baltimore's mighty run defense. The Ravens have now given up 29 10-yard runs on the year. Mathews has 20 percent of them.
2.
LeSean McCoy BUF
19
153
2
3/3
16
0
64
65
-1
CLE
McCoy's success rate against Cleveland was a mind-boggling 79 percent. He had ten total first downs, and five runs that gained at least 10 yards (including 20- and 24-yarders), while getting hit for no gain just once.
3.
Devonta Freeman ATL
20
139
3
2/2
16
0
52
37
15
SF
Freeman had a fumble too. He also had 11 runs of 6 yards or more, including gains of 20 and 34 yards, and eight first downs.
4.
Jordan Howard CHI
17
90
1
4/4
23
0
51
42
9
GB
Howard's five first downs on the ground -- including gains of 11, 13, and 13 yards -- all came in the second half. He was hit for no gain just once. He added another first down as a receiver.
5.
Derrick Henry TEN
9
58
2
0/0
0
0
46
46
0
KC
Henry's success rate against Kansas City was 89 percent. He had six first downs in only nine carries, including gains of 11, 14, and 15 yards.


Five Best Running Backs by DYAR (Rushing)
Rk
Player
Team
Runs
Rush
Yds
Rush
TD
Rec
Rec
Yds
Rec
TD
Total
DYAR
Rush
DYAR
Rec
DYAR
Opp
1.
Ryan Mathews PHI
20
128
1
1/1
5
0
80
79
1
BAL
2.
LeSean McCoy BUF
19
153
2
3/3
16
0
64
65
-1
CLE
3.
Ty Montgomery GB
16
162
2
2/3
1
0
39
51
-11
CHI
Montgomery ran for seven first downs against Chicago, including gains of 13, 26, 36, and 61 yards.
4.
Derrick Henry TEN
9
58
2
0/0
0
0
46
46
0
KC
5.
Jordan Howard CHI
17
90
1
4/4
23
0
51
42
9
GB


Worst Running Back by DYAR (Total)
Rk
Player
Team
Runs
Rush
Yds
Rush
TD
Rec
Rec
Yds
Rec
TD
Total
DYAR
Rush
DYAR
Rec
DYAR
Opp
1.
Kenneth Farrow SD
15
39
0
2/3
14
0
-60
-37
-23
OAK
Odds are most of you have never heard of Kenneth Farrow until right now. But with Melvin Gordon out with a hip injury and Danny Woodhead, Branden Oliver, and even quasi-runner Dexter McCluster all on IR, the Chargers have been forced to use the undrafted rookie out of Houston as their primary runner. Against Oakland, Farrow only ran for one first down, his longest carry gained only 6 yards, and he had a fumble. He also had a fumble on one of his two catches.


Worst Running Back by DYAR (Rushing)
Rk
Player
Team
Runs
Rush
Yds
Rush
TD
Rec
Rec
Yds
Rec
TD
Total
DYAR
Rush
DYAR
Rec
DYAR
Opp
1.
LeGarrette Blount NE
17
31
1
0/0
0
0
-41
-41
0
DEN
Blount's 1-yard touchdown against Denver was his only first down on the day. His longest run gained only 5 yards, and he was hit for no gain or a loss four times. He did this against a Denver defense that has been pretty terrible against the run this season.


Five Best Wide Receivers and Tight Ends by DYAR
Rk
Player
Team
Rec
Att
Yds
Avg
TD
Total
DYAR
Opp
1.
Brandin Cooks NO
7
8
186
26.6
2
92
ARI
Cooks had touchdowns of 65 and 45 yards against Arizona, plus four other first downs.
2.
Brandon LaFell CIN
7
9
91
13.0
0
64
PIT
On top of his nine catches, LaFell's longest play was a 39-yard DPI. He had seven first downs on the day.
3.
Tyler Lockett SEA
7
8
130
18.6
1
54
LARM
Lockett had a 57-yard touchdown against the Rams, plus a 29-yarder and two other first downs.
4.
Jarvis Landry MIA
3
4
108
36.0
1
43
NYJ
In his career, Landry is averaging 6.0 catches and 64.0 yards per game. He only caught three passes against the Jets, but the longest -- a 66-yard touchdown -- eclipsed his career average for yards by itself.
5.
Aldrick Robinson ATL
4
5
111
27.8
0
40
SF
All of Robinson's catches produced first downs, including gains of 20, 21, and 59 yards.


Worst Wide Receiver or Tight End by DYAR
Rk
Player
Team
Rec
Att
Yds
Avg
TD
Total
DYAR
Opp
1.
Brandon Marshall NYJ
1
11
16
16.0
0
-69
MIA
Well at least that one reception converted a second-and-13.

Posted by: Vincent Verhei on 19 Dec 2016

39 comments, Last at 22 Dec 2016, 12:35pm by MC2

Comments

1
by Chappy :: Tue, 12/20/2016 - 10:39am

Not really sure I buy the analysis about QBing being worse. Seems like the cut-offs for bad/good/average are pretty arbitrary. (For example, if you use last weeks DVOA data, I think you have 3 fewer bad QBs). Wouldn't a better measure be some measure of variance in DVOA among qualified QBs?

2
by techvet :: Tue, 12/20/2016 - 10:46am

Wait, so Ty Montgomery had more rushing yards and more TDs with fewer rushes than Ryan Mathews and yet his rushing DYAR isn't even close to Ryan Mathews? Please explain. Is this because he had a bunch of runs that gained little or no yardage?

6
by MinisterCheevy :: Tue, 12/20/2016 - 11:10am

Opponent adjustments?

8
by techvet :: Tue, 12/20/2016 - 11:20am

Good point because Mathews was going up against a big rushing defense.

28
by Arkaein :: Tue, 12/20/2016 - 3:03pm

Seems hard to believe it would add up to that. With 10 YPC and only 39 DYAR on 16 carries, that's about 3 DYAR per carry. It would imply that an average RB should gain about 7 YPC against the Bears. And Chicago's rush D is only slightly below average by DYOA.

Regardless of where Montgomery ranks compared to other players, 39 DYAR just seems really low, especially for a player with 2 TDs and no fumbles.

32
by Eddo :: Tue, 12/20/2016 - 4:47pm

Using P-F-R, I see the following carries (in ascending order by yards gained) for Montgomery:

-2 yards on 3rd and 1
-1 yards on 1st and goal from the 2
1 yard on 1st and 10
1 yard on 1st and 10
2 yards on 1st and 10
3 yards on 1st and 10
3 yards on 2nd and goal touchdown
3 yards on 2nd and 2 first down
4 yards on 2nd and 10
4 yards on 1st and goal touchdown
9 yards on 1st and 10
13 yards on 1st and 10 first down
26 yards on 1st and 10 first down
36 yards on 1st and 10 first down
61 yards on 1st and 10 first down

So seven very good successes, one more strong success (9 yards on 1st and 10), one neutral (4 yards on 2nd and 10), and six failures. Honestly, his placement seems about right.

3
by jtr :: Tue, 12/20/2016 - 10:51am

I would love to see a more detailed analysis of QB DVOA over the years in the offseason when you guys have more time to work on it. Graphing out the distribution of DVOA and seeing how the distribution changes over time would give a much better view of how QB talent in the NFL has evolved over time. It sounds like the top tier of QBs have been getting better faster than the rest, making it harder to compete with a QB who's below the median. But it's hard to draw too strong of a conclusion from a quick-and-dirty analysis like the one above.

12
by garion333 :: Tue, 12/20/2016 - 12:24pm

I'm with you on this. I'd also suggest throwing out a QB's first couple of seasons, if they played. Curious to see if the throw-the-rookie-to-the-wolves mentality has made QBs look worse or if there really is less QB talent.

Also, maybe instead of simply counting the number of good/bad/average QBs you list the percentage of QBs who were good/bad/average for the year. Seems unfair to compare 40 QBs to 30 QBs.

23
by Duke :: Tue, 12/20/2016 - 2:36pm

What jtr said.

My first thought reading this article was that maybe the top QBs are skewing DVOA more than they had in the past. Which does mean there are "more bad QBs" but not in the way that phrase implies

4
by MilkmanDanimal :: Tue, 12/20/2016 - 10:59am

I feel like the calculation for DVOA should be modified so that when the right tackle apparently dies on the field at the start of the fourth quarter and nobody notices, leaving the QB open to repeated free runs from the defensive end.

Also, RIP GOSDER CHERILUS NEVER FORGET.

I'm just saying, I wouldn't mind a Word of Muth focusing exactly on whatever the hell Cherilus thought he was supposed to be doing in that fourth quarter.

5
by jtr :: Tue, 12/20/2016 - 11:08am

Relatedly, perhaps Bryce Petty deserves a break this week too. He took a sack-fumble in the first quarter when the right tackle decided he could just ignore Cameron Wake, and then got brutally double-teamed in the fourth quarter when the center snapped the ball early and neither OT really got out of his stance at all.

34
by johonny :: Tue, 12/20/2016 - 5:46pm

As noted, worst WR of week Marshall did him no favors either. Marshall ran his mouth big all week and then got out played by an injury replacement rookie corner. Although mostly he just dropped passes. Early, often, and critically. Howard's play helped, but Marshall killed most plays all by himself.

15
by Will Allen :: Tue, 12/20/2016 - 12:37pm

I really do wish there was a way to quantify offensive line play across years and eras, in order to confirm or disprove my suspicion that blocking and coordination among the big guys has declined significantly. What has often made the games less enjoyable to me is the consistently poor o-line play, which really narrows the type of offenses that can be utilized. Denver's trying to run their offense with guys who can't begin to execute the backside blocking in the run game, much less anything else. It's ridiculous. The Vikings are going to have a game where, if this keeps up, Bradford is going to go 32 for 35, for 96 yards, and 8 sacks.

The game is more fun to watch when a o-lines can execute basic fundamentals.

16
by PirateFreedom :: Tue, 12/20/2016 - 12:56pm

That's a very good point.
I think almost all QBs would be at least 'good' if they had excellent lines.
Very few can still be 'good' with bad line play.

I wonder if the decline in running plays has made life harder on O-lines.
Maybe they aren't worse, maybe the job is harder.

Continuity might be another difference but free agency has been around for quite a while now, on the other hand I don't remember interior linemen getting expensive until Hutchinson

18
by Will Allen :: Tue, 12/20/2016 - 1:16pm

The CBA's restrictions on practice, especially practice in pads, very likely harms o line play more than any other unit. If you don't get to practice handing off responsibilities at full speed, with full contact, doing it in the game gets a lot harder. Fans just don't commonly appreciate the intricate coordination required, in an extremely violent environment.

Toss in that college spread offenses really don't train o linemen well for the next level, and boring NFL offenses are the predictable result. Yeah, they keep scoring up by making the rules tough for dbs, but raw scoring numbers dont really tell the story.

24
by Duke :: Tue, 12/20/2016 - 2:38pm

If that's true, then figuring out a way to train OLinemen under the newer rules is the new market inefficiency. I wonder how you'd try to solve it.

29
by Will Allen :: Tue, 12/20/2016 - 3:17pm

Or you go the Dallas route, and be very good/lucky at investing in o-linemen in the 1st round, and dominate games with your offensive line. Whether that was lucky or good (and how would we categorize Stephen Jones physically restraining Jerel Jones from drafting Johnny Manziel instead of an o-lineman?), it may not be repeatable.

7
by PatsFan :: Tue, 12/20/2016 - 11:14am

Vincent,

How much of a positive adjustment did Brady get for playing the DEN defense?
How much of a negative adjustment did Siemian get for playing the NE defense?

9
by PatsFan :: Tue, 12/20/2016 - 11:26am

With respect to the change after 2003, something else happened in 2004 -- Bill Polian's jihad (sorry, "point of emphasis") against pass defense.

Maybe making it harder to play pass defense, while it floated all ships higher, benefited the better QBs disproportionally and so increased the "spread" between best-of-best and everyone else, pushing more QBs out of good into average and pushing more out of average into bad?

10
by dmstorm22 :: Tue, 12/20/2016 - 11:53am

The real rise didn't really start until after the lockout though. There definitely was a spike in '04, but there was mostly a course correction in 2005-06.

2007 had another spike, and then the real spikes started 2011 onwards.

To me it has more to do with the rise of short-passing, shotgun offenses and raising the pass/run ratio to favor the pass.

Sure, defensive backfield play has become a lot tougher, but there has to be more than just that since even comparing recent years to 2004 there is a big difference.

39
by MC2 :: Thu, 12/22/2016 - 12:35pm

I think the reason the numbers are better now than in 2004 is because it has taken a while for teams to adjust to the fact that passing is now so much more efficient than running. Most coaches are very stubborn and set in their ways, and it takes a while to weed out the guys (e.g. Jeff Fisher) who simply refuse to adjust.

Also, there have been some other rules changes since then that have helped the offense (e.g. roughing the passer, can't hit a defenseless receiver, etc.). Again, it takes a little while for teams to fully adjust to the impact of these changes, and as they do, the better offenses keep getting gradually more efficient, and the gap keeps growing between the teams that successfully adjust, and those that are unable (or unwilling) to do so.

22
by mrt1212 :: Tue, 12/20/2016 - 2:33pm

This has been my prevailing theory on why some QBs like Brees are so far and away better in this era - he among a handful of others really are adept at making the most of the rule structure whereas most can't capitalize to nearly the same degree and are exposed in their dearth of talent relative to Brees et al.

25
by Tomlin_Is_Infallible :: Tue, 12/20/2016 - 2:50pm

dearth is a really nice choice of words

https://www.youtube.com/watch?v=hHE0sA2qKy0

--------------------------------------
The standard is the standard!

36
by dbostedo :: Tue, 12/20/2016 - 11:40pm

Cool video, but pretty disingenuous to claim that Brees is more accurate than an olympic archer.

Sure, Brees hit the bulls-eye more often, but from 20 yards, where the archers shoot from 70 meters! I'd think Brees would be lucky to get the ball to hit the target at all, much less the bulls-eye, from 70 meters (not that THAT would be a valid comparison either).

11
by big10freak :: Tue, 12/20/2016 - 12:17pm

Randall Cobb was on the field for 51 plays on Sunday and did not have a pass thrown in his direction. Chicago consistently either locked him up at the LOS forcing Rodgers to look elsewhere as the timing was off or had a hand on him downfield where he could not disengage and get separation. Physical play really impacts his game.

13
by tuluse :: Tue, 12/20/2016 - 12:25pm

What is the standard deviation of DVOA over the years? Are QBs spreading out more or just shifting down?

17
by Eddo :: Tue, 12/20/2016 - 12:58pm

Also, do we have more below-average QBs now because the overall average has increased?

Let's say, in 2000, we had ten QBs, with raw (i.e. NOT DVOA) values of:

-10, -8, -6, -4, -2, 2, 4, 6, 8, 10

That average is 0, so we have five below, five above.

But now, let's say, in 2016, we have:

-6, -4, -4, 2, 2, 4, 8, 12, 16, 20

The average is now 5 - a good bit higher than in 2000 - and SIX QBs are below it.

Subjectively, I don't see worse QB play in general - I see BETTER QB play from 2/3 of the starters in the league, but the remaining third or so is about as bad as they were fifteen years ago (i.e. Brock Osweiler is about as bad as the 32nd best starter in 2000). So the overall average is up, which means more QBs are below it.

27
by Chappy :: Tue, 12/20/2016 - 2:59pm

You guys are saying the same thing I'm saying in comment #1. Totally agree. The author could show the standard deviation and/or show say, quartiles of QBing so you see how the top 25% perform relative to the bottom 25%. My guess is there is drift.

An even simpler way would be to show the median. My guess, like yours, is that in the Brady/Manning (Brees/Rivers/Rodgers) era, you have some pretty big outliers bringing up the average, so the median player might be below the mean.

Lastly, I think the cut-off for qualifying QBs should be era adjusted. It seems kind of weird to me that 1995 has about as many qualifying QBs as the previous years, but there were two expansion teams added. Why is this the case? I suspect because using 175 qualifying throws was a pretty high bar when the running game was more valued. Point being, being a back-up QB today means you throw 175 throws in 4 games and that was maybe half a season of throws for a back-up in the 1990s. I think that only shows that back-ups are being more relied upon today to throw than in years past, but I don't think that means that back-ups are worse--they are just leveraged more (or the cost of an injury is higher).

31
by Eddo :: Tue, 12/20/2016 - 4:34pm

I agree that the standard deviation would be good to know, but I'm actually saying I'd like to know the not-adjusted-to-average values, so I'd want to see something besides DVOA. Or DVOA to be recalculated (for this exercise) using the same baseline across all the years (it's done year-by-year now, right?).

14
by serutan :: Tue, 12/20/2016 - 12:36pm

"Winston was sixth among all quarterbacks through three quarterbacks this week, "

Do I want to know how a quarterback goes through quarterbacks? Sounds like it could be grisly.
______
Was wr

19
by Lance :: Tue, 12/20/2016 - 1:21pm

The first paragraph is telling. To anyone who is paying attention, the NFL has made it much much easier to pass in the last 20 years or so, and the numbers reflect that. Amazingly, though, the TV and radio pundits haven't caught on (ahem) to this, and still swoon over the big numbers put up by modern QBs. Sorry, but 4000 yards and 35 TDs is no longer the feat is was back in 1995 or something.

26
by PatsFan :: Tue, 12/20/2016 - 2:57pm

Yeah. I'd like to see all QB and receiving stats normalized to the season they occurred in.

So sure, keep track of the raw yards, TD, etc. as well but also have adjusted yards and adjusted TDs done something like this:
1) pick a season to be the baseline for all the stats
2) player's adjusted stat = league avg for stat in baseline year * (player's raw stat)/(league avg for stat in current year)

The idea is to compute some sort of "constant yards" and "constant TD passes" and the like, analogous to how economists compute "constant dollars" when making intertemporal comparisons.

30
by RickD :: Tue, 12/20/2016 - 3:50pm

Even with that in mind, I would dislike seeing the stats reduced to "number of standard deviations above or below the mean".

20
by Tomlin_Is_Infallible :: Tue, 12/20/2016 - 1:21pm

The description on BigBen sums up what is wrong with nfl play by play logs and our use of the word "deep".
--------------------------------------
The standard is the standard!

21
by Red :: Tue, 12/20/2016 - 2:13pm

Vince - Where did your 6.77 yards per pass figure come from? Standard Y/A is 7.2 and NY/A is 6.4 this year. It seems like you're using different measures in the numerator and denominator, leading to a distorted result.

33
by Vincent Verhei :: Tue, 12/20/2016 - 5:06pm

Not really sure I buy the analysis about QBing being worse. Seems like the cut-offs for bad/good/average are pretty arbitrary. (For example, if you use last weeks DVOA data, I think you have 3 fewer bad QBs). Wouldn't a better measure be some measure of variance in DVOA among qualified QBs?

I'm not exactly sure what you're saying here, but the numbers in the table represent all of 2016, not just one week.

Wait, so Ty Montgomery had more rushing yards and more TDs with fewer rushes than Ryan Mathews and yet his rushing DYAR isn't even close to Ryan Mathews? Please explain. Is this because he had a bunch of runs that gained little or no yardage?

The biggest reason is the massive opponent adjustments Mathews gets for playing Baltimore. Without those, Mathews would only have had 46 DYAR. Montgomery also had more bad runs -- eight runs that were worth negative YAR, even without opponent adjustments. (Most of these gained 3 yards or less, without gaining a first down.) Mathews also had eight, but in four more carries. And again, against Baltimore.

How much of a positive adjustment did Brady get for playing the DEN defense?
How much of a negative adjustment did Siemian get for playing the NE defense?

Brady: plus-79, most of any quarterback this week.
Siemian: minus-18, 26th

What is the standard deviation of DVOA over the years? Are QBs spreading out more or just shifting down?

There's no real pattern.

https://twitter.com/FO_VVerhei/status/811310321612754944

Somebody asked about median DVOA -- there's a lot of noise, but it's definitely on the decline.

https://twitter.com/FO_VVerhei/status/811313434646212608

"Winston was sixth among all quarterbacks through three quarterbacks this week, "
Do I want to know how a quarterback goes through quarterbacks? Sounds like it could be grisly.

Heh. Three QUARTERS. Fixed.

Vince - Where did your 6.77 yards per pass figure come from? Standard Y/A is 7.2 and NY/A is 6.4 this year. It seems like you're using different measures in the numerator and denominator, leading to a distorted result.

Oh, dammit. I just realized the numbers at PFR list net passing yards for teams, but don't distinguish between yards gained passing and yards lost on sacks. So the numbers listed in the opening paragraph are (league-wide net passing yards, including sacks) divided by (total passes, NOT including sacks).

OK, I found the real data. It's fixed now.

I agree that the standard deviation would be good to know, but I'm actually saying I'd like to know the not-adjusted-to-average values, so I'd want to see something besides DVOA. Or DVOA to be recalculated (for this exercise) using the same baseline across all the years (it's done year-by-year now, right?).

Yes, the baseline is recalculated each year so the sum-total for each season is zero. I think you're looking for the baselines, but that's easy to answer just by re-reading the opening paragraph (even with its errors): League-wide baselines for quarterbacks are higher than they have ever been before.

35
by Eddo :: Tue, 12/20/2016 - 6:27pm

"Yes, the baseline is recalculated each year so the sum-total for each season is zero. I think you're looking for the baselines, but that's easy to answer just by re-reading the opening paragraph (even with its errors): League-wide baselines for quarterbacks are higher than they have ever been before."

Thanks, Vince.

So if the baseline - the mean for DVOA purposes - is trending up, but the median is trending down, that would mean there are more "elite" QBs these days, but probably the same amount of mediocre ones as ever, right?

That is, compare these two sets of raw values:

A: -10, -8, -6, -4, -2, +2, +4, +6, +8, +10
B: -10, -8, -6, -4, -2, +4, +8, +12, +16, +20

The baseline/mean for set A is 0. For B, it is 3.

Given that, the values over average for each set would be:

A: -10, -8, -6, -4, -2, +2, +4, +6, +8, +10
B: -13, -11, -9, -7, -5, +1, +5, +9, +13, +17

A's median is 0, while B's is -2, which is what we're seeing in real life, no?

EDIT: Of course, B's standard deviation is going to be higher than A's, which isn't necessarily what we're seeing. But these two sets show a case where (a) the baseline is higher, (b) the median over-average metric is lower, (c) the worse raw scores are about the same, and (d) the elite scores are even better.

37
by Chappy :: Wed, 12/21/2016 - 11:23am

Thanks. I wrote the comment:

Not really sure I buy the analysis about QBing being worse. Seems like the cut-offs for bad/good/average are pretty arbitrary. (For example, if you use last weeks DVOA data, I think you have 3 fewer bad QBs). Wouldn't a better measure be some measure of variance in DVOA among qualified QBs?

What I meant here is that I looked at your qualifying QBs in the database for week 15 and figured out that 3 (terrible) QBs wouldn't have qualified in week 15, but presumably did in week 16 (since you have better data). My main point is that if you did the same analysis one week earlier you'd have 3 fewer "bad" QBs so the cut-offs are arbitrary. I also think you have a problem with your 175 qualifying pass cut-off. Today that means you are selecting in QBs that would have otherwise been true, non-qualifying back-ups in say the 1990s because many fewer passes were attempted. Now 175 passes could be 4 games of a back-up. Back, say, in the 1990s coaches would have stopped playing a back-up before they got to that many passes.

38
by jtr :: Wed, 12/21/2016 - 12:35pm

Hey Vince, you get asked in the comments about the opponent adjustment for some player every week. It might save you some effort if you add YAR onto the tables so that people can figure it out for themselves.