Buffalo Bills QB Josh Allen

Divisional Round Quick Reads

The Buffalo Bills came out for their divisional round game against the Baltimore Ravens with Josh Allen passing all over the place. He threw short, and he threw deep. He threw down the middle, he threw a bunch of passes to his right, and he threw to his left too. He threw to a Pro Bowler, a 30-year-old journeyman, a fourth-round rookie, and a running back. And then after all those passes, Allen … dropped back to pass again, but ended up scrambling for a short gain. After that, he passed and passed and passed some more, and also took a sack. By the time Allen handed the ball off to a teammate, there were barely three minutes left in the second quarter. And that was Buffalo's only handoff in the first half; they finished the first half with five more passes and a kneeldown.

Brian Daboll, Buffalo's offensive coordinator, had a game plan that was probably the most pass-heavy attack we have seen in an NFL playoff game. The Bills pulled ahead in the second half and won 17-3, and they did run a bit in the fourth quarter to kill some clock, but they still finished with only 16 running plays all day. Seven of those were runs by Allen (including scrambles), and four of those were kneeldowns, which means Buffalo had only nine handoffs and a dozen total plays where they tried to run for positive yardage.

Those aren't unprecedented numbers, but they're definitely on the low end of the bell curve. The game database at Stathead found 18 other playoff games -- dating all the way back to a 1949 Browns win over a completely different Buffalo Bills franchise in the AAFC -- where the winning team finished with 20 or fewer runs. We then went through the boxscores and, when available, the play-by-play to count quarterback runs and kneeldowns. Here are those 19 games, sorted by date:

All Teams That Won Playoff Games with 20 Runs or Fewer
Year Tm Opp Week Result Runs Yds Avg TD QB
Runs
Kneels Real
Runs
Handoffs
2020 BUF BAL DIV W 17-3 16 32 2.00 0 7 4 12 9
2017 NE JAX CCG W 24-20 19 46 2.42 1 5 3 16 14
2016 GB DAL DIV W 34-31 17 87 5.12 2 2 0 17 15
2015 NE KC DIV W 27-20 14 38 2.71 1 6 4 10 8
2015 ARI GB DIV W 26-20 (OT) 19 40 2.11 0 0 0 19 19
2014 NE BAL DIV W 35-31 13 14 1.08 1 6 4 9 7
2013 IND KC WC W 45-44 19 100 5.26 1 7 3 16 12
2010 GB PIT SB W 31-25 13 50 3.85 0 2 2 11 11
2009 NO IND SB W 31-17 18 51 2.83 0 1 1 17 17
2002 OAK TEN CCG W 41-24 17 89 5.24 2 8 1 16 9
2002 SF NYG WC W 39-38 20 90 4.50 2 7 0 20 13
2002 PIT CLE WC W 36-33 20 89 4.45 1 2 0 20 18
1999 STL TEN SB W 23-16 13 29 2.23 0 1 0 13 12
1999 STL MIN DIV W 49-37 17 31 1.82 1 3 1 16 14
1993 KC HOIL DIV W 28-20 18 71 3.94 1 3 3 15 15
1991 DET DAL DIV W 38-6 16 84 5.25 1 1 0 16 15
1991 HOIL NYJ WC W 17-10 20 71 3.55 0 3 1 19 17
1985 MIA CLE DIV W 24-21 19 92 4.84 2 0 0 19 19
1949 CLE BUF DIV W 31-21 20 72 3.60 1 1 0 20 19
* Estimated based on boxscores

Four of these teams had fewer total runs than Allen's Bills, but that number falls to three if we remove kneeldowns, and two if we take out all quarterback runs. Both of these teams were Bill Belichick's New England clubs, in back-to-back divisional round wins over the Ravens and Chiefs in 2014 and 2015. (Note that this table only includes teams with 20 total runs or fewer, so we may be missing some games where a quarterback such as Randall Cunningham or Michael Vick ran around all day and rarely handed off.)

Kurt Warner's Greatest Show on Turf Rams that won the Super Bowl in 1999 are the only team to make the list twice in the same season. Tom Brady's Patriots show up in three different years and Aaron Rodgers' Packers appear twice. Obviously, this year's Green Bay and Tampa Bay teams are built in a completely different manner from those clubs, but if things do go south in a hurry for one team in the NFC Championship Game, at least both quarterbacks have had experience in offenses that need to pass every down to win.

None of these teams, however, opened the game so averse to the run as Allen's Bills. Of the 14 games here since 1999, only one other team made it more than seven plays without handing off. The 2015 Patriots matched the Bills with 14 straight passing plays to open the game before Brady handed off to Steven Jackson (a great example of "that guy played for that team?" syndrome). Even then, New England made a choice to run; Buffalo only ran because the Ravens chased Allen out of the pocket. The Bills chose to open the game with 20 straight passing plays, and that is almost certainly an all-time record.

And that really tells you how much faith this team has in their quarterback right now. Look at the passers in this table: Brady, Rodgers, Montana, Moon, Warner, Drew Brees, Dan Marino, and Otto Graham are all Hall of Famers (or will be). Carson Palmer, Andrew Luck, Jeff Garcia, and Rich Gannon were all Pro Bowlers. There are a few oddballers here (hello, Tommy Maddox and Erik Kramer!), but generally speaking you must be an awfully good quarterback for your team to ask you go out and win the game by yourself. And if we're being honest, Allen didn't win this game -- his defense did -- but that still says a lot about the turnaround Allen has made this season.


 

Quarterbacks
Rk
Player
Team
CP/AT
Yds
TD
INT
Sacks
Total
DYAR
Pass
DYAR
Rush
DYAR
Opp
1.
Aaron Rodgers GB
23/36
296
2
0
0
190
182
8
LAR
Rodgers gains 49 DYAR due to opponent adjustments. In our game preview, we noted how dominant the L.A. defense had been on throws to the offense's right. Well, Rodgers must not be a reader -- on throws to his right, he went 15-of-20 for 178 yards and a touchdown. The Packers were also effective on very short throws -- targeting receivers within 2 yards of the line of scrimmage, Rodgers went 11-of-12 for 69 yards and a touchdown.
2.
Patrick Mahomes KC
21/29
255
1
0
0
115
112
3
CLE
If it felt like Mahomes was unstoppable in this game, well, for a little while, he was. At one point in the first half he picked up first downs on six straight throws, picking up 104 yards and a touchdown in the process. But he only picked up four first downs and 110 yards in 18 dropbacks afterwards before leaving the game in the third quarter. He was very efficient on throws that traveled 10 or more yards downfield, going 6-of-8 for 104 yards, though only two of those reached the 16 yards required to be qualified as deep balls.
3.
Tom Brady TB
18/33
199
2
0
1
107
97
10
NO
Brady gains 52 DYAR due to opponent adjustments. His DYAR got as low as -46 as he started out 1-of-6 for 5 yards with a sack; obviously, he got better after that. He was especially effective on throws to his left, going 9-of-13 for 86 yards and a touchdown.
4.
Jared Goff LAR
21/27
174
1
0
4
21
19
1
GB
With about 11 minutes left in this game, Goff and the Rams had the ball, trailing by only seven points. Goff failed to pick up a single first down from that point forward, going 3-of-7 for 10 yards with two sacks. He only picked up one conversion on third downs the entire game, going 3-of-5 for 25 yards with a sack. On the plus side, he was perfect on throws down the middle, completing each of his four passes for a total of 72 yards. He also completed all four of his deep passes for a total of 79 yards.
5.
Chad Henne KC
6/8
66
0
1
1
-9
-9
0
CLE
Henne came into the game with Kansas City up 19-10 in the third quarter. He was nearly perfect on third and fourth downs, completing all four of his passes for 39 yards and three conversions (the one that didn't convert was a 5-yard gain on third-and-6). And that's not even counting his long scramble on third-and-forever that set up his fourth-and-1 conversion to ice the game.
6.
Tyler Huntley BAL
6/13
60
0
0
1
-19
-24
5
BUF
All of Huntley's passes came with Baltimore down 17-3 in the fourth quarter. Two of his three first downs came on throws to running backs, where he went 2-of-4 for 20 yards.
7.
Josh Allen BUF
23/37
206
1
0
2
-29
-20
-9
BAL
In Ravens territory, Allen went 9-of-17 for 48 yards with one touchdown, two sacks, and a fumble. He did not have a good night on deep balls, going 1-of-8 for 21 yards.
8.
Baker Mayfield CLE
23/37
204
1
1
1
-55
-47
1
KC
Mayfield's totals include -10 DYAR receiving for his one catch, an 8-yard loss. That was one of his 10 failed completions, most this weekend. He had a real rough day on throws down the middle -- his five throws in that direction included a completion to Jarvis Landry for a loss of 3, that completion to himself for a loss of 8, an incompletion, an interception, and a 5-yard DPI.
9.
Lamar Jackson BAL
15/23
162
0
1
3
-104
-83
-21
BUF
Jackson's rushing numbers: nine carries for 34 yards. Only two of those carries gained first downs, two were stuffs, and he also had a fumble. He only completed one pass for a conversion on third downs, going 2-of-6 for 32 yards with a pick-six. He had an awful night in Buffalo territory, going 7-of-12 for 40 yards with two sacks, a pick-six, and exactly zero first downs.
10.
Drew Brees NO
19/34
134
1
3
0
-111
-116
5
TB
Brees gains 41 DYAR due to opponent adjustments. He was especially terrible at the beginning and ends of this game. He picked up just one first down in his first 13 dropbacks, going 6-of-13 for 31 yards. Then, after his touchdown pass to Tre'Quan Smith put the Saints up 20-13 midway through the fourth quarter, he went 6-of-13 for 35 yards with two interceptions, plus a 20-yard DPI. He was at his worst, though, between the New Orleans 30- and 40-yard lines, where he went 4-of-11 for 20 yards and all three interceptions. If you're curious, Jameis Winston picked up 42 DYAR for his one pass, a 56-yard touchdown to Tre'Quan Smith.

 

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.
Leonard Fournette TB
17
63
0
5/6
44
1
75
26
49
NO
Fournette gains 34 DYAR due to opponent adjustments. He only ran for three first downs and his longest carry gained only 8 yards, but he was stuffed only one time and 13 of his runs gained 3 yards or more. Each of his catches gained at least 8 yards. Four gained first downs; the other was an 8-yard gain on second-and-10.
2.
Jamaal Williams GB
12
65
0
0/0
0
0
51
51
0
LAR
Williams gains 12 DYAR due to opponent adjustments. All 12 of his runs gained at least 3 yards, an absurdly high minimum gain for somebody with a dozen carries in a game. Eleven counted as successful plays and six gained first downs, the longest a gain of 11.
3.
Aaron Jones GB
14
99
1
1/2
14
0
47
42
5
LAR
Jones gains 20 DYAR due to opponent adjustments. He was not as reliable as his teammate Williams, with only six successful runs and four first downs, but he was more explosive, with a 60-yarder. And he was stuffed just once.
4.
Alvin Kamara NO
18
85
0
3/6
20
0
37
44
-7
TB
Kamara gains 29 DYAR due to opponent adjustments. He ran for five first downs -- including gains of 10, 11, and 17 yards -- while being stuffed just once. None of his catches gained first downs, and only two counted as successful plays.
5.
Cam Akers LAR
18
90
1
1/1
6
0
31
29
2
GB
Akers loses 10 DYAR due to opponent adjustments. He was stuffed just once while running for six first downs, including gains of 10 and 13 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.
Jamaal Williams GB
12
65
0
0/0
0
0
51
51
0
LAR
2.
Alvin Kamara NO
18
85
0
3/6
20
0
37
44
-7
TB
3.
Aaron Jones GB
14
99
1
1/2
14
0
47
42
5
LAR
4.
Ronald Jones TB
13
62
0
0/1
0
0
26
32
-6
NO
Jones gains 20 rushing DYAR due to opponent adjustments. He only had two first downs and a long gain of 12, but he was stuffed just once while nine of his carries counted as successful plays.
5.
Cam Akers LAR
18
90
1
1/1
6
0
31
29
2
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.
Nick Chubb CLE
13
69
0
2/5
4
0
-33
-7
-26
KC
Chubb loses 22 DYAR due to opponent adjustments. He did run for three first downs against Kansas City, but 44 of his 69 yards came on those three carries, while he was stuffed four times. His two catches resulted in no gain and 4 yards, both on first-and-10; he was the target on incomplete passes on first-and-20, second-and-20, and second-and-10.

 

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.
Gus Edwards BAL
10
42
0
0/0
0
0
-20
-20
0
BUF
Edwards is here because he had fumble on first-and-10 that cost him 17 DYAR even though the Ravens recovered. Otherwise he was just fine, running for four first downs while being stuffed just once.

 

Five Best Wide Receivers and Tight Ends by DYAR
Rk
Player
Team
Rec
Att
Yds
Avg
TD
Total
DYAR
Opp
1.
Tre'Quan Smith NO
3
3
85
28.3
2
69
TB
Smith's four targets: 13-yard gain on third-and-11; 56-yard touchdown; 16-yard touchdown; 20-yard DPI on third-and-8.
2.
Stefon Diggs BUF
8
11
106
13.2
1
48
BAL
All eight of Diggs' catches produced first downs, and aside from his 3-yard touchdown, they all gained 11 yards or more. That includes a pair of third-down conversions.
3.
Tyreek Hill KC
8
10
110
13.8
0
37
CLE
Hill's totals include five rushing DYAR for his three carries for 9 yards. Six of his catches produced first downs, the longest a gain of 26; he also had a 15-yard gain on second-and-23.
4.
Josh Reynolds LAR
3
4
65
21.7
0
27
GB
All of Reynolds' targets came on first-and-10. His three catches: gain of 28, gain of 21, gain of 16.
5.
Allen Lazard GB
4
8
96
24.0
1
26
LAR
All four of Lazard's catches produced first downs, the longest a 58-yard touchdown.

 

Worst Wide Receiver or Tight End by DYAR
Rk
Player
Team
Rec
Att
Yds
Avg
TD
Total
DYAR
Opp
1.
Mark Andrews BAL
4
11
28
7.0
0
-41
BUF
Only one of Andrews' catches picked up a first down. All four of them were thrown by Lamar Jackson; he failed to catch any of his three passes from Tyler Huntley.

Comments

45 comments, Last at 20 Jan 2021, 11:38am

1 There are a few oddballers…

There are a few oddballers here (hello, Tommy Maddox and Erik Kramer!)

The Cowboys were stacking the box to stop Sanders and daring Detroit to pass.

https://www.freep.com/story/sports/nfl/lions/2016/10/15/detroit-lions-remembering-playoff-win-dallas-cowboys/92173392/

So Detroit passed.

2 DYAR adjustments

So GB has a -5.7 rush defense DVOA meaning better than the baseline 0 and Akers losses 10 DYAR? That doesn't fully compute for me. I would expect almost no adjustment for facing a D with that rating. I can see how specific situations for individual runs could create a small negative or positive adjustment, but 10 DYAR is a 25% adjustment putting him at 31 vs the 31+10=41 YAR.

Something feels off about that.

7 The baseline of 0 is for all…

In reply to by DisplacedPackerFan

The baseline of 0 is for all offensive plays, and the average run play is less efficient than the average pass play. At -5.7%, the Packers only finished ranked 18th in run defense DVOA. (Hey, that's better than I would have guessed!) So the opponent adjustment for RBs must be based on the strength of the average run defense, which was probably a bit better than -5.7%. An opponent adjustment of 10 yards across 18 carries don't seem too out-of-whack to me for a run defense that was just a little worse than average.

3 Close to 100% passes--Buffalo is on to something

Here are the passing vs rushing DVOA of the 4 remaining teams:

            Pass         Run

GB        52.0         3.0

KC        49.0        -5.6

BUF      43.3        -15.0

TB         37.1        -2.0

I know the counterargument already, if you do not run then you will not have these wildly efficient passing offenses.

My argument:  Let's say you are GB, if you run 50% of your plays you will getting a total DVOA of 27.5.  

(52 X.5 +3 X.5)= 27.5 DVOA

So you are making the assumption that running helps passing.  Running half of the time is clearly crazy in my opinion, do you really think that if GB passes every play they would not have better than 27.5 DVOA?

The 75% passing number is more reasonable:  (52 X.75) + (3 X.25) = 39.75    To assume that using this ratio is better than 100% passing, you are assuming that passing every play reduces the efficiency by close to 24%  39.75/52=.764 DVOA ratio.

I am not talking situational (such as 3rd and 1, or two score lead in 4th quarter).

Do this for any team.  Buffalo refuses to use a -15% run play when they can use a 43.3% pass play.  The season is on the line, it makes sense to me.

Look at the 2 minute drill, or even the 1 minute drill.  When Patrick Mahomes throws the running game away, does he get stopped?  Ever?  What if you played the whole game like that drill?

 

 

 

4 This

This works for essentially every team. Even mediocre QBs are better on average pass plays than top run plays. The envelope has yet to burst because no one has pushed it far enough. One day we'll see a team, locked in their fate, before week 17...erm 18 just go all out. I look forward to that day. 

 

6 Hey, I'm a very pro-passing…

Hey, I'm a very pro-passing person, but sometimes it depends what the defense is doing. Using Saturday's game as an example, the Rams are a defense that goes out of their way to keep their safeties out of the box and use lighter personnel explicitly to try to prevent explosive passing plays. The Packers ended up running the ball a ton (more than they usually do) because they knew they had great looks to run into and their O line was absolutely road-grading the Rams front.

GB ended up with a 56% rushing success rate, their RBs averaged 6 yards/carry, they only faced like two third downs of more than 4 yards to go, and they went 8/12 on third down, one of which I think was a kneeldown. On the TD to ice the game, Lazard ran right past a deep safety who was crashing hard to the LoS on play action - a way more aggressive reaction from the defense than your normal false step or hesitation by the LBs. The plays the Packers made in the passing game are what ultimately won them the game, but it was a game (maybe a rare game) where the defense tried to be patient and invite the offense to run and the offense took full advantage and forced the defense out of what they wanted to do.

9 So you are making the…

So you are making the assumption that running helps passing.  Running half of the time is clearly crazy in my opinion, do you really think that if GB passes every play they would not have better than 27.5 DVOA?

The General Issue
This discussion has been had many times before. Passing efficiency will go down if you never run the ball. How much we can't say, but you will either have more short passes that replicate what the run game does to help make sure you convert downs, or you will end up with more 3 and outs, and more longer passes which likely totals out to lower overall efficiency because not converting a first down or scoring is less efficient than converting or scoring. You will also have more interceptions even if the QB throws them at the same rate simply because of more passes. That means more turnovers since INT are more common than lost fumbles from RB or planned QB runs. Then you run into the defensive adjustment argument. 

The real question is, what is the ideal ratio. How little can you run before your passing efficiency drops. At what point does your run efficiency climb even higher because defenses are selling out to stop the pass? If you know all the factors the equation can be solved. My guess is that for most teams a 65/35 pass/run split is probably ideal. But then you have teams like GB where they use so much play action, and motion, and game plans that are designed to make defenders react thinking it's a run only to have a better passing window. Or react thinking it's a pass only to help set up a big run. They were the #1 offense in the NFL this year by DVOA, was their split actually the best or would they have been better with something different? I admit I would love to be able to boot up an alternate universe and run the experiments. But I can't.

Specific Examples
To illustrate the efficiency drop, as well as the difficulties in decoupling the effects, look at Josh Allen against Baltimore. -20 DYAR. Heck Buffalo had -16% passing VOA (no opponent adjustments) in the game vs a 43.0% VOA for the season. Yes there were other factors, but BAL started to play to stop the pass, and while BUF still won the game, BUF did not pass as efficiently as they did in game where they ran the ball more. It's possible that if BUF had run more that Allen would have had a positive VOA. It's also possible that the overall VOA/DVOA could have been better OR worse as a result because of all the other confounding factors.

Further discussion on the ideal ratio. You are right that more passing is better than rushing. For the regular season GB had 526 passing attempts and 392 non QB rushing attempts. Rodgers had 38 rushing attempts I actually don't know how many of those were scrambles vs kneeldowns vs planned runs. Tim Boyle had 13 kneeldowns I know that. Most of Rodgers plays weren't kneels since they brought Tim in at the end of games to do that, but I do recall 3 for Rodgers. So lets say 30 of the 38 were scrambles, 3KD, and 5 planned runs (I'm confident 2 of his 3 rushing TD were planned runs and I recall at least 2 sneaks). A scramble counts in rushing DVOA, but for this argument and the way GB plays it is a play that was supposed to be a pass but the player on the field decided it would get better results as a run for whatever reasons. So that puts us at 556 planned passes vs 397 planned rushes; throwing out kneeldowns completely. So GB had passed 58/42 pass/rush split during the regular season. 

That worked out to be the #1 offense in the NFL by DVOA (and points per drive) breaking down to #1 passing and #5 rushing. I do contend that if they had thrown the ball 714 times and only run it 238 times for the 75/25 split that yes they likely would have had worse results. I bet their rushing DVOA would have likely gone UP given the actual skill of the line and backs and the fact that teams would have stopped playing the run against them and sold out vs the pass so when they did run it would have been more favorable. But I also believe their passing DVOA would have gone down for the reasons listed above.

There was a lot to be said in the offseason that GB did nothing to improve the offense. An additional year in a new system should bring some improvements but GB with pretty much the same players went from 19.6% passing DVOA and 4.5% rushing DVOA to 52% and 3%. They had a 63/37 pass/run split in 2019. Lots of factors, but if I want to be a cherry picking anecdotal evidence pointer I can claim the BS that running the ball 5% more in 2020 improved passing DVOA by 32% at the cost of 1.5% rushing DVOA. THAT IS NOT WHAT HAPPENED but it was a fun thing to throw out there. It's possible it had an effect, but there are way to many factors to make that claim.

Specific game situations As dank067 points out for a specific case I agree with his analysis. I highly doubt GB would have had the 58 yard TD to Lazard without the running game. Another data point. Davante Adams, Mr All Pro, DYAR leader had 10 targets for 66 yards, so 6.6 ypt, which feels like a pretty apple to apples comparison to Aaron Jones 14 carries for 99 yards or 7.1 ypc. We also know that Adams had less than 26 DYAR or he would have made the charts. So Jones, on the ground, was move valuable than the best receiver in the league this year because the defense was set up to stop Adams. I'm pretty sure, though I have not done so, if you watch all 22 film of the game that you won't see the Rams safeties, Linebackers, and corners coming down towards the line as hard they did early in the game. It was the running the game that starting getting them to react more any play that looked like a run, and that helped get more separation not only for that Lazard TD but a couple of the Tonyan receptions that might have been too well covered otherwise for Rodgers to make the throw. I say that because I've seen it happen in other games this season. I've seen how for the specific case of the GB offense, how the defense reacting to the running the game has helped create situations that lead to completions. I in fact thing part of the reason Rodgers had his highest completion percentage by 2.5% was in part because of how defenses treated one of the best GB rushing offenses I've ever seen.

10 The Turnover issue with more passing

The more I think about it the bigger a factor this is. I think it will increase TO rate in games, which will change behavior because it will affect who potentially scores and how aggressively a team might need to play to insure they win. I'm still assuming that a game heavily titled towards passing can still do things to manipulate the clock, though it's trickier due to the incomplete pass = stopped clock. I think a higher passing volume will lead to more plays in a game as well though I'm not attempting to factor any of that in.

And yeah I've posted multiple times on this not because I think it's a wrong idea, but because, as previous discussions have pointed out, it's a good idea to try and find the right balance and I enjoy thinking about it. I'm having more fun thinking about it than I would just looking up past studies and citing them too so I'm maximizing my fun quotient equation. :-)

Mulling it over
I see no reason for INT% to go down if you throw more passes. I see a few reasons for it go up (defenses having more players back so zones have more people in the passing lanes, pass rush ignoring run responsibility so pressure forcing bad passes increasing) and several reasons why it likely stays the same. But that does mean that actual INTs will go up from higher passing volume. There were 409 INT in the NFL this year so 1.6 a game. Some of those are likely end of half hailmary passes too but I don't have the data. 

Overall lost fumbles will likely go up given that most fumbles in the modern NFL are by QBs either scambling or getting sacked which are both situations that happen because you are intending to pass and not rush. Looks like there were 211 lost fumbles in the NFL this year (0.82 a game) and 162 of them were by QB/WR/TE the only ones I can be reasonably sure should be assigned to the passing game. That's 76.7% of fumble turnovers likely being part of the passing game though I'm sure some of those are actually in rushing DVOA. Some of those fumbles may have occurred on a designed running play (i.e. a Josh Allen or Lamar Jackson designed run or a WR sweep, etc). Of course some of the 49 fumbles by other positions might have been a RB catching a pass a fumbling so should be part of the passing game. I'm willing to just say 65% and use 137 for the passing game and 74 for the running game. That likely helps make up for the hailmary numbers too.

Raw numbers pulled from the ESPN stat pages show 18,018 passing attempts in the NFL and 13,792 rushing attempts in the NFL this year. For a total of 31810 plays. If we say an average game has 4 kneeldowns (1 before half time, 3 to ice the game) that would pull out 1024 rushing plays.

Passing game turnovers 409 + 137 = 546
Rushing game turnovers 211 - 137 = 74

546/18018 = 3.03% of passing game plays result in a turnover.
74/12768 = 0.58% of rushing game plays result in a turnover.

So basically a pass is 6 times more likely to be a turnover than a rush. That is kicking out 1024 rushing plays as kneeldowns which is actually inflating the rushing turnover numbers.

So if we got with 30786 non kneeldown plays and kick up from the 58% passing to 75% passing then you have, 23,090 passing plays now, 5072 more. So that would give us 153.6 more turnovers or .6 more a game. So we go from 2.4 TO a game to 3.02 a game. 

Conclusions
As I mentioned I think the rate of INT would go up, which leads to a lowering of DVOA, and I think more TO per game leads to different decisions. I don't have the number but it would be fun to look at the rate of scoring off of drives that start from a TO vs drives that start from a punt vs drives that start from a Kick off vs a turn over on downs. I believe that the TO starting drives have the higher scoring rate. So a team that wouldn't have turned the ball over because they were running it with it's significantly lower turn over rate might turn it over and give up a score because they were passing more.

As mentioned this all needs to go into the calculus of the maximally efficient run/pass ratio. But it's hard to suss out. Of course depending on the exact equation there can be more than one ratio that gives the maximum efficiency too. So it's possible that the Packers are the most efficient at a 58/42 split, but worse at a 65/35, and then have the same max at 75/25.

11 Great stuff here

two additional comments, one linked to Packers.

I am 99% positive that the huge improvement in Offensive Performance for GB this season was primarily caused by greater familiarity with the LaFleur/Shanahan offense-- including but not limited to the QB and also not excluding some changes by the HC/Play Caller as well. makes you salivate for Year Three-- though the schedule is gonna be a doozy.

In terms of run/pass ratio, I believe down and distance is also hugely important. If you have set up your running game for success on first down-- say, averaging over 5 yards each time you run--then that will help make the passing attack even more efficient. Likewise if you have a play grouping that pretty much guarantees a 5+ yard gain when passing on 1st down and very few INCs-- the essence of what the West Coast began in terms of a possession passing game-- then your running game when facing 2nd and short can also be more effective. 

My bottom line is don't waste first downs with predictable running plays and overly risky pass plays that will far too often put an offense into obvious passing situations where the advantage begins to shift to the defense. Seems ridiculously obvious of course but I swear close to half the teams in the league still don't get it and set up their offenses to fail far too often. take the 6-7 yards if it's being given to you.

of course it isn't an absolute in the typical chess game. A short 2 yard gain from a certain formation might set up a bigger play down the road. And there are times when a longer first down pass is the perfect call.

what i am seeing with the Shanahan tree offense-- is that pre-snap motion is being used increasingly to produce winning plays/matchups that are passes and runs-- in ways that a Mike McCarthy, for example, simply couldn't comprehend merely two seasons ago. And that it also can produce games-- SF v GB both games last year, for example-- where the running attack can simply dominate and make the passing game almost unnecessary. rare events but clearly a possibility.

 

14 Bills

In reply to by oaktoon

The Bills have had games like this before though.  Not this extreme but they are and have been very pass happy.   I watched almost every snap this year and to me it just seemed like a combination of a good D and very tricky passing weather.   The Allen-Diggs over the top connection for instance was particularly bad, and they don't usually miss those.     

That said, they came out in the second half and mixed in the run more and did move the ball better.  But my inclination, for whatever that's worth (!), is that the relatively low passing output was more about the weather and the Ravens than the lack of 3 yard runs by Singletary.  (although it was interesting that they didn't run Allen at all this week).

16 Q

I think you might even be missing the #1 Benefit of running the ball: QB Health.

For most teams, an injured QB means Game/Season over. Every run attempt is another play of nearly guaranteed health for your QB.

Especially in the Regular Season, the goal shouldn't be to have the #1 Highest/Most Efficient Offense.  The true goal should be to be efficient enough to win while maximizing your QB's health for the playoffs.

17 I could see this

In reply to by Q

But we also see Brady and Brees be healthy and throw the ball a ton. Mostly because they're willing (and sometimes only can) throw the short pass/checkdown (that's still more efficient than running).  Same with Peyton. Could be a problem if they refuse to throw the ball but even Russ hasn't missed a single game in his career yet. 

18 Also, the surest way of…

In reply to by ImNewAroundThe…

Also, the surest way of preserving QB health is to build up big leads within games, and to win enough games so that playoffs spots are tied up early. Then they don't have to play at all. 

 

22 I don't know what your point…

I don't know what your point is. That Mahomes should have just been handing off at that point? Or not in the game at all? You can't legislate against injury entirely. 

They led by 12 in the third quarter. A TD on that drive and they would have been most of the way to garbage time. 

25 This isn't directed…

In reply to by ImNewAroundThe…

This isn't directed specifically at you, it was just prompted by your comment that the short pass/checkdown game is more efficient than running. I think some people lose track of why.

I will say if a 5 yard pass on 1st and 10 is more efficient than a 5 yard run on 1st and 10. Or if a 3 yard pass on 3rd and 2 is more efficient than a 3 yard run on 3rd and 2 the system measuring efficiency is broken. For team DVOA my understanding is those plays should measure the same.

The only way one of those plays should be more efficient is if it took less time off the clock. That can certainly happen, but a 5 yard run up the gut should take basically the same time as a 5 yard pass to the outside. A 5 yard pass to the outside likely takes less time than a 5 yard outside run, but a 5 yard pass to the middle can take more time than a 5 yard run to the middle especially if it's too as outside receiver on a crossing route. However I don't think DVOA has any duration of play component. Besides a slower play can be more efficient in terms of win percentage than a day play. I only mention it in case clock efficiency is on someone's mind. DVOA is a yardage/scoring efficiency metric though.

Traditionally the short passing game has been more efficient because short passes tend to be 3-7 yards and short runs tend to be 0 to 4. But if you never run but still care about getting a new set of downs and have to take what the defense gives you, you'll see more of the passes that only produce 0 to 4 yards which would drop overall pass efficiency.

The biggest reason that passing has become more efficient overall is because completion percentages have gone up. 60% used to be good. Now it's poor to OK. Removing 0 yard plays helps a lot. This applies to the run game too of course. Williams was more efficient than Jones because he wasn't stuffed, despite have fewer yards and ypc.

The only inherent reason passing should be more efficient at getting yards is that it is easier to cut down the number of defenders that can make a tackle. But in theory a 52% DVOA run game could exist. Yards are yards and if you figure out how to get 8 yards per carry you should be approximately as efficient as an offense that gets 8 yards per passing attempt. 

So again the real question is what is the calculus for max efficiency. I contend that some running will always be required to hit max efficiency. I also agree that for most situations, teams, and schemes that more passing is probably more efficient and that most, if not all, teams should pass more of they want to win more. 

I just enjoy the complexities behind figuring that out because it's not as simple as pass more = more efficient because passing more will almost certainly make passing less efficient than it currently is for a given team.

29 Maybe it is that simple though

We just haven't tried it yet. At least not at the NFL level.

I do agree it's unlikely 100% as the last couple minutes teams should prioritize making the other team use their timeouts so they either get them to do that, or they just flat out end it with a 1st or two; a win win imo.

But other than that why not pass the entire first 3 quarters? You may be able to bench your starter if you have a big enough lead (but if you consistently don't maybe it's because you don't have a good enough QB).

Sure the efficiency may go down but it doesn't mean you can't mix up the depth routes to get different pass plays. Get creative and do some weird stuff. Is one run good enough to keep teams honest? Oh no we're getting into "guy gets x rush attempts" territory that we know is faulty. 

But I think there's enough history of pro football to know that running is just easier to stop. Throughout all the changes but especially now, even though running efficiency is the best now! But yes if you can get 8 ypc consistently go right ahead. But as we've seen even great backs like Henry, can't sustain it and it eventually hurts the team. 

It's complex but we'll never find out until someone pushes the envelope imo.

32 But other than that why not…

But other than that why not pass the entire first 3 quarters? 

The goal of most strategists is to prevent their opponent from knowing their optimal strategy beforehand. If your opponent can reach Nash equilibrium, you're probably going to lose.

38 Yes

FO tweeted the same thing when the Titans were better at both run and pass than the Packers. But the Packers had a higher overall offensive DVOA because they passed more. Then the Packers ended up beating them head to head because they werent insistent on establishing the less optimal strategy. 

19 The irony, of course, is…

In reply to by Q

The irony, of course, is that Manning's career-ending injury came on a running play.

40 Wait, what?

The history is a little murky, and I assume you are not talking about Cooper in college (injury lead to a diagnosis of spinal stenosis, ending his career) or Eli (no final injury that I recall).  Peyton didn't have a career ending injury either (aside from Father Time stepping on his windpipe a little harder each week of his final season).  So I assume you mean whatever event injured his neck about three years before he missed the whole 2011 season....  IIRC it was against Washington and I assumed it was a sack or maybe a scramble.  Or had he handed off and gotten popped maybe as it was happening?  Man, that would be crappy, but certainly how I'd coach the D.

I DO recall an earlier game vs Miami in which he was clearly hit high and late by Lorenzo Bromell in the 4th quarter, fractured his jaw, missed one snap before finishing the game.  And of course didn't miss any more time that season.

More on the kind-of-terminal injury, please....

Here's a link to the play that Dungy and Manning credit for the start of his neck woes.  (the clip link below leads to a pass play, actually)  https://www.youtube.com/watch?v=9gjdmww3vgM 

 

42 https://www.cbssports.com…

In reply to by Bobman

https://www.cbssports.com/nfl/news/peyton-manning-tries-to-throw-block-gets-blasted-by-chargers/

The torn quad against the Chargers. This was his Waterloo; he would never be Peyton Manning again after this.

28 basic game theory

These values are not independent of the fact that strategies vary.  An offense that passes every down (hello, Steelers!) is going to end up facing a defense that is optimized to defend the pass, and thus will rate worse than and offense that offers a mix of passes and runs.

It is fallacious to say a run "earns -15% per play" and a pass earns "43.3% per play".  They only earn at those rates based on the mix that Buffalo was using.  

Hmm...maybe this isn't "basic game theory" because the basic game theory 2x2 square implies the wrong result.  In reality, the values in the square will change based on expectations of strategies.  

37 Responses from the 100% pass comment, great discussion

I think that the most interesting comments regard the question, "What is the optimal pass-run ratio."  Clearly this varies from team to team, but look at those pass/run DVOA's for the 4 team that are left.  Some bad teams have similar splits (see HOU and LAC). 

Why not pass for the entire first three quarters?  That is an interesting point to me, I can watch a Super Bowl, or NFC/AFC Championship game with any of these teams with that strategy.  I am not convinced that this would not be the best strategy, nor am I convinced that throwing in some runs is not a bad idea.  I am convinced that a 50/50 "balanced offense" is not a good idea, why would you take the ball out of the hands of these great passing offenses on half of the plays?

Yes the DVOA splits are based upon the actual ratio of pass/run that a specific team used .Regarding the Steelers, everyone was lining up to guard the pass, and they still could not run.  Buffalo had a Steelers like game this weekend.  The Ravens were guarding the pass and Buffalo could not run effectively at all.

With regards to interceptions and turnovers, the effectiveness of the passing game as expressed in DVOA includes its higher risk of turnover and for that matter sacks.  

Turnovers do occur more in the two minute drill at the end of the game when passing, this is desperation mode in many circumstances.  When you have a half a game left to play, I doubt that there are many interceptions during the first half during the 2 minute drill.  All QB's get to run the 2 minute drill in the first half.  However, remember who is losing and desperate, and thus throwing those interceptions at the end of the game, much more likely to be the QB's of the Jets or Jacksonville than prime of career Patrick Mahomes or Aaron Rodgers.   Speaking of the 4 QB's remaining, how comfortable are you in getting the game winning interception at the end of the game when protecting a lead of 1-6 points?  (I know its possible, see letter to Pete Carroll at the end of this post)

With regards to passing a great deal and losing, in the old days, losing let to passing and winning led to running.  That is still true, but to a lesser extent today.  Thus, correlating passes and runs to wins and losses does not work.

If you know of a way to come up with a 52% DVOA running game, please come to Baltimore, you can live in my home for free until the Ravens hire you (I will throw you out when your application is rejected).  

Does anyone remember the Thursday night opener after the Saints won the Super Bowl?   The Saints rolled, using no runs at all in the first half.  This was Drew Brees in his prime.

I am not totally against running.  Tom Brady is even more efficient on QB sneaks than on his passes.  

Great quote from Andy Reid regarding his decision this weekend, "I went to BYU, every down was a passing down"

I will finish with a sarcastic dig.  Letter to Pete Carroll:

Dear Pete,

I am a huge believer in passing over running.  However, with the Super Bowl on the line, you should have run on 1st down and goal from the 1 with Marshawn Lynch or run a QB sneak.

Very truly (not understand the strategy that is) yours,

_____________________  Sign name here

 

 

8 What a ridiculous effort by…

What a ridiculous effort by the GB OL (and I'm a Packer-hating Bears fan). Yes, Donald was not 100%, and yes, Aaron Rodgers is very good, but zero sacks and 5.2 YPC (188 yards total) against that defense after losing your all-pro left tackle? Remarkable.

45 I'm really interested in…

I'm really interested in seeing the o-line continuity and adjusted games lost articles in the offseason. Most of the time, the best o-lines rank among the most continuous and have the fewest games lost, but I don't think Green Bay will this year. GB's o-line room (coaches and players) have really done a remarkable job adapting. PFF has them as the second-best o-line in the league this year, and that's with Bakhtiari missing a quarter of the year, Linsley missing multiple games, starting RG Lane Taylor missing almost the entire year, and lots of onesy-twosy missed games that caused the line to shuffle for a week or two at a time. Elgton Jenkins was the only player to start all 17 games so far, and did so across 3 different positions (LG, C, RT). 

FO has them as 5th in adjusted line yards and 6th in adjusted sack rate (sacks are a QB stat, but the eye test says they've been robust in pass pro this year). Any way you measure it, they've been really good, despite a lot of shuffling.

15 teams that won withou throwing

the list of teams that won without throwing 20 passes is probably fairly long so you 'd need to cut it off somewhere else, say maybe 10....  We had the 49ers do this with 8 attempts vs Green Bay iirc last year... and  I remember noting the '73 Dolphins won with 7 and 6 throws in the same playoffs (though that's attempts, not dropbacks).  Still would be interested in seeing a comparison.   (I woulda  thought PFR/SH would be a tool that could spit that out, but it failed to find the two dolphin games so maybe not, not sure what else it is missing)...

23 Looked it up for you

There have been 107 playoff games won with fewer than 20 pass attempts. Drops to 40 with 15 attempts, and 11 with 10 passing attempts.

Season,Team (query: In a single game, since 1940, in the Playoffs, team won game, requiring Pass Attempts <= 10)

2019 49ers

2009 Ravens

1984 Seahawks

1977 Vikings

1973 Dolphins (2x)

1971 Dolphins
1952 Lions

1950 Browns

1949 Eagles

1940 Bears

 

 

27 Brady: His DYAR got as low…

Brady:

His DYAR got as low as 46 as he started out 1-of-6 for 5 yards with a sack; obviously, he got better after that.

 

Well, it is a cumulative stat, right?   

30 Yes, but it can also…

Yes, but it can also accumulate negative value. The more passes you throw, the more it can go up, but also the more it can go down. But I was just pointing out Brady got off to a poor start.

31 46 seems high

Maybe I'm reading it wrong, but did that sentence also imply that Brady when he was 1-6 for 5 yds with a sack had a 46 DYAR?

Seems way high - six failed plays out to seven (maybe all seven depending on D&D for the 5 yard gain?)

34 Phew

Haha, no worries

Glad it was an error. Otherwise I think we were all going to need to have a little talk about DYAR going rogue

39 GWD

Just wanted to check if the Divisional match between TB and NO, Brady should be credited with a GWD? 4th qtr started with the teams tied and Brady led the point scoring drives that netted the win. Has the definition of a GWD been changed?

41 We don't keep track of that…

In reply to by eagle97a

We don't keep track of that stat. You'd have to ask the folks at Pro Football Reference (we use and endorse PFR a lot, but they are a separate site). The first play of Tampa Bay's tie-breaking drive was the last play of the third quarter; that may disqualify it as a GWD.