DVOA Analysis
Football Outsiders' revolutionary metrics that break down every single play of the NFL season

Week 1 DVOA Ratings

Mark Andrews
Photo: USA Today Sports Images

You love them when your team is high! You hate them when your team is low! Once again, the famous Football Outsiders DVOA and DAVE ratings return for 2020.

Football Outsiders readers are familiar with DVOA, our main play-by-play metric. By now, most Football Outsiders readers are also familiar with DAVE, our rating that combines the preseason projection with the results of early games to give us a better prediction of how each team will rank at the end of the year. For those who don't know the story, this metric is called DAVE as a reaction to criticism that our stats are too much alphabet soup. I mean, who can argue with a guy named Dave? (Technically, it stands for "DVOA Adjusted for Variation Early.") In past years, we used DAVE for the first half of each season. This year, based on new research, we will be using DAVE for another month, until every team has played 12 games. In this week's DAVE ratings, the preseason forecast counts 93% and performance in Week 1 is only 7%. The weight of the preseason forecast goes down every week until we fully filter it out after Week 13.

Of course, though I'm calling our main metric DVOA here, it is actually VOA because there are no opponent adjustments right now. We do not apply opponent adjustments until after Week 4, so in Weeks 1-3 DVOA and VOA are the same thing. Please don't get all nitpicky about it. Most readers know what's up, and if you don't, I just told you!

The Baltimore Ravens are on top of our ratings after Week 1 for the second straight year. Last year, they chokeslammed the Miami Dolphins 59-10. This year, they "only" beat Cleveland by a score of 38-6. Once again, just like last year, the Ravens are in the top three for offense, defense, and special teams after one week.

Actually, this is the third straight year the Ravens have done this. In 2018, they beat the Bills 47-3 in Week 1 but "only" ranked second because the Jets (yes, you read that correctly) had an even bigger win over the Detroit Lions.

Based on DVOA, the Ravens weren't quite as good as the Browns were bad. Baltimore ended up with 88.0% DVOA, while Cleveland got spanked to the tune of -107.6% DVOA. Does a loss this bad mean we should give up on the Browns already? I went into the past to get an idea of how much we can learn from one game this bad.

Since 1985, 30 different teams have started the season with a game below -100%. (This is Week 1, so there are no opponent adjustments being considered, even for these past years.) Those 29 other teams averaged 6.0 wins for the season. Eight of them managed to finish 8-8 or better, and four of them made the playoffs. The problem for Cleveland is that no team has started the season with a loss below -100% and made the playoffs in almost 30 years. The four teams that did it were the 1991 Lions, the 1989 Oilers, the 1989 Steelers, and the 1985 Jets.

What about the other side? The Ravens didn't match the 119.3% DVOA they put up against the Miami Dolphins last year in Week 1, but they still were one of just 36 teams to put up a Week 1 game over 80% since 1985. The previous 35 teams averaged 10.1 wins and 24 of them made the playoffs that season. Looking only at teams coming off a top-five DVOA rank the year before, 9 of 11 teams with big Week 1 wins went on to the playoffs. The two that didn't? The 1997 Cowboys (who beat Pittsburgh 37-7 but finished 6-10) and the 2005 Bills (who beat Houston 22-7 in a game that wasn't as close as the score, then finished 5-11).

More interesting from this game may be the splits. Cleveland sold out to stop the run, and it worked! Baltimore ranked 29th this week in run offense DVOA. Of course, the Ravens also blew past the rest of the league in pass offense DVOA, over 140% while Seattle was second at 80.1%.

The more surprising Week 1 results are lower down on the page. Look over the whole table, and you're probably wondering three things.

  1. How on earth did Minnesota end up with a higher rating than Green Bay when the Vikings were losing 29-10 at the end of the third quarter?
  2. How are the Jets not way down with the Cleveland Browns given how badly they got embarassed in the first half of Sunday's game with Buffalo?
  3. What is Kansas City doing all the way down at No. 16 after what looked like a dominating victory over the Houston Texans?

I was also wondering these things! So let's take a bit of a closer look at each of these games. You'll notice that there's a connection between these three games and the disagreement between DVOA and conventional wisdom: the fact that DVOA measures the entire game instead of filtering out "garbage time." Yes, DVOA does count plays at only half strength when there's a huge lead in the game, but that lead has to be over 21 points in the fourth quarter. Readers expect DVOA to devalue plays much earlier than that, but as I've written many times, all my research has shown that excluding most "garbage time" plays actually makes DVOA less predictive for the future.

Green Bay 43, Minnesota 34

Another thing to remember about DVOA is that is an efficiency metric. It's measuring success per play against a leaguewide baseline. Often when DVOA ends up differing with conventional wisdom, it is related to one team running far more plays than the other team. That's one of the things that happened in this game. It will shock you to learn that the Minnesota Vikings gained more yards per play than Green Bay on Sunday, by 7.8 to 6.9. The Vikings did turn the ball over with an interception but the Packers had a fumble, although they recovered it themselves.

The big difference here is that Green Bay ran 76 plays while Minnesota ran only 49 plays. Should that be somehow accounted for in our ratings? I've never figured out a good way to account for teams running more plays that actually makes DVOA more predictive. Green Bay managed to string its plays together into more coherent, successful drives than Minnesota. I have a feeling that means something, but I also have a feeling that it means less than the final score would indicate.

The other issue, of course, is the timing. Minnesota scored 24 points in the final quarter, even though the game was mostly out of reach at that point. Should those plays be discounted by the DVOA system? Well, the problem with that is that if you discount Minnesota's best offensive quarter of the game, you also discount Green Bay's best offensive quarter of the game. Minnesota was much more efficient when they were losing in the last 15 minutes, gaining 9.9 yards per play. But Green Bay with that big lead managed to put up 10.7 yards per play! If we were to discount plays once win probability for one team is very low, we would be taking out Green Bay's best plays too.

Here's a look at both offenses by quarter. In case you are wondering, the Kirk Cousins interception was in the second quarter and the Aaron Jones fumble was in the fourth.

GB-MIN Offensive DVOA and Yards by Quarter, Week 1
Qtr GB DVOA GB Yd/Play MIN DVOA MIN Yd/Play
Q1 23.5% 6.4 53.2% 7.5
Q2 -10.6% 6.0 -158.6% 4.1
Q3 16.8% 6.2 1.2% 7.5
Q4 107.8% 10.7 113.8% 9.9

Buffalo 27, New York Jets 17

Here's another game with the same issues as Green Bay at Minnesota. Both teams turned the ball over twice. Buffalo gained only 5.0 yards per play compared to 4.8 yards per play for the Jets. The biggest difference was that Buffalo ran 81 plays while the Jets ran just 53. The Buffalo offense also shut down after a fantastic first quarter. In particular, the Bills had trouble running the ball, gaining only 2.7 yards per play on non-scramble runs. And the Jets got some offense going in the second half of the game. Buffalo was the better team in this game, certainly, but the idea that the Jets were a total garbage fire is a bit of an overreaction to just the first quarter (combined, let's be honest, with our priors about the current state of their locker room and the head coach).

BUF-NYJ Offensive DVOA and Yards by Quarter, Week 1
Qtr BUF DVOA BUF Yd/Play NYJ DVOA NYJ Yd/Play
Q1 52.1% 6.5 -90.1% 1.6
Q2 -22.9% 4.0 -23.7% 7.9
Q3 -19.2% 4.9 5.5% 7.7
Q4 -4.7% 5.5 9.4% 4.0

Kansas City 34, Houston 20

Once again, the yards per play numbers here are going to surprise you. Houston averaged 6.2 yards per play while Kansas City averaged only 5.5 yards per play. The Chiefs ran more plays, but not by very much: 67 to 58. Houston had an interception, and Kansas City had no turnovers. So it makes sense that Kansas City came out higher in DVOA. But it's not by much. How did the Chiefs end up with a -1.2% DVOA despite winning the game?

Again, the big question here is how to count plays when a team has a big lead in the fourth quarter. The Chiefs offense was completely different in the fourth quarter. Was it because they were playing with a big lead, so they didn't care about sending Clyde Edwards-Helaire to run over and over into a stacked box at the goal line? Their defense didn't really have a similar breakdown despite the big lead. In the first three quarters of the game, Kansas City averaged 6.4 yards per play with a 59% success rate. In the final quarter, Kansas City averged 2.3 yards per play with a 33% success rate. That dropped their overall numbers for the game.

Compare Kansas City's final numbers for the game to the league averages for Week 1, and you can see why their DVOA comes out right around the league average. Kansas City's 5.51 yards per play almost exactly matched the NFL average of 5.52 yards per play. The Chiefs had a higher success rate than league average, and thus had a positive offensive DVOA. But there are two sides to every story. The Texans had basically the same success rate as the Chiefs, and thus also had a positive offensive DVOA. So both offenses were just a little bit better than the NFL average based on both success rate and DVOA.

KC-HOU vs. NFL Average on Offense, Week 1
Offense Yd/Play DVOA Suc Rate
KC 5.5 17.6% 54%
HOU 6.2 11.1% 55%
NFL Avg Week 1 5.5 0.0% 47%

* * * * *

Football Outsiders playoff odds are updated through Week 1. The annual stats pages are now updated with 2020 data, although some of that data can be kind of sketchy after just one week (in particular the offensive line and defensive line pages). Snap counts and the FO+ DVOA database are also now fully updated through Week 1.

A quick bit of housekeeping: As you may have read when we discussed the new DVOA version 7.3, we are now correctly applying adjustments to both offense and defense for playing indoors. Therefore it is important to note that we have designated the new Sofi Stadium in Los Angeles as an outdoor stadium, not an indoor stadium. Although Sofi Stadium has a roof, the sides are open to the weather and the wind. In the long run, we'll have to see how Sofi Stadium plays, but for now the Rams and Chargers (and their opponents) will not be getting the same penalty to offense and bonus to defense as teams such as New Orleans and Detroit.

* * * * *

These are the Football Outsiders team efficiency ratings through one week of 2020, measured by our proprietary Defense-adjusted Value Over Average (DVOA) system that breaks down every single play and compares a team's performance to the league average based on situation in order to determine value over average. (Explained further here.)

OFFENSE and DEFENSE DVOA are adjusted to consider all fumbles, kept or lost, as equal value. SPECIAL TEAMS DVOA is adjusted for type of stadium (warm, cold, dome, Denver) and week of season. As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.

Please note that there are no opponent adjustments in DVOA until after Week 4. (It's still listed as DVOA instead of VOA because I don't feel like going through and changing all the tables manually.) Our second weekly table which includes schedule strength, variance, and estimated wins will appear beginning after Week 4.

DAVE is a formula which combines our preseason forecast with current DVOA to get a more accurate projection of how a team will play the rest of the season. DAVE is currently 93% preseason forecast and 7% actual performance.

To save people some time, please use the following format for all complaints:

<team> is clearly ranked <too high/too low> because <reason unrelated to DVOA>. <subjective ranking system> is way better than this. <unrelated team-supporting or -denigrating comment, preferably with poor spelling and/or chat-acceptable spelling>

RK TEAM TOTAL
DVOA
TOTAL
DAVE
RANK W-L OFFENSE
DVOA
OFF.
RANK
DEFENSE
DVOA
DEF.
RANK
S.T.
DVOA
S.T.
RANK
1 BAL 88.0% 18.9% 2 1-0 39.9% 1 -38.7% 3 9.4% 2
2 NE 52.3% 3.0% 11 1-0 36.4% 3 -23.8% 5 -7.9% 25
3 JAX 38.0% -7.1% 24 1-0 35.8% 4 0.5% 13 2.8% 5
4 SEA 37.0% 9.7% 5 1-0 38.6% 2 9.5% 21 7.8% 3
5 PIT 36.0% 10.6% 4 1-0 6.5% 14 -39.0% 2 -9.4% 27
6 WAS 35.9% -15.3% 30 1-0 -32.6% 28 -69.4% 1 -0.8% 13
7 BUF 16.7% -0.7% 19 1-0 2.6% 17 -19.0% 7 -4.8% 22
8 LV 13.7% -3.3% 23 1-0 31.7% 5 16.5% 23 -1.4% 15
9 NO 10.8% 22.2% 1 1-0 -13.7% 24 -14.3% 9 10.2% 1
10 MIN 10.8% -0.2% 17 0-1 28.2% 6 20.1% 26 2.6% 6
11 LAC 4.1% -2.2% 20 1-0 -7.6% 22 -20.4% 6 -8.7% 26
12 CAR 2.5% -13.1% 28 0-1 23.3% 8 24.0% 27 3.2% 4
13 CHI 0.3% -2.8% 22 1-0 0.9% 19 -1.6% 12 -2.2% 16
14 DAL 0.3% 6.0% 7 0-1 9.2% 12 3.9% 18 -5.1% 23
15 DET -0.1% 1.1% 14 0-1 -0.3% 21 1.2% 15 1.4% 9
16 KC -1.2% 18.8% 3 1-0 17.6% 9 17.5% 25 -1.3% 14
17 GB -5.0% -0.7% 18 1-0 24.2% 7 28.7% 30 -0.4% 12
18 ARI -5.2% 0.5% 16 1-0 0.5% 20 2.9% 17 -2.8% 17
19 SF -6.1% 3.9% 9 0-1 2.3% 18 0.6% 14 -7.8% 24
20 HOU -9.3% -2.3% 21 0-1 11.1% 11 16.7% 24 -3.6% 19
21 LAR -13.5% 3.9% 10 1-0 7.4% 13 9.5% 20 -11.4% 28
22 TEN -14.3% 0.9% 15 1-0 3.4% 16 -11.1% 10 -28.7% 32
23 TB -16.3% 7.1% 6 0-1 -16.1% 25 -16.4% 8 -16.7% 30
24 DEN -17.1% -7.5% 25 0-1 -13.3% 23 3.9% 19 0.1% 11
25 ATL -26.0% 2.9% 12 0-1 11.8% 10 38.9% 32 1.1% 10
26 IND -27.0% 4.3% 8 0-1 3.8% 15 26.0% 28 -4.8% 21
27 CIN -28.4% -15.5% 31 0-1 -33.2% 29 -8.5% 11 -3.7% 20
28 NYJ -35.2% -14.0% 29 0-1 -21.1% 26 2.7% 16 -11.4% 29
29 PHI -41.1% 1.9% 13 0-1 -71.3% 32 -33.6% 4 -3.4% 18
30 MIA -48.9% -17.3% 32 0-1 -23.2% 27 28.3% 29 2.6% 8
31 NYG -54.7% -9.5% 26 0-1 -45.4% 30 11.9% 22 2.6% 7
32 CLE -107.6% -12.0% 27 0-1 -57.5% 31 33.3% 31 -16.8% 31

Comments

116 comments, Last at 18 Sep 2020, 3:14pm

1 Washington ranked 6th…

Washington ranked 6th despite posting an ugly -32% offensive dvoa. How in the world did...Oh...their defense has a nearly 70% DVOA on defense. I would love for this to remain true all season. 

98 Not so fast.......

...baby slotter.

Your Skins are in Zona on Sunday and it's the home opener.  Pass rush is negated by baby Kyler's dipsey do scrambles and your offense would never be able to keep up.  Your QB has a rough day.  A couple sacks and a pick or two.

Zona dominates.   34-15

The Ron Rivera sympathy tour hits a bit of a rough skid in the desert.

113 Nope.

In reply to by scraps

"Raiderjoe" doesn't hold multiple degrees or state licensure.

 

2 GB-Minn

It pains me to say good things about announcers, but the team doing the game in Minnesota did point out late in the 4Q that the Vikings had more yards per play than Green Bay. So I'm not all that surprised. The Packer D couldn't do anything in the last quarter and I don't think it can all be attributed to being in prevent mode.

How does an offense (something called 'LV', isn't that the next superbowl?) have a 3.8% adjusted sack rate when it didn't give up a sack? What is the adjustment being made there?

3 Prevent Defense

In reply to by ammek

The idea behind prevent defense is that you sacrifice yards in order to make the other team chew up clock, so basically allowing just dinking and dunking down the field.  

The Vikings last three TD drives were 1:16, 2:07, and 1:58, so if the Packers were playing prevent they weren't doing a very good job.

5 I felt FO's preseason…

In reply to by ammek

I felt FO's preseason projection for GB's defense was really pessimistic - the reasons for regression (especially turnover rate) all made sense, but I also thought they were capable of baseline level of play that was still high enough for them to be much closer to an average unit. Well, the 4th quarter was definitely a peek at what could happen this year to their secondary, especially if the pass rush gets neutralized... hope Clark gets healthy soon.

At least you can still feel good about the offense sustaining 7 yards per play on offense for 76 plays to the tune of 522 yards and 31 first downs, and actually getting better as the game went on - they really never found that gear last season, except maybe vs. the Raiders.

6 To me, GBs get out of jail…

To me, GBs get out of jail free card was always Rodgers. No matter how far the defense sank, Rodgers ensured a competitive team that could win nearly any game. That Rodgers hasn't been there for a while now, but man did he come back roaring this week. 

 

I will be curious to see if hes back to being peak Rodgers because his apparent demise appeared inexplicable to me. 

9 Rodgers commented during training camp that he went back to 2010

video of his play and saw something that clued him in on something that needed fixing.  He didn't say what it was, so there has been all sorts of speculation (footwork, get rid of the ball faster, look for the dumpoff out of the back field rather than peering downfield for the big one, and so on).  He definitely looked on it and I wonder about his numbers if they factored in drops - MVS had some great catches but also some egregious drops.

32 I thought he went on the…

I thought he went on the record saying it was footwork. He started doing more leg and core strength training this off season. That's actually something I think ALL athletes over 30 should focus more on, regardless of sport, more core training has been huge for my mid 40's self with the distance running. 

I think they even showed some video during the game of how it had gotten worse and how this one game sample showed improvement. Or maybe I'm just thinking that because I've been thinking that has been his issue for awhile. After the collarbone injury he focused too much on his arm, which makes sense mentally, that's where the injury was you think more about it. I think last year I was saying he had lost arm strength and that explained a lot of things, if you are trying harder, plays are a bit slower, throws can be less precise. But I really need to think about it in terms of "throw strength" which has always been a full body thing. 

Favre was a better QB when he had a coach who hounded him on footwork. Makes sense that it applies to every other QB too. The tricky thing with both of the Packers HoF QB's I've been spoiled with is that I've had years and years of seeing two of the best ever at throwing "off platform". As a Wisconsin fan I started following Russell Wilson as well and watch more Seattle games than other teams, so there's a 3rd who is amazing at it. I've live in Kansas City for the last 7 years so I get to see a lot of Mahomes too. Yes I'm spoiled. LOL  

Rodgers didn't seem to decline in the improvised play aspect as much, which should have pointed out that his decline from amazing to merely above average NFL starter had to do with the feet in the pocket. I was chalking it up to age and supporting cast issues. I hope that he continues to reverse the trend. I said this after the draft, I figured the Love pick would do something with the chip on his shoulder, and for that reason it might not be fully wasted this year. I like to think I was right and that's part of why he went into deeper film study (the pandemic and extra time at home didn't hurt that either). We'll see.

 

As to general comments on the DVOA, I wasn't surprised I commented about it a few times in the Week 1 discussion thread. Watching the game, after Clark went out I was never comfortable. I kept seeing an MVS alligator arm tipped int and suddenly it's a completely different game. 

20 "something called 'LV'"  …

In reply to by ammek

"something called 'LV'"

 

Even weirder, to me, is the Las Vegas hockey team, when they are represented on sports tickers as: VGK.

22 This bothered me too at…

This bothered me too at first. I don't know anyone who lives there, but when I searched around for an answer on why they use "VGK" I think I read that they deliberately market themselves as "Vegas" as opposed to "Las Vegas" because that's in line with how the locals refer to the city (not just the rest of us). The official team name is in fact "Vegas Golden Knights."

39 Huh.  I had no idea that…

Huh.  I had no idea that they were officially "Vegas".  Interesting.  But in that case, shouldn't their abbreviation be: "VEG"? 

I worry too much about abbreviations, ever since I was commissioner for fantasy football in the pre-computer days, and had to write team abbreviations for each player.

"JAX" for Jacksonville also drives me crazy, because there is no "x" in Jacksonville.

53 NFL Abbreviations

The worst official NFL abbreviation is that the Rams are just "LA" while the Chargers are "LAC".  It's like the Chargers are officially relegated to second-tier status.

63 Yeah, I think you’re right. …

In reply to by The Ancient Mariner

Yeah, I think you’re right.  So I guess the Chargers need to be LACH.  Probably not enough people care for them to change it (I just happen to be OCD about symmetry).

64 LARM and LARD

In reply to by The Ancient Mariner

LARM and LARD are Football Outsiders inventions; the NFL at the time used RAM and RAI.

65 Back to my fantasy football…

In reply to by The Ancient Mariner

Back to my fantasy football commissioner comment.  I used the Los Angeles Times to tabulate my results.  The Times used RAM and RDR as the abbreviation for those teams, which caused me problems when trying to sort alphabetically by city.  But I assume the Times used those abbreviations since they were both LA teams, so the "LA" part was assumed.  Also the problem with LARM and LARD is that they are 4-character abbreviations, which can cause problems.

66 Among other larger reasons,…

In reply to by Aaron Brooks G…

Among other larger reasons, one disappointment of the new Sofi Stadium being off-limit to fans was that I was hoping that the Rams would sell out the stadium and the Chargers would still have Stubhub Center levels of attendance.  What a travesty that they moved to Los Angeles.  Still makes me sick.

99 Too early...

In reply to by ammek

...to make a definitive determination on what both teams actually are.

Honestly, I see both the same as middling NFC Central contenders in a slog.  Neither is good on defense and every game is probably a shoot-out.

Rodgers also wanted to make a statement about his "ability".  If I'm a Viking's fan, I'm worried that the Zimmer era could be hitting a downward slide.  I remember a few years ago during the Arians era in Zona that Zim's Vikes were putting a real game on the Cards.  Almost won.  That defense is no more.

4 Adjusted Sack Rate

How does an offense (something called 'LV', isn't that the next superbowl?) have a 3.8% adjusted sack rate when it didn't give up a sack? What is the adjustment being made there?

Derek Carr got flagged for intentional grounding, which is essentially a sack.

7 I said things were wonky…

I said things were wonky with the OL/DL stats early!

ASR measures sacks and intentional groundings, and Carr got one of the latter. But yeah, that ASR seems very high considering. He just threw a lot in down-and-distance situations where sacks are less likely, so the system adjusted him upwards.

8 I've never figured out a…

I've never figured out a good way to account for teams running more plays that actually makes DVOA more predictive.

Well you have, and I think we used it once upon a time -- it would be comparing DYAR instead of DVOA.

13 We've heard many times that…

We've heard many times that DVOA favors a move-the-chains offense over a boom-and-bust offense, right? So one would think that a team going for 80 yards on 8 plays would be more highly regarded than a team that goes 80 yards on 4 plays, then follows that up by going 0 yards on 4 plays. On a points-per-drive basis, the former is much better. But on a yards-per-play basis, they're exactly the same. 

So if you have two teams with similar yards per play, the one running more plays would seem to have the better offense, wouldn't it? What am I missing?

 

 

 

10 I was curious how badly…

I was curious how badly Detroit would have to cock up their season to get rid of Patricia and you brought up the Jets game. There is no result so putrid that Quinn will fire Patricia or Ford will fire Quinn. The 2018 Jets weren't exactly the 1991 Redskins.

101 I got good news...

After the Motor City Kitties get rocked in Arizona, the ownership will finally start considering cutting Patricia loose.

Last year they tied.

This year, Zona dominates.

 

11 -18.0 ALEX

After seeing that Cam had a -18.0 ALEX, I checked the box score and saw that he only attempted one pass on 10 third downs. The three runs on 3rd and 1 and 3rd and 2 make sense, but they also ran it on 3rd and 12, 3rd and 5 and two 3rd and 6's. Were some of those called pass plays where Cam decided to scramble?

18 Per the PBP, Newton only had…

In reply to by Perfundle

Per the PBP, Newton only had one scramble: a 12-yard gain on first-and-10 in the fourth quarter. 

He had three runs on third/fourth downs, and picked up a first down three times. That includes conversion on third-and-5 and third-and-6. 

He did also take a sack on third down in the second quarter. 

This Tweet has a four-minute video of Newton highlights. I don't know if it has all his runs, but it does show one of those third-down conversions on a QB draw.

https://twitter.com/PatsNationTM/status/1305886156501012480

102 Newton Magic..

..dries up as the schedule gets harder.

Seattle isn't Miami.

Neither is Arizona (dry heat vs. humidity)

As teams gear up for Newton's scrambles, he'll be increasingly one-dimensional.  I still see NE being a .500 club at best.  Their defense lost too many people and their offense didn't have much talent to begin with.  

115 .500?

In reply to by DIVISION

The Pats face the Dolphins twice and the Jets twice and should at least split with Buffalo. That's five wins. Hard to believe they can't win half of their non-division games simply by virtue of having Belichick on their sideline.

I imagine that Miami 'geared up' for Newton's scrambles as he's been doing it for a decade now. I also think it's unlikely that the coaching staff doesn't come up with various other wrinkles on offense.  Newton also was 15-19, with one drop and two tipped balls. So his arm seems fine.

The one place NE is talent poor on offense is the WR's. The o-line is excellent and the running back corps is deep. They also have talent (albeit unproven) at TE. Although they are thin at linebacker, the Pats still have the best, and deepest, defensive backfields.

Ten wins seems to be a reasonable expectation.

12 I bet CFB at fairly high levels and

I bet CFB at a semi-serious level and so I like to look closer into each game/box score and with that said the YPP numbers are something I like to closely analyze. I run into same problem and here's how I feel I have made things a little better: 

1A) first off I throw out all 4th qtr garbage stats. I define garbage time when a team is up by 17 points or more because 17 points equals 3 possessions. I think you should look closer into lowering that garbage time number as I really believe there's a different mindset when a defense is up by 3 full possessions. 1B) I also throw out all offensive plays ran inside the red zone unless the first play inside the red zone gains more than 6 yards on a play. Once there's an offense play of 6 yards or less  then I start throwing the red zone plays out as they only end up penalizing offenses too much. Without getting too complex I will also keep numbers if they score a TD from inside red zone if they averaged more than 6 YPP (so a 1 yd play followed by 18 yard play still counts as they ran 2 plays for 19 yards or more than 6 YPP inside RZ). Bottom line is there's almost upside inside the red zone for an offense YPP wise.

2) I always felt the problem with YPP stats is the above scenarios and one other thing which is a team that runs way more plays seems to get hurt by this. So what I do is I like to do is compare total running plays for the game (cause usually a team behind will throw more and teams ahead will run more) and round up the differential so if a team has 18 more run plays than their opponent I round that up to 20 and add a decimal in the middle so I give the run team 2.0 more YPP overall for the game differential overall (I don't care if its offense or defense just looking at overall differential). With all that said there's obviously a bigger differential overall in CFB YPP numbers per game than an NFL game so maybe make the NFL differential half of that adjustment.

This is far from a perfect way of doing things but I feel it's an improvement overall and a better representation of how teams played as it gives credit more to teams who are running the football a lot late which tends to mean a bigger lead, trying to kill the clock and in doing so they are usually killing their YPP numbers as well for obvious reasons. Just my two cents. Keep up the great work here!!

15 You need to refine that…

You need to refine that definition more though. Up 17 points but at what juncture of the game? Is it the entire fourth quarter?

 

What happens if the 17 points resulted from random things like a pick six, a fumble recovery for six, or big special teams play? Is it still considered garbage time? Furthermore is it contingent on what quarterback and offense you're facing? I would be terrified of Patrick Mahomes even if I were up 17 points with 8 minutes left.

 

I guess I'm more reasonable metric would be something like win probability percentage, but that has some flaws as well.

80 College football talent…

College football talent mismatches are way bigger than those in the NFL. A 17 pt 4th quarter lead is no guarantee of victory, unless it's in the final two minutes. There have been games where teams have scored 28+ points in the 4th quarter.

14 At what point do we start asking what happened to Carson Wentz

Two years into his career he looked like the next great QB and since then just looks average at best. Week one he was the worst rated QB in the NFL in DVOA, DYAR and QBR.

Maybe it was Frank Reich being there, who knows? He's never had many weapons BUT he's always had one of the best offensive lines in the NFL but that's not the case this year. 

103 Settle down...

...Lil' Cove.

Wentz had an injured O-Line and tried to force things.

Honestly, I'm wondering how Philly has been cursed with injuries so much the last few years.  They were down to a QB turned receiver in the playoff game.

Last year he was making chicken salad out of chicken sh!t, so I blame the team around him more than the QB.

16 Early season small data sample size is amusing

Look at the delta between the #1 offense and the #2 offense. 39.9-38.6=1.3.  Okay.

Now look at the same delta for the 1/2 defenses:  69.4-39.0= 30.4.  Holy Christmas, that's roughly the same as having the WA defense being as good as the Steelers, Pats, and Bengals on the field all at the same time!  (yeah, I know, that's not how it works... if nothing else, they'd get flagged for 33 men on the field every single play.)

But still, not sure that's sustainable.  ;-)

 

104 Don't forget...

...it was Ryan's defense that won the '85 SB, much more than anything else.  I never thought Ditka was a quality coach and that is probably the best empirical evidence of such.

You could have had Rich Kotite running the '85 bears and still won...

 

 

19 Very much confirmation bias…

Very much confirmation bias on my part, but I'm very unsurprised that NE's offense looks better than TBs. 

I remain very convinced that Tom Brady was the vast majority of the problem last year.  I will be very surprised if things go well in Tampa. 

35 Confirmation Bias

At the risk of being that guy (and let's be honest, I really am that guy).

Pre-season predictions being what they are it's hard to know what the difference means. Sure Miami looks on course for 4-11 and NO on course for 12-4, but time will tell. I neve dreamed the Titans would win 9 games last year and the Panthers would only win 5.

What we can say for certain was that Brady was bad week 1, every bit as bad as guys like Hoodie (a harsher critic of him then I am) thought that he would be based on last season's performance.

36 I didn't see the game but…

I didn't see the game but the general gist I got from people was that a lot of it seemed like poor timing and communication between him and his receivers. It was Drew Brees who looked shot according to people.

42 The problem is that "poor…

The problem is that "poor timing and communication" is what last year looked like. It was just "rookies and bad receivers" , the line, etc. 

 

And yet Cam looks like he's on the same page with them, and Brady is having the same issues as last year. 

 

Brady's game has always been predicated on being able to work through a pretty complicated route tree very quickly, and then make and execute decisions quickly. He didn't look like he could do that last year.  It's one game and covid and all that - but it's not a good sign that the problems look similar. 

Brady in the second half of the season last year had a 55% completion percentage - but still didn't turn the ball over much. The pick sixes in each of his last 3 games are fluky, but that's an awful strong deviation from a guy who averaged one every 2 years and was known for his decision making.  

Lots of caveats, etc. But he was really bad last year, and he's old. 

45 Decision Making

This is right - If you carefully watched Brady over the course of his career, last year it did not appear that his physical had waned badly. Instead it appeared that his trademark decision making and field vision was starting to go. One aspect of middle age that is not well discussed is the tendency to find it more difficult to form new relationships and for those relationships to take longer to develop, especially at the level of trust. I find myself wondering if Brady's decline has more to do with his aging mind then it does to do with his aging body.

This perspective is why I have been hesitant to blame him for the collapse of the 2019 Pats and did not predict a flop in Tampa. The evidence on the field last year might simply have been his idiot WR's, and it's possible he gets it figured out in Tampa - he still has enough physical tools to run his game the way he has for the past 5 years as far as I can see. That said, the evidence for decline in his mental game is mounting.

47 Physical decline also…

In reply to by sbond101

Physical decline also affects your timing. Because you need to put more of your available effort into a pass of the same velocity profile, it takes you longer to load up and decreases your accuracy. That half-beat of hesitation can be the difference between six and pick-six.

Manning talked about it; how after his last injury, he struggled for weeks to recalibrate which throws he could still make and which he couldn't, and when he needed to make them. You basically have to relearn your reflexes.

67 This was what I saw with Dan…

This was what I saw with Dan Marino.  In his final season, it seemed like suddenly the passes that would just zip through the defenders were instead getting intercepted.  The passes still looked good, but it's as if it took him an extra fraction of a second to make the throw.

46 And yet Cam looks like he's…

And yet Cam looks like he's on the same page with them

Cam threw for 155 yards against the worst defense in the league. NE rushed for 217 yards. Non-Cam runners were 27-142. He had the luxury of taking whatever he wanted against a defense that couldn't stop anything.

Teams with large turnover are going to struggle early this year, with no preseason and restricted practice time.

We'll know more about Cam in the next six weeks.

55 I'm happy to be corrected by…

I'm happy to be corrected by people with better stats / access to the all-22's, but from the roughly 75% of the MIA game I saw, it looked to me like NE's receivers mostly just weren't open.

It's just one game and NE never really needed to move away from it's run-first approach, so who knows what the rest of the season brings.  My guess, however, is that NE's going to run the ball a lot because they'll have to.  Unless teams sell out against the threat of the run, I doubt NE will have sustained success passing.  Not because of Cam, but because no one can stretch the defense and no one besides Edelman can create separation.  I suspect we'll see more use of TEs as the season goes on, but they're rookies, so that'll be back half of the season or possibly not until next year.  Until then, I suspect we'll see Cam throwing a lot of balls to a fully-covered Harry in the hope that Harry can outmuscle the DB for the ball.

The SEA game this week will be interesting.  If SEA gets the lead, we'll see if NE has the ability to respond. Conversely, if NE has success running the ball early, will Carroll and Schottenheimer be able to resist their own tendencies to try and run the ball, too?  SEA needs to leave the ball in Wilson's hands and not let Belichick suck them into a three-yards-and-a-cloud-of-dust game. 

72 As a Seattle fan I'm a little terrified of New NE...

Seattle blitzed A LOT against Atlanta.  With their current personnel Seattle probably has the first major scheme change on defense since the Pete Carrol era started in 2010.  A blitzing style of defense that minimizes pass rush weaknesses on the defensive line and confuses the type of Shanahan variant West Coast offenses that the Rams and Niners play. 

With that, I'm afraid of seeing a rehash of the early days of the 2010-2013 era when the early adoption of zone read option abused the type of Biltzberg derived defenses being run in Pittsburgh, Green Bay, and other locations.  Guys like Kaepernick having 300/200 days and Tebow winning a playoff game on 50% completions.  I doubt this year's Seattle team can create meaningful pass rush or run push without blitzing... and blitzing can be DESTROYED by option and RPO attacks.

84 I agree, blitzing would seem…

I agree, blitzing would seem to be a bad strategy for SEA versus NE.  This is a game, however, where if SEA's DL pressure is as weak as advertised, it still shouldn't hurt them.  If they play patient, and their secondary is as good as anticipated, SEA should be able to contain the NE passing attack without generating pressure.

If NE's OL is as good as it looks on paper, pressure may be hard for most teams to get consistently on Newton.  Blanketing the receivers and out-waiting him will probably be the best counter.  Will be interested to hear what Ben Muth has to say about NE's OL

112 Should you be?

It's a good early season test.

I don't think NE is all that good, honestly.  Decent defense but horrible offense.

If Seattle can't beat them in Seattle, then you aren't winning the NFC West and my Cards will have a much easier road.

I would not be worried, nor should you.  A win will confirm what we thought.  A loss at least tempers overconfidence.

Seattle should win this by 14 at least.

 

 

61 Cam's 155 yard day was a…

Cam's 155 yard day was a better day than Brady has had with the Patriots since about week 7 last year. 

 

Tom Brady was an awful quarterback the second half of last year. Cam Newton was completely acceptable - which is a huge upgrade. 

 

Might I remind you that in the last game of NE's regular season last year against same said defense, Brady went 16/29 for 220 yards and a pick-6 and a -48 DYAR?

71 Considering Brady's age, and…

Considering Brady's age, and his sharp decline in the second half last year, I'm not sure how relevant that game is at this point. 

 

Sure, it happened, but he's not 30. It's very possible that he had very different physical capabilities in those games. 

He was clearly a much better player early in the season than late in the season. 

 

It could just be random/sample sizes/etc, but considering his age, and things that were very visible to people watching the game (like how long he was taking to throw) - or it could be transient injuries - but I wouldn't bet on it. 

73 Time Will Tell

Pats receivers definitely struggled to get open. That's even with the real threat of runs loading the boxes against them.

At the same time, having Newton back there provides more space/time for receivers on routes. He doesn't need to be as accurate as prime Brady because of that.

Newton went 15/19 and the incompletes included an Edelman drop, two tipped balls and what was basically a throwaway. He was either told or decided to throw only when guys were wide open. So there is definitely more room there for him pass more.

It's funny: Belichick has been moving toward a power running game the past few years, and now he has the perfect QB for it. I think he decided that there was no way he was going to keep up with Mahomes and the other pass happy offenses (with an aging Brady) and that the only way to stay competitive was to play keepaway. Hence the resources devoted to RBs. On the flip side, he's invested heavily in the secondary as well in order to keep up with the demand for nickel and dime looks.

38 well

I believe Arians when he says Brady's two INTs were the results of mental mistakes as opposed to anything else.  And he should fit in with the Bucs' receivers better than he would with the NE run-based offense.  But people also have to come to terms with his arm strength diminishing.  

I don't think he was "the vast majority of the problem" since his receivers were also bad.  But he was a significant part of the problem.  I expect him to have better games than he did on Sunday.  But also Cam is a better fit for the Patriots right now.  

44 I think where we disagree is…

In reply to by RickD

I think where we disagree is his receivers - I don't think they were any worse than a typical Patriots low year. I think there have been a lot worse years, and Brady has been much better (hey, Reche). 

I watched a good amount of all-22 last year, and I saw an awful lot of teams blanketing Edelman and largely ignoring the younger guys - to the point that they were often uncovered, or had several yards on defenders. 

 

And I saw an awful lot of Brady having plenty of time, but not making it down that far in his progressions - often spending 2-3 seconds just staring at Edelman despite him being bracketed. These are things I'd never seen him do before. (He did have a bit of a habit of staring down Gronk, but not this bad). 

 

I'm not sure if he couldn't do the progressions fast enough, or he just didn't trust the young guys, or what - but I'm not sure it matters. 

 

 

105 Young Hoodie...

...two different scenarios.

Brady running a new offense with unfamiliar receivers outside of the Gronkster, plus NO had continuity because they are becoming a machine.  

NE's offense was horrible.  What game were you watching?  They didn't score many points against a rebuilding Dolphins team and looked horrid when they were moving the ball.  When the competition improves, they will fall back.  Did you forget about the second half of last year?  This defense is even less talented than that one was.

21 Laughing So Hard it's Almost Painful

Listen I get it-- the 2019 Packers were anything but a 13-3 team. And DVOA played a leading role in showing just how and why.

But in a game where Packers take a 29-10 lead-- on the road-- and then basically exchange scoring drives with the Vikings from then on, to see DVOA say Minnesota was the better team is simply about the funniest thing I have ever read.  OK-- I'll give you this, Aaron-- you admit you don't know how to measure one team dominating time of possession. Wake me up in October when perhaps this metric might make some sense. But for now, I can simply laugh about it. 

27 You know, the Lions had a…

In reply to by oaktoon

You know, the Lions had a better record against the Packers than the either of the other two NFC North teams this past decade. They also broke the 24-game Lambeau losing streak. Something might be in the air...

50 Matt Patricia:  0-9 against…

Matt Patricia:  0-9 against Minnesota and Chicago, 2-2 against Green Bay (and the two wins weren’t particularly close, the two losses were by razor thin margins).

However, the assigned referee for Sunday is Clete “hands to the face” Blakeman.  I feel like the NFL is trolling the Lions at this point.

29 Ive made data science models…

Ive made data science models, including ones related to football. There are always cases, especially in complicated problems, that produce predictions that look odd and hard to understand. Only you manage to see them with GB every week.

33 Reverend Bayes

I was very interested to see Aaron use "priors" casually, assuming most readers will know what he's talking about. ("combined, let's be honest, with our priors about the current state of their locker room and the head coach."). FO's statistical model is fundamentally frequentist, but the constant reminder that you should combine what DVOA "says" with all the other information you have at your disposal, especially what you see with your own eyes watching games, is effectively saying you should combine the likelihood distribution given by DVOA with your own priors to get the best posterior distribution possible. This would be a neat way to explain the Bayesian/frequentist landscape to a certain type of student.

Oh, and Jags-Squirrels Super Bowl, baby! Bring it on!

34 Just curious, are you…

In reply to by TomC

Just curious, are you inferring it's a frequentist model because it was created in a time when Bayesian work was mostly done by academics and was probably beyond what FO was capable of in 2003? 

I have no idea how it works but my first stab at it would have assumed it was some kind of Vector Autoregression model or maybe a series of independent armax models.

76 At the most basic level, I…

At the most basic level, I mean that the parameters of the model are determined by minimizing some distance between the model predictions and the data (or maximizing some likelihood-like statistic), with no priors explicitly applied in the process. I'm not saying anything about exactly how that statistic is minimized (i.e., how the regression or fit is performed), just that the statistic is not modified based on some prior knowledge.

And, to be clear, that is just my understanding of the process, and I could be completely wrong.

95 I learned to be a better…

In reply to by TomC

I learned to be a better statistical thinker from reading this website than I did in 9 years of undergrad and graduate engineering / computer science education (that included a fair bit of simulation and modeling).  

37 SoFi Stadium

Similar design to baseball's stadium in Seattle, which has an "umbrella" roof. Not enclosed.

41 DVOA oddities..

You almost need a way to account for coaching style.  Anyone who watched the HOU @ KC game saw that Houston was outmatched on both sides of the ball.  Reid is notorious for going ultra-conservative with big leads even before the 4th quarter. 

While I'm sure it gets washed out over the course of a season, because those big lead games aren't super common in the NFL, I do find it very interesting how those ultra-conservative, clock killing drives on both offense and defense can skew whole game statistics.

Have you all ever looked at historical DVOA of individual coaches by score differential?  I'd be willing to bet a metric like that would generate a pretty good indication of coaching philosophy.

108 Let's not get carried away, Tooster.

In reply to by cstoos

The Texans slowed the KC offense down in the first few series.  Andy Reid made some adjustments and Bill O'Brien "the coach" couldn't match up.

That is what happened.

That was a game where a better coach could have possibly schemed his way to a closer game, maybe a win, but O'Brien to me is lackluster as a coach, almost as bad as he is a GM.

It was ironic to me how badly they missed the threat of Hopkins when you see what a difference maker he is for Arizona.  

With Hopkins, Arizona is a playoff team.  A plug-n-play USB #1 WR.  

43 Sad

Wow, not much improvement from Miami's dreadful performance opening day a year ago. 0-6 start seems very possible. 

49 This is just more classic…

This is just more classic front-running Green Bay. Get out to a big early lead, sure the defense makes a couple key plays along the way, 4th quarter comes along and they get conservative on offense (maybe not so much in this particular game) and the D gives up huge chunks of points. Fans chalk it up to playing soft with a big lead and hype ourselves up that they're SB contenders.  Then when it's actually a close game in the 4th quarter against a top team, it happens again. Then come playoff time we give up 37 in the divisional round or NFCCG.

57 I suspect this is why DVOA's…

I suspect this is why DVOA's predictability isn't improved by removing more "garbage time".

A big half-time lead that ends up in a 37-6 game likely does provide additional information about the teams than a game that ends up 37-24.  What happens in the late 3rd, early 4th quarters may in fact provide an insight into how replicable in the future were the events that happened to create the early lead.

69 It occurs to me that the…

It occurs to me that the real problem with accounting for garbage time isn't the winning team, it's the losing team.

DVOA simply has no way of parsing whether a sudden offensive and/or defensive improvement is real or not -- whether you're looking at GB's 13 point dead-cat bounce in the 4th or KC's 21 point comeback. From a play-by-play perspective, they look similar, because DVOA doesn't know the DBs are playing a disinterested prevent to run out the clock and not suddenly being picked apart by an MVP.

58 Lord you guys are spoiled. …

Lord you guys are spoiled. 

GB did in the Seattle game what they did in the Minnesota games - front run and then coast across the line. SF is just a bad matchup for them. Green Bay was 3-2 against the other semi-finalists last year; 3-0 against non-SF teams. They are a SB contender.

60 Coast Across the Line?

Maybe on defense-- Rodgers and Co kept scoring-- LaFleur was noticeably less conservative than Mccarthy had been... 29-10  and the game went  MINN TD +2  GB  TD  MINN TD +2 GB TD  MINN TD +2  Onside kick recovered GB and Game over....

75 Rams worse than Cowboys

I've come to avoid the FOMBC, but I just don't see why the Rams ended up being seen as worse in the LAR-DAL game. I understand that special teams are independent, but the cowboys also had bad special teams. The game I watched, and of course I'm biased, the Rams really were just better. Better and more consistent on offense, and better and more consistent on defense. 

Is it possible that FO underrates teams that successfully run the ball? Because the rams were very successful running the ball, but if even successful runs are considered not particularly high DVOA then it's possible that we could see part of our discrepancy.

Actually, I would be fascinated to see what DVOA would think of a team that simply ran the ball every single play for exactly 4 yards. We could see that such an offense would be perfect, scoring on every drive, but what would DVOA think of such a team? I would be fascinated to see where they come out.

79 I don't think overall DVOA…

I don't think overall DVOA rating makes a distinction between runs and passes; all that matters is yards gained for each down and distance.

As for your hypothetical team, yeah their offense wouldn't be ranked very high. No turnovers is excellent, but needing to convert a third down on every single series is highly inefficient. 

81 It truly is fascinating to…

It truly is fascinating to me that such a team would actually have a mediocre DVOA, despite scoring on every drive. While I am strongly in favour of pass-heavy offenses, I can't help but wonder if perhaps I've gotten slightly ahead of myself condemning the run.

86 Catch-22

If your hypothetical team was to become true and able to sustain that success then the  DVOA model would be broke. Because that level and type of success has never been repeatable. Hence why it’s just hypothetical 

82 This little thought…

This little thought experiment intrigued me, so I looked it up. Using 2019 data...

Plays that gained exactly 4 yards on first-and-10 had an offensive DVOA of -13.4%.

Plays that gained exactly 4 yards on second-and-6 had an offensive DVOA of 3.1%.

Plays that gained exactly 4 yards on third-and-2 had an offensive DVOA of 93.4%.

Since this team would be getting to third down every single time, we can just average those three together and get a DVOA of 27.7%. For comparison's sake, Baltimore led the league last year with a DVOA of 28.2%.

If that seems low to you for a "perfect" offense, DVOA doesn't "know" that this team will get exactly 4 yards every time. It assumes that after getting a first down, that team will then behave like a normal offense -- which means eventually they will get a penalty or a drop or fumbled snap or something that will set them back. Or, even if they do get exactly 4 yards every single play, at the end of the half they will run out of time. This is why a 20-yard gain on third-and-2 is worth more than a 4-yard gain on third-and-2. 

83 What is the DVOA for a 3.5…

What is the DVOA for a 3.5 yard gain every play?

We tossed about this thought experiment a few years ago. I think it was the Leroy Hoard Scenario.

"If you need one yard, I'll get you three. If you need five yards, I'll get you three."

 

100 DVOA for 3.5 yards per play

The DVOA for exactly 3.5 yards on every play would be whatever the DVOA is for a 4-yard gain on first down, a 3-yard gain on second down, and a 4-yard gain on third down, as that's how the official play-by-play would score them. 

(After the 3.5 yard gain on third down, the referees would spot the ball on the next whole-number yard line to begin the next series of downs, so the scoring sequence would continue to be 4-3-4 rather than changing to 3-4-3.)

85 The penalty impact is a big…

The penalty impact is a big one.  Far from being "perfect", this offense would never score on any drive in which it incurred a penalty before entering FG range.  I also agree with the point about being able to score on end-of-the-half drives.

The ability to overcome penalties and negative plays and still sustain a drive is a definite real life advantage.  Incurring those penalties and negative plays, of course, is a counter-balancing weakness in the offense.  And the ability to score quickly with little time on the clock is another factor in favour of quickstrike offenses.

The major counterbalancing effect is that the highly consistent offense limits the opposition's number of possessions.  I think the evidence, however, has shown that by also limiting your number of possessions, this effect ends up being a neutral factor.

Also, it's probably worth pointing out that DVOA doesn't favour the pass over the run.  If runs routinely generated 10+ yard gains the way passes do, DVOA would like runs just fine.  DVOA dislikes the 3-yard pass on 1st down every bit as much as it dislikes the 3-yard run.

88 The difference between pass and run

There is a difference between pass and run in the DVOA system, in the "baseline available." Passes get a bigger penalty indoors than runs do. Passes and runs are treated differently as far as the baseline available in the fourth quarter depending on score and time remaining.

As far as how many "success points" a play gets -- the numerator in DVOA -- there's no difference.

91 Good point.  I forgot this…

Good point.  I forgot this type of offense wouldn't need to ever consider punting in these circumstances!

PS, that would boost their DVOA.  Converting 4th-and-3 consistently would surely be scored more highly than taking the 5 yard penalties.

92 Failed plays

Remember that a 4 yard gain on 1st down is a failed play by DVOA (well, not a successful play anyway). Not converting on a third down is also a failed play (even if it brings the team closer to the first down).

90 That actually seems quite…

That actually seems quite high to me; I guess being able to convert every single 3rd and short still outweighs everything else.

Now all we need to do is wait for a team to sign a 15-foot player that falls forward arms outstretched after getting the ball every time. 

116 "DVOA doesn't "know" that…

"DVOA doesn't "know" that this team will get exactly 4 yards every time."

I can't help but think that there is something here that is a little off. After all, running the ball is more consistent than passing. It's possible that better offenses intentionally choose lower variance playcalls, since they don't need big downfield plays in order to sustain long drives. This would also be true of the short passing game, and might fit with my subjective takes on NE passing, where they seemed to do just fine with tons of 4-6 yards passes all game long.

I think for a team that can actually consistently get those 4-6 yards gains, those 4-6 yards gains are actually more valuable. As someone who struggled through the Jeff Fisher 2016 Rams, I can say that team might as well have just thrown the ball downfield every play, since they simply were not stringing together enough short yard gains to sustain a drive. Hell, getting a first down was an ordeal, so for them, a 20% chance of completion 20-30 yard pass was actually an improvement in offensive efficiency. In contrast, such a play would be horrible for the uber-efficient NE offenses of years past.

I guess what I'm saying is that we might have to add a little something to teams where low-variance is considered good, if the team is already good. That might paint a more accurate picture than simple aggregated success. Maybe Aaron already does this?