DVOA Analysis

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

Week 5 DVOA Ratings

by Aaron Schatz

We have a new No. 1 team in the Football Outsiders DVOA ratings this week. The San Francisco 49ers had the best single-game performance of the season so far against Cleveland on Monday night (100.9% DVOA), and that catapults them into the top spot past the New England Patriots even though both the 49ers and Patriots see their actual ratings move up from a week ago.

Last week, I was writing about the Patriots and where their first four games stood among the hottest starts of the DVOA era, since 1986. Now we have to talk about the 49ers and where their first four games stand among the hottest starts of the DVOA era. The answer is: even higher.

Yes, the 2019 49ers now have the sixth-highest DVOA ever measured through four games. The Patriots now rank 13th... but they rank fifth all-time in best total DVOA ever measured through five games. And that list is even more impressive than the list of the best DVOA through four games, with pretty much every team on the list at least making it to the conference championship game. Whether or not the 49ers can make it onto the second list depends on how they play against the Rams this week. Most of the teams that were great through four games dropped a bit after their fifth game.

Usually I do these lists with 12 teams but I'll do these with 13 teams so that you can see where the 2019 Patriots were a week ago.

BEST TOTAL DVOA
THROUGH 4 GAMES, 1986-2018
  BEST TOTAL DVOA
THROUGH 5 GAMES, 1986-2018
Year Team DVOA   Year Team DVOA
2007 NE 73.4% x 1991 WAS 71.5%
1991 WAS 71.8% x 2007 NE 68.1%
2013 DEN 62.2% x 2006 CHI 56.3%
1999 STL 62.1% x 1999 STL 55.4%
1996 GB 60.8% x 2019 NE 55.1%
2019 SF 58.1% x 1996 GB 53.5%
2009 PHI 57.1% x 2015 ARI 51.5%
2015 NE 55.9% x 2013 DEN 51.3%
1989 CLE1 54.8% x 1986 CHI 50.9%
1998 DEN 54.3% x 2009 NO 49.1%
1992 PHI 54.2% x 1998 DEN 48.0%
2009 NO 53.6% x 2001 PHI 47.8%
2019 NE 52.8%   1994 DAL 47.6%

Both the 49ers and the Patriots are being driven by their defenses. The Patriots are still the best defense in DVOA history, with a slightly bigger lead on the rest of history than they had a week ago. The Patriots' defensive rating got a little worse this week as opponent adjustments got stronger, but the ratings for the other great defenses in dropped even more. Meanwhile, the 49ers rank fourth among defenses through four games.

BEST DEFENSIVE DVOA
THROUGH 4 GAMES, 1986-2018
  BEST DEFENSIVE DVOA
THROUGH 5 GAMES, 1986-2018
Year Team DVOA   Year Team DVOA
2019 NE -48.1% x 2019 NE -46.2%
1997 SF -47.5% x 1991 PHI -41.2%
2008 BAL -46.3% x 1986 CHI -38.6%
2019 SF -45.3% x 2002 TB -38.5%
1991 PHI -42.7% x 1997 SF -37.2%
1996 GB -40.0% x 1991 NO -36.5%
1989 CLE1 -36.8% x 1996 GB -34.6%
2006 BAL -35.6% x 2011 BAL -34.5%
2000 MIA -34.7% x 2008 BAL -34.0%
2011 BAL -34.4% x 1991 WAS -33.2%
2002 TB -34.2% x 2012 CHI -32.4%
1988 MIN -34.1% x 2015 DEN -31.9%

 

Obviously, strength of schedule is still an issue here. Right now, opponent adjustments are only at 50% strength. So I ran a version of DVOA with opponent adjustments at 100% instead, and the Patriots ended up with a defensive DVOA of -43.0%, which still would be the best ever measured through five games.

It's certainly reasonable to believe that this year's Patriots schedule is so extreme that there's no way we can measure it correctly, even if we apply our usual opponent adjustments. Part of the problem is just how bad Miami is. Overall, the Patriots have played the offenses ranked 26, 28, 30, 31, and 32. I'm pretty sure there's never been anything like that when it comes to easy schedules. However, when it comes to overall schedule strength, the Patriots aren't that much different from the 1999 Rams, who currently have the easiest season-long schedule strength in the DVOA database. The difference is just how bad Miami is at No. 32. When the 1999 Rams were 5-0, their five opponents ranked 15, 25, 29, 29 (Atlanta twice), and 30. The Patriots' five opponents rank 19, 20, 29, 30, and 32. That's really not much different, except that team No. 32 this year has been the worst team in DVOA history.

Meanwhile, San Francisco's schedule has been easy but more run-of-the-mill easy. Their four opponents rank 16, 20, 26, and 31 overall. On offense, their four opponents rank 18, 25, 28, and 29.

Below our top two, almost all the teams in last week's top ten dropped this week. Kansas City moved up a spot to No. 3 but its DVOA dropped from 27.3% to 22.1%. Dallas and Baltimore both dropped. Tampa Bay and Chicago dropped out of the top ten entirely. Moving up are Philadelphia (from 6 to 4) and especially Minnesota (from 15 to 7) and Houston (from 17 to 10).

Philadelphia had negative total DVOA in each of its first three games, but have single-game DVOA over 60% for both Week 4 and Week 5. Their Week 4 win, 34-27 over Green Bay, was a huge offensive performance, and then in their Week 5 win, 31-6 over the Jets, the offense was average but the defense was fantastic (even after adjusting for how bad the Jets offense has been this year). The Vikings at No. 7 is probably a bit of a surprise, especially since we're only a week separated from all the articles and discussion of how they were struggling with turmoil in the locker room. Their season has been very up and down. More specifically, the offense has been very up and down. The defense, which ranks fourth in DVOA, has been pretty consistent. The offense has been good in the wins and bad in the losses. It will be fun to watch these two teams face each other in Week 6.

The whole rest of the top ten is all packed together, far below the 49ers and the Patriots. We've never seen anything like the gap that currently exists between the top two teams and the rest of the league. The Patriots at 55.1% have the highest DVOA to ever rank No. 2 after five weeks. The Chiefs at 22.1% have the lowest DVOA to ever rank No. 3 after five weeks.

(In case you are curious, the previous record-holder for best No. 2 team after five weeks was the 2009 New Orleans Saints. The previous recordholder for worst No. 3 team after five weeks was the 1987 Cleveland Browns.)

One other team I want to address is the Buffalo Bills. The Bills have been very good on defense this year, currently ranking third behind the Patriots and 49ers. They've also been an incredibly one-sided team this season. The Bills are down at 27th on offense and 31st on special teams. It all combines to make the Bills 19th in DVOA. They're 22nd in DAVE, which takes into account our preseason projections.

And yet Buffalo right now ranks fifth in playoff odds, making it to the postseason in 70% of our simulations. Other simulations available on other sites (for example, FiveThirtyEight) have similar results. It's really nice to bank some close early wins when your conference is in a down year. Right now, only seven teams in the AFC have winning records. The Bills already have a one-game lead on any other competitors for the wild card, and only Pittsburgh has an easier remaining schedule by average DVOA of opponent. I don't think we can pencil in the Jets as an automatic win now that Sam Darnold is back, but we can pencil in Miami and Washington as pretty sure wins. The Bills still get to play Miami twice, so those three wins put them at 7-1. All they have to do is go .500 in their remaining games to finish 11-5. And two of the most likely losses are to NFC teams (Philadelphia and Dallas) that won't matter in a tiebreaker.

Right now, the Bills make the playoffs in 99.9% of simulations where they finish 11-5. They make the playoffs in 96.3% of simulations where they finish 10-6. And they make the playoffs in 62.1% of simulations where they finish 9-7. Heck, they even make the playoffs in 10.3% of simulations where they finish 8-8.

* * * * *

Stats pages should now be updated through Week 5, including playoff odds, the FO Premium DVOA database and snap counts.

* * * * *

These are the Football Outsiders team efficiency ratings through five weeks of 2018, 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.

Because it is early in the season, opponent adjustments are only at 50 percent strength; they will increase 10 percent every week through Week 10. As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.

DAVE is a formula which combines our preseason projection with current DVOA to get a more accurate forecast of how a team will play the rest of the season. Right now, the preseason projection makes up 45 percent of DAVE for most teams (55 percent for teams with just four games played).

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
LAST
WEEK
TOTAL
DAVE
RANK W-L OFF.
DVOA
OFF.
RANK
DEF.
DVOA
DEF.
RANK
S.T.
DVOA
S.T.
RANK
1 SF 58.1% 2 25.8% 2 4-0 15.0% 4 -45.3% 2 -2.2% 24
2 NE 55.1% 1 37.9% 1 5-0 9.8% 8 -46.2% 1 -0.8% 16
3 KC 22.1% 4 19.6% 3 4-1 26.9% 2 2.8% 18 -2.0% 23
4 PHI 21.6% 6 16.0% 4 3-2 12.7% 6 -8.7% 6 0.2% 12
5 DAL 20.7% 3 14.3% 6 3-2 27.2% 1 2.9% 19 -3.6% 27
6 BAL 14.6% 5 10.8% 8 3-2 14.0% 5 9.4% 24 10.0% 1
7 MIN 13.7% 15 8.6% 10 3-2 2.3% 16 -12.0% 4 -0.6% 15
8 GB 12.8% 9 8.9% 9 4-1 6.8% 9 -6.1% 9 -0.1% 14
9 SEA 12.6% 11 11.2% 7 4-1 19.6% 3 5.2% 22 -1.7% 22
10 HOU 8.3% 17 8.2% 11 3-2 3.8% 13 -4.5% 11 0.0% 13
11 DET 7.5% 10 5.2% 13 2-1-1 1.2% 17 -4.0% 12 2.3% 8
12 NO 5.6% 21 14.7% 5 4-1 3.2% 14 4.1% 20 6.5% 2
13 JAX 5.3% 18 -1.5% 18 2-3 11.4% 7 9.8% 25 3.7% 6
14 CHI 4.9% 8 4.4% 15 3-2 -9.8% 26 -10.7% 5 4.1% 4
15 CAR 3.4% 13 3.0% 16 3-2 -4.9% 21 -7.1% 8 1.2% 9
16 TB 2.4% 7 -4.3% 21 2-3 0.6% 18 -3.4% 13 -1.7% 21
17 LAR 1.3% 12 7.9% 12 3-2 4.3% 12 1.9% 17 -1.1% 20
18 OAK -0.4% 25 -3.6% 20 3-2 4.6% 11 4.2% 21 -0.9% 17
19 BUF -2.2% 20 -5.4% 22 4-1 -11.4% 27 -15.4% 3 -6.2% 31
20 PIT -2.8% 24 -1.2% 17 1-4 -9.2% 25 -7.5% 7 -1.1% 19
21 LAC -5.9% 19 4.6% 14 2-3 2.6% 15 10.8% 28 2.4% 7
22 DEN -7.1% 27 -9.2% 26 1-4 -0.2% 19 -0.1% 16 -7.0% 32
23 IND -7.4% 26 -6.7% 23 3-2 6.4% 10 10.2% 26 -3.6% 26
24 TEN -9.3% 16 -2.5% 19 2-3 -8.7% 24 -3.2% 14 -3.9% 28
25 NYG -13.3% 22 -15.8% 27 2-3 -5.0% 22 9.4% 23 1.2% 10
26 CLE -14.2% 14 -7.5% 24 2-3 -19.8% 29 -0.6% 15 4.9% 3
27 ATL -16.3% 23 -8.6% 25 1-4 -1.3% 20 10.4% 27 -4.7% 30
28 ARI -21.9% 30 -17.1% 28 1-3-1 -6.9% 23 12.8% 30 -2.2% 25
29 WAS -32.3% 29 -25.0% 30 0-5 -24.0% 30 12.0% 29 3.8% 5
30 NYJ -37.9% 28 -19.1% 29 0-4 -43.3% 31 -5.0% 10 0.3% 11
31 CIN -39.8% 31 -26.6% 31 0-5 -19.5% 28 19.4% 31 -0.9% 18
32 MIA -90.9% 32 -57.0% 32 0-4 -46.1% 32 40.8% 32 -4.1% 29
  • NON-ADJUSTED TOTAL DVOA does not include the adjustments for opponent strength or the adjustments for weather and altitude in special teams, and only penalizes offenses for lost fumbles rather than all fumbles.
  • ESTIMATED WINS uses a statistic known as "Forest Index" that emphasizes consistency as well as DVOA in the most important specific situations: red zone defense, first quarter offense, and performance in the second half when the score is close. It then projects a number of wins adjusted to a league-average schedule and a league-average rate of recovering fumbles. Teams that have had their bye week are projected as if they had played one game per week.
  • PAST SCHEDULE lists average DVOA of opponents played this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
  • FUTURE SCHEDULE lists average DVOA of opponents still left to play this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
  • VARIANCE measures the statistical variance of the team's weekly DVOA performance. Teams are ranked from most consistent (#1, lowest variance) to least consistent (#32, highest variance).
RK TEAM TOTAL
DVOA
W-L NON-ADJ
TOT VOA
ESTIM.
WINS
RANK PAST
SCHED
RANK FUTURE
SCHED
RANK VAR. RANK
1 SF 58.1% 4-0 58.8% 5.0 1 -13.6% 28 -2.4% 20 8.4% 13
2 NE 55.1% 5-0 65.8% 5.0 1 -33.2% 32 -10.1% 28 11.8% 21
3 KC 22.1% 4-1 21.9% 3.2 8 3.9% 13 5.4% 9 14.1% 24
4 PHI 21.6% 3-2 29.0% 3.6 4 -13.3% 27 -2.2% 19 21.4% 28
5 DAL 20.7% 3-2 34.3% 3.7 3 -23.6% 31 3.6% 11 8.6% 14
6 BAL 14.6% 3-2 22.1% 2.8 12 -21.6% 30 -0.1% 17 17.0% 26
7 MIN 13.7% 3-2 19.5% 3.6 5 -2.5% 22 5.9% 7 13.9% 23
8 GB 12.8% 4-1 6.1% 3.1 9 10.8% 6 5.9% 6 10.4% 18
9 SEA 12.6% 4-1 18.6% 3.3 7 -11.5% 26 11.0% 1 4.7% 6
10 HOU 8.3% 3-2 14.8% 3.0 10 -1.6% 20 5.3% 10 3.1% 4
11 DET 7.5% 2-1-1 12.5% 3.0 11 4.0% 12 2.7% 14 3.8% 5
12 NO 5.6% 4-1 4.5% 2.3 18 9.1% 7 0.6% 16 9.5% 17
13 JAX 5.3% 2-3 -2.5% 2.7 13 3.5% 14 -9.8% 27 5.1% 8
14 CHI 4.9% 3-2 8.2% 2.6 16 -2.7% 23 8.5% 2 8.1% 11
15 CAR 3.4% 3-2 5.0% 2.7 14 -0.9% 19 1.4% 15 4.8% 7
16 TB 2.4% 2-3 6.8% 2.3 20 11.0% 5 -2.6% 21 6.2% 9
17 LAR 1.3% 3-2 1.2% 2.7 15 1.9% 16 6.0% 5 11.7% 20
18 OAK -0.4% 3-2 -3.4% 3.5 6 5.2% 11 -4.5% 23 31.3% 32
19 BUF -2.2% 4-1 3.3% 2.0 24 -9.1% 25 -14.9% 31 2.0% 3
20 PIT -2.8% 1-4 -7.4% 2.2 21 20.1% 2 -19.9% 32 17.4% 27
21 LAC -5.9% 2-3 -2.5% 2.2 22 -17.9% 29 5.5% 8 8.8% 16
22 DEN -7.1% 1-4 -5.4% 2.0 25 3.3% 15 3.1% 13 6.6% 10
23 IND -7.4% 3-2 -10.1% 2.1 23 -2.0% 21 -6.5% 25 8.2% 12
24 TEN -9.3% 2-3 8.8% 2.4 17 -7.0% 24 3.1% 12 16.8% 25
25 NYG -13.3% 2-3 -19.5% 1.5 26 0.4% 17 -3.5% 22 22.7% 30
26 CLE -14.2% 2-3 -9.7% 2.3 19 5.3% 10 -11.4% 29 27.6% 31
27 ATL -16.3% 1-4 -27.2% 1.2 27 5.4% 9 7.1% 4 10.4% 19
28 ARI -21.9% 1-3-1 -22.9% 1.2 28 -0.4% 18 8.4% 3 2.0% 2
29 WAS -32.3% 0-5 -37.4% 0.0 31 17.8% 3 -0.6% 18 22.4% 29
30 NYJ -37.9% 0-4 -43.7% 0.6 30 15.1% 4 -14.8% 30 8.8% 15
31 CIN -39.8% 0-5 -45.6% 0.7 29 8.8% 8 -6.3% 24 13.0% 22
32 MIA -90.9% 0-4 -93.3% 0.0 31 21.1% 1 -9.5% 26 0.5% 1

Comments

127 comments, Last at 11 Oct 2019, 12:34pm

1 Wow, is the Norris division

Unlike the AFC East, the black and blue division is competitive this year.

All four teams are in the top half of DAVE and clustered around 10-15.

Will be great to watch down the stretch.

2 So what are the chances the…

So what are the chances the Pates D is actually historically good? I haven't watched them much this year, but I have a hard time believing this rating. Someone here (ABGT maybe?) that when you're facing inept offenses, you can't even really evaluate a defense because none of their opponents have even been good enough to test it.

4 About as good as the Eagles…

About as good as the Eagles actually being a solid defense. Hint: they aren't.

Honestly this is going to be a question all year and beyond, and the Rams comparison Aaron made isn't right: bad offense *specifically* is a problem for evaluating defenses: if an offense screws up, it looks like good defense, even if the defense screws up too. The Patriots schedule in the end is going to be almost all bad offenses.

 

 

9 "About as good as the Eagles…

"About as good as the Eagles actually being a solid defense. Hint: they aren't."

I would agree with this if the question had been...what are the chances that the Patriots are the best defense ever. 

Historically good, which I guess means top 15 or so, is certainly realistic. It may not finish there, it might not even be the most likely outcome, but I don't think its some farfetched probability. 

19 Unsupported Conclusions

"About as good as the Eagles actually being a solid defense. Hint: they aren't."

This is garbage; Putting aside the question of the validity of the signal of over-performance against a set of crappy offenses - by your own standard there is insufficient evidence to draw a conclusion as to the quality of the Patriots defense one way or the other. At most you could say that there is a good chance the Patriots won't be historically good because most teams aren't historically good and we have insufficient data to say otherwise. What annoying is that we will have to put up with at least a few more weeks of litigating this issue as the Giants seem primed to get blown out on Thursday night, and we'll be no further toward resolving the matter - though we may well be further toward a historically high defensive DVOA.

 

 

21 I have a simple acid test…

I have a simple acid test for this. Pat can offer up some conditions he expects out of the Patriots defense this Sunday. As I said before, lets assume the Pats are the greatest defense ever. What would this giants team do on average? If your answer is literally 0 first downs, 20 incompletions, ypc of 1, and 10 turnovers...yeah ok. 

More realistically, set a threshold. I believe if this were against the 85 bears, 2002 bucs, etc etc...the giants would do what?

 

Me personally, if the Giants scored more than 21 points, I would actually ding the patriots quite a bit. As it is, I expect them to score 13 points or fewer. 

25 hey, let's just throw out DVOA

and use the 'eye test' instead!  

Me personally, if the Giants scored more than 21 points, I would actually ding the patriots quite a bit. As it is, I expect them to score 13 points or fewer. 

So...the same as every other team they've faced this season?  

The defense has given up 20 points in 5 games.  

I wouldn't be surprised to see the Giants held to less than 7 points.  

39 The Giants appear to be…

The Giants appear to be missing most of their best offensive players heading into this game.  Anything other than a complete demolition is probably a ding against any "best ever" talk.

47 I can't. Bad offenses are…

I can't. Bad offenses are too noisy. It's not a one to one thing anymore: bad defenses can get the same results too.

The schedule's a huge problem this year. In the end you'll only have like 4 or 5 real pieces of information on the Patriots and a few other teams. I would *not* like to be someone whose job depends on predicting outcomes this year.

54 Those defenses would…

Those defenses would outscore the Giants offense. That 'd be "holding serve" to remain in the argument.  Anything less would be a significant demerit (the argument's foolish, anyway; they haven't just played a series of bad offenses, they've played a series of bad offenses playing second- and third- string qbs, and this week again they're facing a seriously crippled opponent down their best two rbs and wr- seriously, Josh Allen's the best quarterback they've faced, and, while he's improving, he's still probably a bottom-ten guy)

123 Roethlesberger was decidedly…

Roethlesberger was decidedly worse than Mason Rudolph in the only other game he played (they each played a half) and it's been suggested he was already hurt before the regular season began; there's nothing to suggest the Pittsburgh offense was ever any good this year

70 Actually, I figured it out…

Actually, I figured it out.

If the Giants offensive line holds up and gives Jones enough time and multiple wide receivers that we know are at least league-average can't get open downfield on even, say, half of those plays, then that would totally vindicate that the Patriots are in the top half of "best defenses in a year" of the modern era. Wait, what's that? The Giants don't have multiple league-average wide receivers? Huh.

Realistically, what do I expect? The Giants are a bad offense at best and they're (likely) missing their 2 leading rushers and receivers. They're basically playing a college offense, and that's not even an exaggeration since I'm pretty sure their starting QB, RB, and WR will all be rookies a year out of college.

20 Yes, Pat continues to…

Yes, Pat continues to stretch a reasonable skepticism about NE's performance beyond the point of necessity or even recognition.  Why not just say it's too early to tell and leave it at that?

23 Because there's a *huge*…

Because there's a *huge* prior against it. To be clear, I interpreted "historically great" as "best in history." They're on track to shatter the points per game record, I think, for instance. If you want to take it down to say "top 15 since 1980" that's a much weaker criterion.

But for First"best" you've already got the prior against "best in history." You'd ballpark around a ~1% chance of them being top in history, just based on numbers. Then add the fact that in order for them to be *really* top 15, the fact that they've played the worst set of offenses in the past ~30 years and that's just random is another 3% chance. The combination of those two is far more unlikely than the alternative, which is that the two results are correlated.

32 There's certainly a huge…

There's certainly a huge prior against best ever, less so against *one of* the best ever, which is how I think everyone else interpreted the question.  Especially when you are starting with a five game sample that is tentatively moving in that direction.

The null hypothesis is already on your side, you don't need to make any additional claims.  "Let's wait and see" is enough for now.

49 The only point I've been…

The only point I've been making is that in my opinion we don't have a 5 game sample of the Patriots. We have a 1 game sample. The Jets, Dolphins, and Redskins games are useless. The Steelers game is a giant question mark thanks to the timing. Literally the only game we have is the Bills game.

I mean Adrian Peterson's quote just totally says it. The Patriots defense *didn't have to try*. Not because they were so great or anything. Because they knew the Redskins couldn't play football.

119 Why did you change …

Why did you change "historically great" to "historically good"? Clearly, you are axe ginding!!!!!

OK, now that the mind-reading part of the post is over, I'll simply note that when I read or use "historically great", I interpret the term as identifying a player or group of players as being reasonable candidates for the best ever.

 

 

45 Just for ha has, I'll offer…

Just for ha has, I'll offer a more legitimate response.

I haven't the slightest problem with someone doubting the historical greatness of NE's defense.  By sheer numbers, being a top 10 group is incredibly hard to penetrate.  It'll only happen once every five years and only one of 32 teams will pull it off.  Even among the select group of teams that are in there after five games, most tail off to be just "ordinary" good/great.  

The null position is that NE won't be historically great by season's end, so "let's wait and see" captures pretty much all forms of a doubting arguments.  Despite this, Pat keeps going too far and adopting wholly unnecessary burdens.

What I find funny is that my response to Pat would be the same regardless of what team we are discussing.  ABGT, on the other hand, wouldn't bat an eye if we had been talking about any team other than the Patriots.  So the great irony is that it is those complaining about Patriots sensitivity that are often the ones engaging in it!  

52 You're really not…

You're really not understanding what I'm saying. I'm not criticizing the Patriots defense. I'm not saying they haven't done enough. I'm saying I literally have no context for understanding their performance because no one has ever faced a slate of offenses that bad. Ever. Which means any attempt to compare them to other teams in other seasons, right now, is an extrapolation.

The problem is that my instinct is that this will not change at all over the year.

55 It's like trying to measure someone's height

...with tape measure that's only 6' long.

All we know for sure is that the Pats are over 6'0". We can't tell if they're 6'3", 6'5" or 7'0", bc the offenses we're using as a measure are all so very short.

Like Pat is saying, it's very unlikely they're a sigma 7ish 7'0".  (But we can't rule it out!)

 

63 "You're really not…

"You're really not understanding what I'm saying."

No, the problem is that you keep wording it more strongly that you need to.  Everyone you've discussed this with on this site has agreed it is much too early and the data too noisy to form any concrete conclusions.  Are you under the impression that anyone disagrees with this?

69 " Everyone you've discussed…

" Everyone you've discussed this with on this site has agreed it is much too early and the data too noisy to form any concrete conclusions."

The point that I'm trying to make is that I'm pretty sure that we will never be able to make concrete conclusions about the Patriots defense this year. There simply won't be enough real data.

78 Say the patriots go…

Say the patriots go undefeated and lose a close one in the SB - allowing sub 7 points per game through the season.  Does that make them historically great?  Or do we still not know? 

The 2007 NE offense is considered historically great, scoring like crazy men, but played against some pretty crappy defenses as well. 

When did we know they were "historically great"?

87 The "rest of the NFL" isn't …

The "rest of the NFL" isn't "historically bad".  This is an unsustainable stance picked up solely out of convenience.

I'd wager your average football team today would have little problem demolishing any team from the 1970s or 1980s.  Today's linebackers are the size of linemen from those days, and everybody is faster.  Offenses are far more sophisticated, as are defenses.  

The 1972 Dolphins that went 17-0 had a line populated by five guys who were about 250 lbs. each.  There isn't a defensive line in the NFL that they could block today. 

90 Do we really need to…

Do we really need to explicitly state that all discussions of "historically great" in competitive sports are adjusting for the general improvement of human beings in subsequent decades? Doesn't everyone grasp that "historically great" is generally synonymous with "historically dominant, relative to the competition of the era"?

98 (sigh)

Ah yes, Sports Fans are allowed to bastardize the English language at will.  I forgot.

Regardless, there is a lot of serious obtuseness going on in this thread simply from a statistical standpoint.  Pat would have us through away the entire foundation of DVOA because he thinks their schedule has been too soft.  He says this in spite of the fact that DVOA is designed to account for strength of schedule.  

101 Oh, good grief, it does not …

In reply to by RickD

Oh, good grief, it does not "bastardize language" to use "historically great" without explicitly mentioning that, yes, athletes of 2019 are considerably bigger and faster than they were in 1989. Writers do not need to assume that readers are morons.

I'm not really all that interested in debating the relative greatness of this Patriots defense, but I think you are being a little unfair. It isn't obtuse to argue that a statistical model that tries to account for a variable may do so far less successfully in circumstances where the variable is an extreme outlier.

107 "He says this in spite of…

In reply to by RickD

"He says this in spite of the fact that DVOA is designed to account for strength of schedule.  "

No, I'm saying it because the chance probability of the best team in the past 30 years (which you'd assume should be order(5%)) combined with the chance probability of facing the worst set of offenses in the first 5 weeks (again, order(5%)) should only happen every several hundred years or so. Whereas the chance of the schedule correction being off is likely significantly higher than that, given that we have no evidence that it works in a situation like this.

Plus I *know* how DVOA accounts for strength of schedule. It does it via a second-order linear correction. Aaron's explained it before. He introduced the second-order correction (adjusting the Patriots defense based on the *adjusted* Jets offense, for instance, as opposed to the raw offense) back in 2005 due to the fact that the bottom of the NFL was *so bad*: Houston, and San Francisco had DVOAs of -33.1% and -67.4%, and we'd never seen anything like that before. Gee, where have I seen this before?

118 You went a lot further than…

You went a lot further than that.

You said:

The point that I'm trying to make is that I'm pretty sure that we will never be able to make concrete conclusions about the Patriots defense this year. There simply won't be enough real data.

You said this with less than 1/3 of the season done.  

I don't know what you expect out of DVOA, but the number of games played by the Patriots this season will be the exact same number as the season lengths for every year in the FO database thus far.  You're not just rejecting the data we've already seen, you're rejecting hypothetical data.

 

120 "but the number of games…

"but the number of games played by the Patriots this season will be the exact same number as the season lengths for every year in the FO database thus far."

Except that's not how DVOA works. It's a play by play measure. And if the plays that DVOA sees have more intrinsic fluctuation for the Patriots defense than other teams, you'd expect that the final measure would have more error, too. And that's the entire point - because the Patriots remaining schedule is *still* dominated by bad offenses, even though the current data (the plays from the first 5 games) is poor, the *remaining* data doesn't get dramatically better.

It's also a completely valid criticism that because we have *no data* on disparities this large (a ~ -50% defense playing  an average ~-25% offense) that any correction we make will be an extrapolation, with unknown systematics. That's just basic math.

Also note that when I said "pretty sure" I should've said "worried that" or "have a strong feeling that." I also don't think you're entirely understanding my worries. Suppose the Patriots finish the year allowing 110 points. I'm pretty sure that's *far and away* the best point-per-game performance *ever*. So you'd *really want* to be able to look at this year's Patriots squad and figure out what they were doing defensively, and really understand it. That type of performance *should* be something that you point to and say "this marked the beginning of a defensive shift back," because the rest of the league learns something.

Except people could point to those 110 points, and put a big asterisk by it, saying "it's great, but it's not the best ever, because no one *ever* played against that many bad offenses in a year, either." And that's a *completely valid argument*. That's what I'm worried about. I don't worry about people claiming New England's epic when they're *not actually* epic, because I don't care. I worry about New England *actually* being epic and not being able to understand it.

82 1) Offense is not defense…

1) Offense is not defense. There's *far* less noise in a good offense vs. a bad defense than the reverse. Receivers can run free and open all they want, but completing deep passes is still difficult.

2) They played against *some* bad defenses. Not *only* bad defenses. In week 2 they played San Diego, who ended up the sixth-ranked defense. By week 8 they had already played against 3 top 10 defenses.

This isn't really a stretch - these statements are easy to show. The Patriots have faced *vastly* worse offenses than any of the other teams in the "top defensive DVOA by week 5" list, as you can see below in another post.

86 Why do you think the nature…

Why do you think the nature of data collection has changed so much in 2019, compared to seasons past, that it would be impossible to measure the strength of its offense?  

This is a fairly large claim that you are making not only about this year's data (most of which has yet to be produced), but also the data collected over the past ~30 years.

The theory behind DVOA is that it accounts for the strength of the opposition.  Basically you're tossing that claim aside with no justification.  It's one thing to say that it's probablematic after only five games, of which three were against very weak competition (let's not forget that the Steelers didn't start the season plagued with injuries).  You're saying that, no matter how the next 12 weeks play out, it's not going to be enough for DVOA to be trustworthy.  That's an exceptionally strong claim.

96 "Basically you're tossing…

"Basically you're tossing that claim aside with no justification."

I've justified it a lot of places. Bad offenses tend to have higher variance than good offenses. You can try to correct the *mean*, but there's nothing you can do to fix the *noise*.

99 this isn't enough

It's just not close to enough for the kinds of claims you are making.  If you want to say "I think it's likely that the Patriots' D is overrated right now after playing some weak offenses," that's reasonable.  But you've dismissed the possibility that they'll be accurately rated at the end of the season.  

103 I don't know what algorithm…

I don't know what algorithm Dvoa is using for its opponent adjustments, but assuming its statistical in nature, I suspect its already incorporating team level variation and not one big population sigma, so it is capturing the fact that Jets offense is wildly variant compared to some other offense(the chiefs?). 

Pat may be reaching even further by saying not only are team sigmas significantly different from overall population sigmas, but they are conditional on the opponent they are facing. ie - the jets offense has some natural variability but somehow explodes when they face a team like New England. 

Obviously all of that is plausible, but I don't think there will be enough evidence to ever prove that one way or the other since the sample sizes are small and there's so much more going on to add to the time varying volatility - like injuries. 

In any case, at some point, you have to just accept that these are quantitative parables and live with them. I think we can live in a world where NE isn't the greatest defense of all time but something like the 10 or 20th best. They are certainly not the first defense to benefit from a horrible schedule. 

106 "I don't know what algorithm…

"I don't know what algorithm Dvoa is using for its opponent adjustments, but assuming its statistical in nature"

It's not Bayesian, no, it's just basic (second-order) correction. A Bayesian type correction would be better, but you'd obviously have to assume a basic prior on the likelihood of a team being a great defense, and obviously that'd reduce it overall. In some sense DAVE is doing something similar to that, and hey lookie, I have no problem with believing that NE is a -25.3% defense.

"Pat may be reaching even further by saying not only are team sigmas significantly different from overall population sigmas, but they are conditional on the opponent they are facing. "

I never said that...?

"I think we can live in a world where NE isn't the greatest defense of all time but something like the 10 or 20th best."

Dude, the difference between "best of all time" and 10th-20th is *huge*. Especially if you restrict yourself to, say, 1980+, since the game's basically completely different prior to that. I have *no problem* with saying they could easily end up in the 10-20 range since 1980. I mean, If they keep on this path I'd have *no* problem saying they're top 10. Right now I'd say 10-20 is probably safer, which is just another way of saying "they're basically in the top half of best defenses each year."

Look back at the original post I responded to: the statement was "I haven't watched them much this year, but I have a hard time believing this rating." As in, the rating that has them as *the best defense in the past ~30 years*. And that's *entirely* what I'm saying is extremely unlikely, and that's *all*.

110 I did use the word may and…

I did use the word may and it was mostly interpreting your statement about variation of bad offenses specific to New England.

Also one can do opponent adjustments without needing to go Bayesian. And as you pointed out, Bayesian models have their own shortcomings, especially when choosing a proper prior. 

116 The problem is that you are…

The problem is that you are saying that we "literally" have no evidence whatsoever regarding the Pat's defense, and that's obviously not true.  You're not saying that your assumption right now is that the Pat's defense is average are you?  Even games against very bad teams tell us something.

We could be looking at the average offensive DVOA of the '85 Bears, '00 Ravens or '02 Bucs.  Those would be pieces of information that would help us evaluate the data that we have on the Pat's.  We could look at how well defensive DVOA through 5 games tends to predict end of season DVOA.  We could look at the individual games that historically good defenses have played against very poor offenses.

You refuse to do any of that and just keep repeating "we have literally no evidence".  That's not an argument; it's just absurd.

122 "The problem is that you are…

"The problem is that you are saying that we "literally" have no evidence whatsoever regarding the Pat's defense, and that's obviously not true."

No, it really is! We have no analog for a great defense playing offenses this bad. None. It's never happened before. The offenses the Patriots have faced are over twice as bad, by average DVOA, than any of the other teams on that list. Which means if we want to be strict about it, the measure that we're assigning to New England's defense is a pure extrapolation. It's not bounded *at all* by data.

"We could look at the individual games that historically good defenses have played against very poor offenses."

It's not the *individual games* that matter, it's the overall combined effect, because it's the combined total rating. If you try to extrapolate from individual games, it's *still an extrapolation*. Of the teams in that list, the closest set of "bad offenses" that any team ever faced was San Francisco in 1997, with an average opponent offensive DVOA of -12% (compared to New England's -28% - note that I'm grabbing these numbers from memory from another poster here, so if I'm wrong it's my mistake, but scale-wise I'm about right).

If I try to shuffle San Francisco's 1997 schedule to get the *worst possible* set of offenses from the first 5 weeks, I *can't get* to New England's value. You'd have New Orleans, San Diego, the Rams twice, and then the Panthers. That would get you to -20%, which still isn't there.

Part of the difference here might be that we're using "evidence" differently. Of course we have *data*, but we have no *context* to put that data in, because a situation like this has never occurred before. We can try to make reasonable assumptions (extrapolations) to deal with it, but in the end we have to acknowledge (as Aaron has!) that this simply is a situation that's significantly different than anything that's happened before.

56 Balancing the Scales

I'm in agreement that the terrible offenses skew a clear look at the true ability of the NE defense. On the other side, this defense was very good through the playoffs last year. They stomped the Chargers, shut down KC for a half, and suffocated the Rams. They lost only one key player (Flowers) and have more then replaced him with what they've added. At a certain point, it's fair to say that this is a very, very good defense. Top 10 all time? Too early to know but ti wouldn't be a total shock if it were.

58 They also got shredded by…

They also got shredded by Kansas City for a half, to the point that, if  not for a play completely beyond their control (the 'lined up offsides' call that negated Brady's third pick), the game is over and they lose; Pats fans should stop using that game in their argument (you want to brag on the Super Bowl performance, more power to you, but anything earlier weakens your case)

64 They held KC to 7 points…

They held KC to 7 points through three quarters (albeit with Mahomes missing on an easy TD), so it definitely counts no matter how poorly they played in the 4th.  And given that one KC drive amassed 23 total yards and another was primarily composed of a big play that would get overturned on review for OPI this year, "shredded" isn't quite the right description.  They certainly lost control of the game, but that's more a credit to KC than anything else.

71 The Patriots defensive DVOA…

The Patriots defensive DVOA in the Chargers game was -15% and -10% for the KC game. I have no problem with someone hanging their hat on the Super Bowl defensive performance, which was -56%, although the opponent adjustment there is wrong - I don't think the Rams are anything more than a league-average offense since midway through last year - and there's plenty of evidence to support me on that - which would put the Super Bowl performance at more like -35%. Averaging by game (which is wrong, but whatever), that's -20%, which absolutely is a great defense, but not "86 Bears"-level defense.

I'm really not contorting myself or twisting data here, either. It's actually more frustration and disappointment that at this point it really looks like there's a team that's going to set an NFL record for points allowed and there will be *so many* caveats to their quality because of scheduling quirks.

73 It's amazing the length you…

It's amazing the length you'll go to discount data when it involves the patriots and runs counter to your narrative.  

 

I'm gonna guess that in a couple weeks were gonna find out that the Ravens, Eagles and Cowboys all have some sort of temporary issue that caused DVOA to misjudge a poor team as a good one. 

 

 

 

Yes, the Patriots schedule has been weak, but the idea that we don't know anything about their defense is absurd. Are they the best defense in history? No - probably not - but literally nobody is fucking arguing that. 

 

Are they significantly better than average? Absofuckinglootly, because average-to-good defenses don't consistently do to bad teams what the Patriots have done the last 7 or 8 games. 

75 "I'm gonna guess that in a…

"I'm gonna guess that in a couple weeks were gonna find out that the Ravens, Eagles and Cowboys all have some sort of temporary issue that caused DVOA to misjudge a poor team as a good one."

I sure as hell hope not! Are you going to argue that the Giants missing skill players responsible for 65% of their total yardage due to injury is *not* an issue?

"Are they the best defense in history? No - probably not - but literally nobody is fucking arguing that."

Uh, yes they are? That's what people are talking about. You don't go around saying "are the Patriots historically great" without insinuating that they might be the best defense in history. That's *literally* what the phrase "historically great" means.

I have *never once* said that I believe the Patriots are anything other than a great defense. I've flat out said, multiple times, that I think that in the end they basically look like a -20% DVOA defense. Which, for the record, would basically put them at the 6th best defense in the past 10 years.

76 Measurement Techniques

Pat actually raises an interesting point (though I think he's been a lousy advocate on this thread). The question is whether there is indictive value for defensive quality in plays vs. very low quality offenses; his position is that there definitely isn't. For Offenses with DVOA's below -20% it's a point worth taking seriously as there isn't a great deal of history of teams playing many games vs. teams with such low offensive DVOA's - though I suspect there probably is enough data in DVOA history to look at the result. The key to his statement is that he believes the season won't generate enough indicative data to make a judgement, which is much different then the "a few more games will improve the picture from DVOA" perspective.

I think referencing last years playoffs isn't a great argument here - last years playoffs combined with the data from this year clearly indicate that the Pats defense is good, but it doesn't really address the crux of the argument. The best way to address that is that repeated study of DVOA history clearly indicates that success against bad offenses is predicative of future success against good offenses - therefore you shouldn't really look at the Pats defensive DVOA through 5 weeks much differently than any other 5-week DVOA. The question worth studying is whether there is a lower limit to the DVOA of opponent offense beyond which opponent adjustments cannot properly operate. One poster used a few prior cases from great defensive teams to try to make that argument but the evidence presented here is extremely limited. To get enough data you would really have to look at the whole of DVOA history to factor-analyze final defensive DVOA vs. defensive DVOA against teams with <-20%, offensive DVOA, <-30% offensive DVOA etc... and look at the correlation to determine whether and at what point the data should be dismissed. My inclination is that it's like garbage time in that it's more predictive than you would ever imagine just by looking, but no one here has presented data to that effect.

80 Shit's hard, ain't it?…

Shit's hard, ain't it?

Purely from a entertainment enjoyment perspective, I'd like to see this Patriots defense play a road game against a great offensive line and even a just good enough qb, meaning one who wasn't easy to bamboozle. My favorite offense that I ever watched was the '91 Gibbs edition. Rypien was just a pretty decent qb, but when was allowed to feel comfortable he didn't make mistakes, and that offensive line consistently got everybody, and I mean everbody, blocked. At home, without crowd effects to hamper communication, I think they were much more than a match for any defense ever constructed. Even had 22 points, 285 net yards, 35 minutes time of possession in Philadelphia, in a week 17 game which was meaningless to Gibbs' team, but with the Eagles, perhaps the best defense ever, still trying to get into the playoffs. Don't think I'll ever see an offense that fundamentally sound again.

Unfortunately, it is very, very unlikely that we will see any combination of even a good offensive line, and smart, veteran, qb matched up against the Patriots this year, away from Foxboro. It only has a chance of happening in the last game, and who are the candidates? So few teams make o-line construction a priority anymore, and those that do tend to have young qbs. Saints, I guess, but their line is just above average, I think.

 

91 Oh, I know the Eagles had an…

Oh, I know the Eagles had an awful offense that year. However, to go into the stadium of perhaps the best defense ever, in a game that has no meaning to you, while that defense is still trying to get into the playoffs, and perform that way, is really impressive.

81 " The best way to address…

" The best way to address that is that repeated study of DVOA history clearly indicates that success against bad offenses is predicative of future success against good offenses"

It's predictive with lower correlation, because bad offenses have high variance (see footnote). So when you play only against bad offenses, the outcomes have more noise, and less signal, and your resulting measure has more noise. Combine that with the strong prior against being historically great, and it pushes the "likely" value of the measure way up.

(At least, they do *normally*. Because normally, when teams are bad, they try to not *be* bad, and so they bounce around a lot. Obviously when teams are good, they don't, and so intrinsically, good offenses tend to have less variance. If you want to understand why the Dolphins are so effed up this year, it's because their DVOA is -90% and their variance is currently 0.5%.  That, to me, is plenty of evidence that they're not actually trying to be an NFL team this year.)

93 Just looking at the…

Just looking at the historical DVOA records, it's not hard to find teams that are both bad and have low variance in their performance.  For example, the team with the second lowest variance after five weeks this year is Arizona, who is clearly making some effort to win games this year.

Miami is definitely a historical anomaly, but it's not clear that focusing on the variance in their play adds anything to the conversation.

108 To be clear the only reason…

To be clear the only reason I mentioned that is because the scuttlebutt is that the Dolphins aren't actually trying to win this year. The Dolphins have the worst DVOA ever at week 5. From what I can tell, they also have the lowest variance ever for the worst team at week 5. Obviously that could be chance, but given the scuttlebutt, it's reasonable to conclude those two things are related. For reference, from historical DVOA ratings (not from the actual current values themselves) in week 5 the worst team has a variance of 15% +/- 11%, so again, it's not *strong* evidence, but given the rumor mill, it's a reasonable conclusion.

10 Research elsewhere has shown…

Research elsewhere has shown that the vast majority of defensive variance (by EPA) is correlated with the strength of the opposing offense it faces, while offensive variance is *not* particularly correlated with the defenses it faces. 

So yeah, there's a decent chance that the NE and SF numbers are heavily inflated by having faced a series of genuinely putrid offenses. 

62 Of the Top5 All-Time listed…

Of the Top5 All-Time listed defense here the Pats have the easiest SOS for Offense faced (-26.80%) in the first five weeks. Their remaining offensive schedule is -5.50%, as of today.

Closest one was -12.36% for SF97, which had the remaining at -4.27%. They ended with -21.20% defensive DVOA, starting from -37.20%. 

Actually, only PHI91 improved (to -42.40%). PHI had -0.02% offensive SOS the first 5 weeks, then +4.22%. 

CHI86 went from -38.60% to a final -33.60%. CHI had +4.98% offensive SOS the first 5 weeks, then -4.38%.

TB02 went from -38.50% to -31.80%. TB had -4.78% offensive SOS the first 5 weeks, -7.34% the remaining ones.

3 I wonder if there's any kind…

I wonder if there's any kind of way to incorporate Vegas odds into early-season DVOA. I mean, there is no way the 49ers are one of the top 5 teams since 1986. Their performance so far might suggest that, but we all know (or strongly suspect) that's not true. For instance, they are currently 3.5-point underdogs to the Rams in LA. By DVOA, based on the 56.8% gap, they should be double-digit favorites. I'm guessing the truth is somewhere in between, i.e. that Niners should be slightly favored.

I know this isn't a super helpful comment, but I'm trying to think of alternative ways to discount early season DVOA (besides DAVE, which is excellent as it is).

50 I'm pretty sure the Vegas…

I'm pretty sure the Vegas lines already incorporate DVOA and/or a similar formula.  I do find it odd, though, that the Rams are favored by 3.5.  If you look at DAVE, which seems to be much more predictive than raw early-season DVOA, you get about an 18% difference, in favor of the Niners.  But the Rams are at home, and I seem to recall a comment Aaron made many, many years ago, to the effect that HFA is worth about 13% DVOA.  So, that would drop the difference to about 5%, which means the Niners should be favored by a small amount (about 1 or 2 points).  That's a difference of about 5 points between the "DVOA line" and the Vegas line.  I wouldn't expect them to be identical, but I would expect them to be closer than that.

59 Well, the line is closer to…

Well, the line is closer to Sagarin's prediction of SF by 0.5 points (odd that HFA is only 1.39 there). Of course, LA's rating is inflated by getting an easy win against the Saints when Brees went down early, and I have to think the other reason that LA is favored so much is simply inertia from last year.

113 Well, subjectively, Sagarin…

Well, subjectively, Sagarin's line seems best to me.  It seems like basically a "pick'em" game, with the 49ers being perhaps a very slight favorite.  But I think you're right about the public perception of the Rams.  Many people probably haven't really noticed that, compared to last year, Gurley isn't playing as much, and Goff isn't playing as well (and, for what it's worth, I don't think those 2 things are a coincidence).

60 Vegas odds are only designed…

Vegas odds are only designed to get the same amount of money on either side of the line. They are not a representation of how the bookies think that each team will perform, they are a representation of how the bookies think that the betting population thinks how each teams will perform.

66 The 49ers hype train was…

The 49ers hype train was last season.  This year it was Browns-Browns-Browns.  Turns out the hype was right, it was just a year early for SF.  After the Monday night game, I would have expected more people to jump back on the SF band wagon, but it's a west coast team that got almost no press in the preseason.  That's my best guess at trying to explain why SF isn't giving up a FG rather than than getting one in the betting line.

125 That's true, but in reality,…

That's true, but in reality, the final line ends up correlating very strongly to win probability. Maybe it's a "wisdom of the crowd" thing.

If you disagree, feel free to quit your day job and make millions exploiting mispriced Vegas spreads!

126 Yes, although I don't think…

Yes, although I don't think it's necessarily a "wisdom of crowds" sort of thing.  It's more a case of there being a very small number of "sharp" bettors, who make very large bets whenever they see an exploitable line.  The oddsmakers then adjust the line to encourage action on the other side. This continues until the line has been "corrected" (that is, until it reaches an equilibrium).  Of course, the oddsmakers try to anticipate all of this, and they often succeed, which is why many lines experience little to no movement.

5 WSH-MIA next week

Is the -123.2% combined DVOA in Washington/Miami the lowest ever for a game in the DVOA era?

12 Maybe after week 5, but…

In reply to by Mountainhawk

Maybe after week 5, but lowest ever, not even close. Remember that teams can have DVOA ratings around -100% after week 1, and sometimes two teams with horrible game 1's face each other in week 2. So 2018 saw -56.9% SF and -123.4% Detroit meet up (total of -190.3%) and 2017 saw -61.2% Arizona and -119.2% Indianapolis meet up (total of -180.4%).

42 At least until they roll…

At least until they roll Haskins out at starting QB this week, assuming Gruden was correct that Haskins shouldn't be playing, yet.

Miami's had a bye week, so it's possible the extra time will help the team catch up to what Flores wants them to do on defense.  Combine that with a possible first start by a rookie, and you have the potential for Miami's D to have a good game. 

On O, Rosen has presumably been taking the snaps with the starters, has also had a couple of more weeks to get familiar with the offense, and maybe replaced all of Miami's receivers with guys he knew from high school so maybe possibly some of them know how to catch?  Not sure that helps if he keeps getting sacked, though.

This one may come down to whose defence scores more points, or at least sets their offense up inside of easy field goal range.

88 Haskins won't start

The word out of  DC is that, even though Gruden was fired for "not using Haskins properly" (at least if you believe the media), Bill Callahan won't  actually diverge from that strategy.

Haskins might not even be active if both Keenum and McCoy are healthy.  

Honestly, I don't think anybody in the DC area thinks Haskins is "ready".  There seems to be a lot of bickering about what it would take to get him to that point.  The big change from last week to this week is that he's getting more "1st team reps".  I don't quite see why that's such a big deal if he's not going to play anyway.

 

6 By my count, NFC is now 16-8…

By my count, NFC is now 16-8 against the AFC. I’d be surprised if we don’t see one or even two 8-8 playoff teams like we did in the NFC back in the mid 00s.
NFC teams are likely favoured in 4 of the 5 inter-conference games this week with, guess who, the Giants v Patriots being the exception.

13 If the playoffs started…

If the playoffs started right now all the qualifying teams in the AFC would be above .500. You are going to have to have a whole lot of teams have mediocre records the rest of the way for there to be even one 8-8 make the playoffs. I don't think it's particularly likely at all.

14 Well yeah, because almost…

Well yeah, because almost every team has played an odd number of games. You only have to look at the mean wins of the teams on the playoff odds page to see that there's a decent chance. There are nine NFC teams with at least 8.5 expected wins, versus only five for the AFC.

18 I'm not sure that does mean…

I'm not sure that does mean there's a decent chance. Those are just means. It's the same thing with preseason projections, which regularly predict pretty much every team clustered in the 6 to 10 win range. That doesn't mean you should actually expect that to happen, because some teams are going to exceed expectations and some are going to fall well short. Two 8-8 teams in the playoffs would require basically half the conference to play at near .500 the rest of the season, which doesn't seem likely.

26 Two 8-8 teams in the…

Two 8-8 teams in the playoffs doesn't seem likely, no. But one 8-8 team isn't that unlikely, and doesn't require half the conference to play at near .500 the rest of the season either, if it happens to be the record of the winner of the AFC South or North.

27 8-8 Playoff Team

I think theres a different issue at play here; The AFC is so bottom-heavy right now that I find it hard to believe that there won't emerge a second Buffallo-like team that is pretty good, and ends up with 10 wins because, well, there are a lot of bad teams in the AFC on which to build your record. I think the bad teams in the AFC will have to inch toward mediocre over the next year to see an 8-8 playoff team next year.

It was interesting to watch SF on Monday night - the DVOA on them is no accident (though it's early) and I think they ought to be clear favorites to win the NFC (head and shoulders above the rest because they are less exposed to injury risk with how their offense operates). That perspective could be distorted by the (possible) misapprehension that CLE is a decent team/defense - we shall see.

61 Good point about the AFC…

In reply to by sbond101

Good point about the AFC being bottom-heavy, though it might be countered with the fact that the AFC East and the AFC North are playing each other. Cleveland and Pittsburgh are in the best position to take advantage of their schedule, and I don't think either team is up to the task.

Then you have the AFC South and West playing each other, with no really bad team (currently) for the others to leach wins from. I think the best bet for an 8-8 playoff team is from the AFC South, either as a wild card or a division winner.

28 weird things going on

Look at which AFC teams have which played NFC teams.  

AFC East is 2-2 vs NFC East, with each AFC East playing one game.

AFC North is 1-7 vs. NFC West.  Yikes!  Bengals already have three losses there, while Ravens are 1-0.

AFC South is 3-3 vs. NFC South.  

AFC West is 2-4 vs. NFC North.

Mostly what's happened is the AFC North has face-planted vs. the NFC West.  But they've already played half their interconference games, and the Bengals, in particular, can only contribute one more loss.  On the flip side, Pats and Chiefs have each only played one interconference game. 

94 After hosting Saints this…

After hosting Saints this week, Jags schedule gets really soft (and I'm not sold on Bridgewater on the road).  Between them, the Bills, the Raiders, and the Chargers, we'll get two wild cards with nine wins (you could still sell me on 8-8 winning the AFC North)

7 One of the issues with…

One of the issues with current DVOA is that comparison across years becomes muddled by in season averages. It works for teams like Miami or the 07 Patriots, but it starts to become spurious when you have multiple severely awful teams dragging down the mean. VOA might make some sense, but then we lose the opponent adjustments. Maybe some kind of resampling methods with outlier drop out??

Going off of subjective views - as of now - this season has the largest collection of toilet teams I've seen in a very long time(I am willing to hear alternatives). No one even remembers how terrible the Bengals are because the Jets and Skins have already shoehorned themselves as foils to the Dolphins. And amazingly, the Patriots will be playing every single one of these dumpster fires.

74 And you're not even…

And you're not even mentioning the ridiculously-hyped Browns, whose offense could easily free-fall into godawful-level. Mayfield actually strikes me as the kind of player who could easily start self-destructing badly to try to prove people wrong if he sees his reputation start to change. And the Patriots play them too!

That's *seven games* against teams with offenses at -20% or worse. *And* they manage to play the Giants on the week when the guys responsible for *65% of their yardage on the season* are all going to miss the game. Assuming that pushes the Giants to at *least* a -20% offense level, that's literally *half* of their season against teams that would probably rank last in the league in offense on any given year.

It's seriously a tragedy for real football fans.