Bills Move to No. 1 in DVOA with Dominating Defense

Mario Addison
Mario Addison
Photo: USA Today Sports Images

NFL Week 4 - We have a new No. 1 team in DVOA this week, the Buffalo Bills, and they are No. 1 in DVOA by a hefty margin after putting up one of the biggest games in DVOA history with their 40-0 shellacking of Davis Mills and the Houston Texans. There's a gap of over 16 percentage points between Buffalo and our No. 2 team right now, the Dallas Cowboys. The undefeated Arizona Cardinals are right behind in third place, followed by Cleveland, the Los Angeles Rams, and Baltimore.

But the Bills are the big story, and we've got a lot of fun data to share about the incredible defensive performance they put together on Sunday. Right now, with the current opponent adjustments, the Bills come out with -134.4% defensive DVOA for Sunday's win. That would be the best defensive performance in a single game ever measured by DVOA, going all the way back to 1983.

OK, you are probably thinking, but that's only with early-season opponent adjustments. Regular readers know that early in the season, we don't do our opponent adjustments at full strength. Right now, they are only at 40% strength because we don't truly know yet how good teams are. Certainly once the opponent adjustments are stronger, Buffalo's rating for this game will be less impressive, right?

Don't be so sure. The opponent adjustment for playing against Houston may not be as strong as you think. Some of that is due to the specifics of how our system works. The opponent adjustment will be based on how Houston's offense plays for the entire season, and that will include a few games of Tyrod Taylor at quarterback. But Davis Mills also may not be as bad as he looked on Sunday. Houston's pass offense DVOA was -2.0% against Carolina the week before, far from the worst in the league.

I was curious what would happen to Buffalo's defensive rating for this game with opponent adjustments at full strength, so I ran a set of DVOA ratings with those adjustments. And it turns out Buffalo's defensive performance in this game falls from the best ever to ... still the best ever, just by a smaller amount! Buffalo would be at -127.2% DVOA with full opponent adjustments. Here's a look at the best single defensive games in DVOA history. I'm listing Buffalo's rating with the full opponent adjustment, but of course that rating is subject to change over the course of the entire season as we gather more data on both Houston and Houston's other defensive opponents.

Top Single Games by Defensive DVOA, 1983-2021
Team Year Week Opp Score Opp
Off Rk
BUF 2021 4 HOU 40-0 30 -127.2%
CLE1 1989 1 PIT 51-0 24 -126.2%
BAL 2000 SB NYG 34-7 8 -120.9%
PIT 1999 1 CLE 43-0 26 -120.9%
PHI 1991 3 DAL 24-0 4 -116.0%
TEN 2000 17 DAL 31-0 24 -114.6%
NYJ 2009 17 CIN 37-0 19 -113.7%
BUF 2004 14 CLE 37-7 27 -112.0%
CHI 1985 SB NE 46-10 11 -106.1%
DEN 1999 6 GB 31-10 11 -103.9%
NYJ 2010 17 BUF 38-7 26 -102.7%
NE 2004 10 BUF 29-6 21 -102.5%

Buffalo's defensive performance might be even more impressive if we only look at pass defense DVOA, although it is no longer the best game ever measured. Buffalo's pass defense DVOA is -208.4% with the current opponent adjustments and would be -204.7% with full opponent adjustments. Mills had just 87 yards on 21 passes that included four interceptions. He had a fumble on a sack which Houston recovered, plus Jordan Akins fumbled after a reception. Buffalo's pass defense DVOA for this game is actually better than what the New Orleans Saints put up last year when they played against a Denver team that had no real quarterback!

Top Single Games by Pass Defense DVOA, 1983-2021
Team Year Week Opp Score Opp Pass
Off Rk
Pass Def
Main QB Cmp Att Yds TD INT Sk
NE 2009 6 TEN 59-0 17 -234.5% K.Collins 2 14 -7 0 2 0
NYJ 2009 17 CIN 37-0 15 -207.5% C.Palmer/J.T.O'Sullivan 4 19 31 0 1 3
BUF 2021 4 HOU 40-0 28 -204.7% D.Mills 11 21 87 0 4 3
NO 2020 12 DEN 31-3 31 -200.9% K.Hinton 1 9 13 0 2 1
CAR 2010 5 CHI 23-6 (L) 26 -185.7% T.Collins 8 19 51 0 4 3
GB 1987 10 SEA 24-13 (L) 13 -183.3% D.Krieg 9 15 105 0 3 3
TEN 2000 17 DAL 31-0 25 -181.5% A.Wright 5 20 35 0 2 4
GB 2006 17 CHI 26-7 27 -176.6% B.Griese/R.Grossman 8 28 194 1 5 2
TB 2007 15 ATL 37-3 26 -176.4% C.Redman 4 15 34 0 2 1
MIA 1983 4 KC 14-6 12 -171.8% B.Kenney 10 22 105 0 4 3
PHI 1991 3 DAL 24-0 4 -167.8% T.Aikman 11 25 112 0 3 11
DAL 2003 15 WAS 27-0 19 -167.5% T.Hasselbeck 7 27 59 0 4 1

As an aside, we could also look at these stats from Houston's perspective and get similar lists, but today I wanted to concentrate on the Bills defense. It will be fun to see where this game stands from Houston's perspective when we get to the end of the year and we know how well Buffalo's defense has played against other opponents.

Also, how crazy is it that two of the teams with the best pass defense ever recorded in a game lost those games?

Here's another list. Buffalo's defensive performance was so strong against Houston that the Bills also break into the list of the ten best overall games in DVOA history. Once again, the ratings I'm listing here are the ratings with full opponent adjustments applied, subject to change over the rest of the season.

Top Single Games by Total DVOA, 1983-2021
Team Year Week Opp Score Opp Rk Off Def ST Total
SF 1993 DIV NYG 44-3 7 67.3% -78.2% 10.9% 156.4%
NO 2020 9 TB 38-3 2 50.2% -96.7% 6.7% 153.5%
CLE1 1989 1 PIT 51-0 20 15.7% -126.2% 9.2% 151.1%
NYG 1986 DIV SF 49-3 2 66.5% -73.0% 10.0% 149.4%
WAS 1991 1 DET 45-0 15 58.6% -67.7% 20.7% 147.1%
PIT 1999 1 CLE 43-0 31 20.6% -120.9% 5.5% 147.0%
MIN 1987 WC NO 44-10 2 15.7% -102.0% 27.1% 144.9%
BUF 2021 4 HOU 40-0 29 9.2% -127.2% 5.7% 142.1%
WAS 1983 DIV LARM 51-7 6 55.5% -59.7% 25.4% 140.7%
DET 1983 13 PIT 45-3 14 31.9% -77.0% 30.4% 139.4%
CAR 2005 WC NYG 23-0 10 34.8% -92.3% 11.7% 138.8%
PHI 1994 5 SF 40-8 3 60.8% -73.9% 3.8% 138.5%
NE 2008 16 ARI 47-7 21 40.9% -74.6% 22.6% 138.0%
ATL 2002 12 CAR 41-0 25 47.9% -91.5% -1.8% 137.6%
CAR 2015 CCG ARI 49-15 2 57.1% -65.3% 12.8% 135.1%
BAL 2019 12 LAR 45-6 12 67.4% -58.7% 8.8% 134.9%
SEA 2012 14 ARI 58-0 24 56.2% -63.0% 15.5% 134.7%
NE 2004 10 BUF 29-6 3 40.0% -102.5% -8.5% 134.0%
KC 2002 13 ARI 49-0 31 49.1% -71.6% 12.5% 133.2%
NE 2009 6 TEN 59-0 22 58.0% -80.6% -5.4% 133.2%

When you have a game this strong and it counts as one of just four games in the current sample, it's no surprise to see the team involved moving into the No. 1 spot in DVOA overall. The surprise might be the way that Buffalo's defense is so outplaying its offense so far this season. The Bills offense ranks just 16th in DVOA through four weeks, with positive ratings only in Weeks 3 and 4. Buffalo is second in the league in scoring because the defense keeps putting the offense in great field position. Buffalo starts its average offensive drive at the 35.9-yard line, the best figure in the league by a signficant margin. The Bills are also second in the league with 46 offensive drives because the defense keeps getting them the ball back.

Although Week 4 was by far their best game, the Buffalo defense has been better than average in all four games so far this season, and so they also have one of the best defenses ever tracked through four games of the season. 

Best Defensive DVOA
Through 4 Games, 1983-2021
Year Team W-L DVOA
1984 CHI 3-1 -52.4%
2019 NE 4-0 -50.3%
2021 BUF 3-1 -49.4%
1997 SF 3-1 -46.8%
2008 BAL 2-2 -45.9%
2019 SF 4-0 -45.4%
1991 PHI 3-1 -42.6%
1996 GB 3-1 -40.3%
1985 DAL 3-1 -36.7%
1989 CLE1 3-1 -36.6%
2011 BAL 3-1 -36.1%
1991 NO 4-0 -36.0%

What's most remarkable may not be how good the Buffalo defense has been through four games but how big the gap is between Buffalo and the rest of the league. Right now, the No. 2 defense in DVOA is New Orleans -- surprising given that Daniel Jones led the league in passing DYAR this week playing against the Saints. But the Saints were good enough in their first three games, especially against the Packers in Week 1, to still rank second in defensive DVOA. The Saints are at -18.4% defensive DVOA right now. This means the gap between the Bills and the Saints is more than 30 percentage points! As you can probably imagine, this is the largest gap between the No. 1 and No. 2 defense ever recorded after four weeks of a season.

Largest Gap Between Top 2 Defenses After Week 4, 1983-2021
Year No. 1 Def DVOA No. 2 Def DVOA Gap
2021 BUF -49.4% NO -18.4% 31.0%
1984 CHI -52.4% SEA -28.7% 23.7%
2003 TB -44.3% IND -21.4% 22.9%
1997 SF -46.8% STL -26.8% 20.0%
2008 BAL -49.2% TEN -32.8% 16.4%

Now, here's the downside for Buffalo: we know that defensive performance tends to be more variable than offensive performance, and the power of regression to the mean is stronger. If we want a reminder that great defensive performance doesn't always continue for the entire season, we only have to remember back to the New England Patriots of two years ago. It's possible that our opponent adjustments just aren't correcting enough for the value of playing against a series of young and/or backup quarterbacks. It would be better for the Bills if their offense was ranked first and their defense was 16th. However, based on our preseason projections and how good their offense was last year, there's a strong chance that the Buffalo offense will be better than this for the rest of the season. So even if the defense does come back to earth -- and it will -- the Bills are still set up to be a top Super Bowl contender.

This all sets up for an awesome Sunday night game this weekend, because our best defense is going to be playing our best offense, the Kansas City Chiefs. Although the Chiefs' offense is not quite as impressive as the Bills defense right now, it's still very impressive. The Chiefs are No. 1 on offense at 41.1%, with no other offense above 30%. Kansas City has the 14th-best offense we've ever tracked through four games. The problem is that their defense is almost as bad as the offense is good, easily ranking dead last so far this season.

How Good Is Arizona?

The Bills are also now No. 1 in our DAVE ratings which combine early-season DVOA with the preseason projections. However, even after four weeks, the DAVE ratings are still 70% based on those preseason projections, so the next two teams are the two teams we had on top of our preseason forecast, Kansas City and Tampa Bay. Arizona fans might be annoyed to see the Cardinals down at No. 10 in DAVE even though they are the final unbeaten team in the NFL this season. Our preseason projection for the Cardinals had them as a league-average team, or very close to it. How likely is it that we were totally wrong about them?

The ratio of DVOA to projection used in DAVE is based on looking at what ratio best predicts DVOA over the rest of the season, but it's certainly possible that we've painted with too broad a brush. Perhaps things are different if we specifically look at teams that had average DVOA projections and then way outplayed those projections in the first four weeks? I went and looked at all teams with a projection between -5.0% and 5.0%, going back to 2010. Here's a look at the teams that most outplayed their projections in Weeks 1-4, and then what the final DVOA ratings were for those teams.

Biggest Outperformance of an Average
Preseason DVOA Projection After Week 4, 2010-2021
Year Team Week 4
Rk Projection Gap Final
2019 SF 57.7% 1 -2.0% 59.7% 29.0% 5
2017 KC 43.4% 1 3.8% 39.6% 10.9% 10
2021 ARI 33.2% 3 -1.0% 34.3% -- --
2012 SF 33.9% 2 4.2% 29.7% 30.2% 4
2016 MIN 27.4% 3 -0.6% 28.0% 1.2% 18
2010 NYJ 30.7% 3 3.0% 27.8% 17.7% 7
2021 CLE 29.3% 4 2.5% 26.8% -- --
2018 KC 24.1% 3 -2.4% 26.5% 33.2% 1
2020 GB 25.2% 5 -0.3% 25.5% 25.8% 3
2016 ATL 20.5% 5 -4.8% 25.3% 23.9% 3
2013 IND 22.2% 5 -2.5% 24.7% 2.7% 14
2016 DEN 24.7% 4 0.9% 23.8% 2.9% 15
2015 ATL 22.1% 5 -0.1% 22.2% -12.3% 22
Only includes teams with projection from -5.0% to 5.0%.

What can we learn from this table? I went to 13 teams and included the 2015 Falcons to point out that sometimes hot early teams really crash out. But overall... I guess it looks like these early strong performers are more likely to keep at a high level all year than DAVE might otherwise indicate. Five of them were very strong for the rest of the year, six if you want to count the 2010 Jets. These teams had an average DAVE of 9.0% after four weeks, but their final actual DVOA averaged 15.0%. So for these teams, the early performance meant a little bit more than it did for the entire sample of teams from 2010-2019 that I used last year to create the new DAVE ratios. Perhaps that's a good sign that the Cardinals can continue outperforming their mediocre preseason projection. 

* * * * *

Football Outsiders playoff odds, snap counts, and the FO+ database are now all updated through Week 4.

A reminder that all our free stats pages, including DVOA and player position stats, now require registration to view. This is not a paywall! You only need to register (for free) and then log in to the site to view these pages. While you're at it, you can get a seven-day trial of FO+ and check out the FO+ features like a deeper DVOA database, weekly fantasy projections, fantasy football research tools, and picks against the spread.

* * * * *

Here is the Football Outsiders Top 16 through four weeks of 2021, 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 for opponent and performance indoors and consider all fumbles, kept or lost, as equal value. Because it is early in the season, opponent adjustments are currently at 40% strength. 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.

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 70% preseason forecast and 30% actual performance. It is not currently adjusted for any backup quarterbacks.

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>

1 BUF 49.8% 6 22.4% 1 3-1 0.0% 17 -49.4% 1 0.5% 12
2 DAL 33.5% 9 13.9% 5 3-1 27.8% 3 -6.3% 10 -0.6% 17
3 ARI 33.2% 4 9.3% 10 4-0 17.1% 6 -11.4% 6 4.8% 3
4 CLE 29.3% 2 10.5% 9 3-1 8.6% 9 -16.4% 3 4.2% 5
5 LAR 24.9% 3 13.0% 6 3-1 29.4% 2 0.2% 20 -4.3% 28
6 BAL 21.0% 13 16.5% 4 3-1 7.2% 10 -5.0% 12 8.7% 2
7 NO 17.5% 8 8.5% 11 2-2 -0.7% 18 -18.4% 2 -0.1% 14
8 KC 16.4% 15 19.4% 2 2-2 41.1% 1 27.3% 32 2.6% 6
9 TB 16.0% 7 19.3% 3 3-1 18.6% 5 -0.1% 19 -2.7% 27
10 CIN 14.8% 11 -0.6% 16 3-1 -1.5% 20 -11.6% 5 4.7% 4
11 SEA 13.4% 10 11.1% 8 2-2 24.0% 4 10.4% 25 -0.2% 15
12 CAR 10.0% 1 -4.7% 20 3-1 -0.8% 19 -15.4% 4 -4.6% 30
13 SF 8.6% 12 7.7% 12 2-2 10.2% 7 -0.7% 16 -2.4% 24
14 LAC 8.2% 22 -3.1% 19 3-1 6.7% 11 -6.5% 9 -5.0% 31
15 DEN 7.8% 5 4.5% 14 3-1 1.5% 15 -11.0% 7 -4.6% 29
16 MIN 4.1% 16 4.8% 13 1-3 2.2% 14 -0.9% 14 1.1% 9

Click here for the full table.


68 comments, Last at 09 Oct 2021, 4:37am

#1 by Will Allen // Oct 05, 2021 - 4:34pm

Vikings have had the hardest DVOA schedule so far, and their three losses have been in ot, by a missed 39 yard field goal at 0:00, and 14-7.

"So you're sayin' there is a chance!"

-Lloyd Christmas

Points: 0

#9 by theslothook // Oct 05, 2021 - 5:43pm

It's so hard to get a handle on this Vikings team, especially their offense.

Points: 0

#14 by DIVISION // Oct 05, 2021 - 6:22pm

When I was watching the Vikings/Hawks game, it seemed like two teams playing mirror images of themselves. 

Both teams are boom or bust.

The Vikings are so dependent on Cousins having a clean pocket that their offense can look amazing or horrible depending on the opponent.  It's not all that different from years past.  You saw what the Browns did.  

The Vikings will have a shot at the NFC North because the Packers don't have a great defensive front and that's the key to beating the Vikings.

Points: 0

#19 by Will Allen // Oct 05, 2021 - 7:16pm

When their offensive line doesn't get a can opened on them, they are a good team. That can being opened is not a low probability event when playing a good defensive front, to engage in a bit of understatement.

Points: 0

#29 by bedfordp // Oct 06, 2021 - 3:24am

CIN preseason win total of 6.5 poor team

ARI 8.5 average

SEA 10 playoff fringe

CLE 10.5 playoff fringe

So schedule forecasted to be a shade above average in terms of difficulty and may be proved to be a little harder than that. A shade above average brings us to the quarterback a shade above average highly unlikely to be anything else and lead you to the promised land

Two brilliant QBR ratings one average and one clunker - Cousins in a nutshell. I would suggest you expect nothing else

Points: 0

#34 by AFCNFCBowl // Oct 06, 2021 - 8:11am

ARI and CIN are obviously better than Vegas Odds expected. SEA is probably a little worse, and CLE is around where they thought. MIN's schedule is tough.

Points: 0

#41 by justanothersteve // Oct 06, 2021 - 10:56am

All four NFC North teams have killer schedules as they play the NFC West and AFC North outside of division. Don't know about Minn, but the Packers get KC for their Game 17. I believe Detroit started the year as the toughest schedule. 

Points: 0

#45 by AFCNFCBowl // Oct 06, 2021 - 12:15pm

The 17th game pairs the NFC North with the AFC West (tied with NFC West for best record in football), so yikes.

The only "easy" games left for MIN are the two DET games and PIT. They got the short end of the stick facing DAL in the NFC East and CAR in the NFC South is a tough out as well.

Points: 0

#2 by mehllageman56 // Oct 05, 2021 - 4:43pm

DAVE was really down on the 2010 Jets, more so than I thought it should have been.  They had gone to the AFC Championship game on the back of their #1 running game and #1 defense the year before, were starting a second year quarterback who should have improved (and did, slightly), and added a Pro Bowl cornerback (Antonio Cromartie) and Pro Bowl wide receiver (Santonio Holmes) while losing only Kerry Rhodes (Pro Bowl safety, so something).

The only team on that list to not beat its projection was the 2015 Falcons.  I have my doubts about the Cardinals, but I'm probably wrong to doubt them.  Not as surprised with the Browns though.

Points: 0

#6 by theslothook // Oct 05, 2021 - 5:37pm

To be fair that 2009 Jets team was pretty flawed and got to the playoffs with the help of opponents missing field goals, the Colts resting starters in the second half ( and dropping tds along the way), and the Bengals completely laying down for that week 17 game (if ever a week 17 game should have an asterisk next to it).

To their credit, the defense smothered the perpetually disappointing Marvin Lewis led Bengals, and Chargers ( once again benefitting from special teams woes).

And well I know I'm sounding pretty dismissive, but I actually have a soft spot in my heart for the Rex Ryan lead Jets who did about as well as they could given who their QB was.

I also hold Rex Ryan the head coach in quite a bit of esteem. I know he does himself 0 favors in press conferences and is flawed when it comes to decision making, but he's one of the few defensive coaches that raises the floor of his defenses and gets a lot of mileage out of defenses with severe weaknesses in pass rushing. 

Points: 0

#15 by DIVISION // Oct 05, 2021 - 6:26pm

The Brown have a solid running game and elite defense.  Their weak link is Baker Mayfield. 

The Cards are more balance on offense and defense than Cleveland, referenced by DVOA.  Also, Kyler is world's better than Baker even though they had the same offense in college.  Kyler has already beaten Baker in the pros.

Why would you doubt Zona's sustainability rather than Cleveland's?  


Points: 0

#23 by Kaepernicus // Oct 05, 2021 - 9:29pm

I think people are doubting them based on history. This Cardinals team looks like the best since the 2015 Carson Palmer team to me. I would say the defense is probably a touch overrated because of the huge turnover number driven primarily by fumbles lost by the other team. Also the offense is a legitimate top 5 offense with deceptively low turnovers due to fumble luck again. At worst some bad luck makes them a 3-1 team with a top 5 offense and top 15 defense. That is still a really good team. The 49ers are literally the opposite team where they have forced 5 fumbles and recovered none while fumbling 7 times and recovering only 2. The big reasons to assume the Cardinals are going to continue to be really good is the pass rush, 3rd down/redzone offense, and Kyler playing out of his mind. If I had to bet on the season long strength of the Cardinals vs. the Browns I would probably lean on the Cardinals because of the QB play. AZ has been very good and lucky at the same time (fumbles) which typically makes them a great fade play at this point. I would definitely bet the 49ers at -5.5 this week given the divisional aspect, public enthusiasm,  and the fact that AZ has a terrible rushing defense. Long term though AZ looks like the best or second best team in the toughest division in the NFL.

Points: 0

#46 by AFCNFCBowl // Oct 06, 2021 - 12:17pm

Not sure how much fumbles are factored into DVOA. TEN has had horrid fumble luck this season and yet is ranked 27th in DVOA even with a 2-2 record and being a clear division favorite. DAL has been the luckiest team in the league with fumbles and yet is ranked 2nd in DVOA.

Points: 0

#47 by Ranccor // Oct 06, 2021 - 1:13pm

Haven't looked at it in a while, but I'm pretty sure DVOA thinks of causing and dropping fumbles as skills, but who recovers them is pretty much luck.  

Points: 0

#58 by AFCNFCBowl // Oct 06, 2021 - 6:49pm

Yes, that's right. Even so, teams that recover lots of fumbles (no matter who fumbled or not) usually get penalized big by DVOA.

Points: 0

#60 by jheidelberg // Oct 06, 2021 - 7:08pm

Recovery of a fumble, despite being the product of hard work, is almost entirely random.

Stripping the ball is a skill. Holding onto the ball is a skill. Pouncing on the ball as it is bouncing all over the place is not a skill. There is no correlation whatsoever between the percentage of fumbles recovered by a team in one year and the percentage they recover in the next year. The odds of recovery are based solely on the type of play involved, not the teams or any of their players.

Fans like to insist that specific coaches can teach their teams to recover more fumbles by swarming to the ball. Chicago's Lovie Smith, in particular, is supposed to have this ability. However, in Smith’s first three seasons as head coach of the Bears, their rate of fumble recovery on defense went from a league-best 76 percent in 2004 to a league-worst 33 percent in 2005, then back to 67 percent in 2006.

Fumble recovery is equally erratic on offense. In 2008, the Bears fumbled 12 times on offense and recovered only three of them. In 2009, the Bears fumbled 18 times on offense, but recovered 13 of them.

Fumble recovery is a major reason why the general public overestimates or underestimates certain teams. Fumbles are huge, turning-point plays that dramatically impact wins and losses in the past, while fumble recovery percentage says absolutely nothing about a team's chances of winning games in the future. With this in mind, Football Outsiders stats treat all fumbles as equal, penalizing them based on the likelihood of each type of fumble (run, pass, sack, etc.) being recovered by the defense.

Other plays that qualify as "non-predictive events" include blocked kicks and touchdowns during turnover returns. These plays are not "lucky," per se, but they have no value whatsoever for predicting future performance.

  • Pro Football Prospectus 2005, New Orleans chapter

Points: 0

#62 by greybeard // Oct 06, 2021 - 7:25pm

Maybe the QB and RB fumble recoveries are all random. Is it true for all fumble types that it is random though? I feel like a WR fumbling the ball has a much higher recovery chance by defense than offense. It does not require skill, just statistically there are more defenders around fumbled ball than offensive players.


Points: 0

#63 by jheidelberg // Oct 06, 2021 - 7:50pm

Pulling a specific line from my prior post:

The odds of recovery are based solely on the type of play involved, not the teams or any of their players.

Points: 0

#66 by Eddo // Oct 07, 2021 - 12:39pm

Football Perspective did a study on recovery rate by type of fumble.  Their site appears to be down right now, but as I recall, aborted snaps were recovered by the offense more, while WR fumbles were more likely to be recovered by the defense.  I think sack-fumbles were closest to 50/50.

Points: 0

#65 by Eddo // Oct 07, 2021 - 10:35am

"Penalized" is the wrong way to look at it, though - it's more that VOA just ignores the actual recovery and gives credit based on the play type (not all types of fumbles are 50/50, for example, aborted snaps get recovered more often by the offense, and WR fumbles by the defense).

It's true that a team's score might look better than its VOA for a game if they recovered a disproportionate number of fumbles, but they don't lose VOA for recovering fumbles.

Points: 0

#3 by Tutenkharnage // Oct 05, 2021 - 5:24pm

They won't stay this impressive all year, but I'm not worried too much about regression. IIRC, they were the best defense in the league last year after week 6, so this actually fits with who they were coming into the season, all the usual disclaimers about defensive variability aside. (Defenses have been great for three or four years at a time. It's possible we're just about a year into seeing it happen again.)

Points: 0

#4 by muscle417 // Oct 05, 2021 - 5:28pm

In reply to by Tutenkharnage

They were #2 in 2018 and #6 in 2019, but slipped at the start of 2020 and turned it on late to end #12, though #6 in weighted defensive DVOA, IIRC.

Points: 0

#5 by NYChem // Oct 05, 2021 - 5:37pm

Wow, steelers have had the 2nd hardest schedule so far, and the 3rd hardest going forward.

I guess things will improve, if ever so slightly? 

Points: 0

#17 by Bobman // Oct 05, 2021 - 7:04pm

If you check out this week's Gridiron Heights on Bleacher Report or Youtube, the Pitt SOS gets a brief mention.  "Our SOS is number one, man!"  "Uh, analytics says that's a bad thing, bro."

Points: 0

#20 by ImNewAroundThe… // Oct 05, 2021 - 8:37pm

that GH is back. Love it. 

Points: 0

#25 by dbostedo // Oct 05, 2021 - 9:47pm

"I guess things will improve, if ever so slightly?"

I'm thinking that they are going to make more teams look really good, and make their schedule strength appear to be even tougher. *sigh*

Points: 0

#30 by kramerthefirst // Oct 06, 2021 - 3:50am

Yeah, the Steelers have had a tough schedule but Rothlisburger loses arm strength and football IQ weekly and the offensive line is shambles.

They just got beat up by a bad Packers' d-line.

I suppose their defense can maybe win a few games on its own but it's time to start rebuilding.

Points: 0

#39 by NYChem // Oct 06, 2021 - 10:03am

the fact that i seriously entertain the possibility that they would have been more competitive in 3 of their 4 games so far if Mason Rudolph had been playing QB, well, yeah... They HAVE the defensive personnel, when healthy, but it's tough to keep dealing with 3 and outs, and have decent special teams and skills players. Their O-line is young, but has some talent, so they may need time to gel (or maybe they need some studs in at their weaker positions, but time will tell). 

My offseason optimism has given way to the new reality. I don't want Aaron Rogers or a stopgap. But I fear they are destined for a 6-11 season and a Ben retirement, and we will reprise the end of the Chuck Noll and middle of the Bill Cowher eras for a bit with a series of underwhelming QBs, with the chance of occasional defensive-led playoff runs with Bubby Brister and Tommy Maddox-level QBs. Nick Foles in Black and Gold may be coming to a Heinz Field near me...

Points: 0

#49 by Eddo // Oct 06, 2021 - 2:05pm

"I don't want Aaron Rogers or a stopgap."

Could you imagine if the Broncos had felt this way in 2012, or the Buccaneers in 2020?

Points: 0

#56 by Tutenkharnage // Oct 06, 2021 - 6:25pm

As a Bills fan, I’m terrified of the idea of Rodgers as a Steeler. I’m sure fans of the Bengals, Browns, and Ravens feel the same way. Maybe Chiefs fans, too. 

Points: 0

#7 by bochner // Oct 05, 2021 - 5:38pm

This week will be a good test.  Somehow I don't think the Bills' defense will keep giving them the ball back at KC.  Sometimes I think we would be better off not looking at DVOA until at least Week 7.  Though that would deprive us of a lot of good conversation.  So far the Bills have played the number 25, 13, 30, and 28 offense.  KC has played against 22, 14, 4, and 6.  That's a world of difference.  The Chiefs will make the Bills play defense.  The big money will be on KC, but the squares will be all over Buffalo getting a field goal.  That said, I have a nice Buffalo to win Super Bowl ticket purchased in June and I'm not tearing it up--yet.  

Points: 0

#8 by theslothook // Oct 05, 2021 - 5:42pm

As 2007 showed, not even stomping your way to the all-time point differential record is a guarantee that you're going to win the super bowl. 

Points: 0

#12 by Tutenkharnage // Oct 05, 2021 - 6:09pm

Yes, the Bills played weak offenses. They also completely vaporized them. They’ve allowed 44 points on the year, 7 of which came on a blocked punt, with another 7 coming on a l00000ng, wind-aided fluke of an onside kick. They also gave up a completely meaningless touchdown at the end of the WFT game. Overall, they’ve been about as dominant on defense as a team could be over four games, even accounting for the weakness of the opposition.

The real test comes this week, as you said. I still have visions of Tyreek Hill going absolutely crazy in the AFCCG last year, plus Allen and McDermott coming up very small on the big stage, so I recognize the possibility that I’m going to end up disappointed. But if there’s a team that could keep the Chiefs to 28, this might be the one. Maybe it happens Sunday, maybe it doesn’t. If it does, though, the Bills are going to win the 1 seed and represent the AFC in the Super Bowl this year unless injuries destroy their season 

Points: 0

#27 by IlluminatusUIUC // Oct 06, 2021 - 12:25am

The concern I have for the Buffalo D is that Milano, Poyer, and Taron Johnson are all gimpy this week. Obviously injuries always stink but Milano and Poyer are the ones you'd want to take on Kelce. So yikes on that. 

Points: 0

#33 by Mike B. In Va // Oct 06, 2021 - 8:05am

That is absolutely a huge concern. The offense will need to step it up to compensate.

Points: 0

#36 by muscle417 // Oct 06, 2021 - 9:27am

There is optimism that Poyer and Milano, at least, suffered only minor injuries and were held out mostly as a precaution against an inferior opponent. Both were seen running and/or moving well after their injuries.

Points: 0

#57 by Tutenkharnage // Oct 06, 2021 - 6:26pm

Just heard the local radio report. There were so many names on it, I thought Bill Belichick wrote it for them. 

Points: 0

#16 by DIVISION // Oct 05, 2021 - 6:29pm

I have the Bills or Browns coming out of the AFC this year.

That said, I think whomever makes it out of the gauntlet NFC will beat anyone from the AFC, whether it's Cards, Rams or Bucs.  

Josh Allen's regression is concerning.  I don't watch too many Bills games, but these blowouts have been masking his inaccuracy.

I wish we knew why he's lost his touch.  The worst part is they haven't played any good teams yet.

Points: 0

#37 by muscle417 // Oct 06, 2021 - 9:37am

Allen played skittish for 1.5 games, it's true. He wasn't hitting throws that he can make and seemed like he was lofting throws he should have lasered. A decent part of that was the entire OL tripping over itself vs PIT's front.

In the second half vs Miami, he settled back in, getting TDs on 3 of 4 drives. Vs WFT he was unstoppable, and vs Houston he sailed his first 2 passes in a downpour and then went 20 of 27 for 258yds and 2 TDs.

It seems pretty clear that he's back in his groove now.

Points: 0

#10 by jheidelberg // Oct 05, 2021 - 6:00pm

Is there any comparable team in DVOA history to the Bills at any point in the season week 4 and beyond?  There exceptional DVOA is entirely due to defense, if there are any comparable teams, what have they done going forward? 

I see the chart of the 12 best DVOA defenses through 12 games, maybe there is some comparable team or teams in there.


Points: 0

#28 by DisplacedPackerFan // Oct 06, 2021 - 12:46am

Well that 1996 GB team had a young QB entering his 5th year as a starter, a QB who flashed promise in his first two years but really improved in his 3rd. His TD progression as a starter went 18, 19, 33, 38, 39.  His INT progression as a starter went 13, 24, 14, 13, 13.

This 2021 BUF team has a young QB entering his 4th year as a starter, a QB who flashed promise in his first two years but really improved in his 3rd. His TD progression as a starter went 10, 20, 37, ??. His INT progression as a starter went 12, 9, 10, ??. 

GB was in it's 5th year under a new coach who got them to the playoffs in 3 out of his 4 first seasons after a long run of the team not making the playoffs. BUF is in it's 5th year under a new coach who got them to the playoffs 3 out of his 4 first seasons after a long of the team not making the playoffs.

Something feels a bit similar about those teams. Sure Favre ended up starting in game 3 of that first season under Holmgren and Allen didn't start until McDermott's second season so that is some of the difference there. 

There are some QB comp differences too Favre had a much better completion % vs league average than Allen did in his first 3 years. I'm gonna list some of the traditional stats vs league average. If they were better than league average that will be a +, if they were worse it will be a - so a +1.1 on INT means they threw less int since INT are bad and a + TD means they threw more TD since TD are good. Little weird but it's for a quick visual to be consistent. A 0 means they were league average in that stat. All data from PFR. In 1992 Favre's first year as a starter he had a 64.1 comp% the league average was 57.5 so he was +6.6.  He had a 2.8 INT% vs the league average of 3.9 so again he was +1.1 (1.1 better than league average). I don't know where I an quickly get current averages if I find them I'll fill them in.

Year          Player     Comp %        TD %          INT %           DVOA
1 - 1992    Favre      64.1 (+6.6)     3.8 (0)        2.8 (+1.1)     5.9%
1 - 2018    Allen       52.8 (-12.1)    3.1 (-1.7)    3.8 (-1.4)     -35.9%  

2 - 1993    Favre      60.9 (+3.0)     3.6 (0)        4.6 (-1.3)      -5.8%
2 - 2019    Allen       58.8 (-4.7)      4.3 (-0.2)    2.0 (+0.3)     -11.8%

3 - 1994    Favre      62.4 (+4.4)     5.7 (+1.8)   2.4 (+0.7)     14.5%
3 - 2020    Allen       69.2 (+4.0)     6.5 (+1.7)   1.7 (+0.5)      25.9%

4 - 1995    Favre      63.0 (+4.8)     6.7 (+2.7)   2.3 (+0.8)     27.5%
4 - 2021    Allen       63.5 (??)         5.8 (??)      1.3 (??)         1.1%

I'm not saying Allen is Favre (though he does give me Favre vibes at times). Also Favre didn't play his rookie season, he was drafted in 91 by Atlanta, Wolf traded for him, and he took over in week 2 after Majikowski that planned GB starter got hurt. Allen may not have had that really awful rookie season if he could have sat it out. The improvement trends aren't exactly the same either but Favre's career ended up with a 62.0 comp% so he settled into that quick. a 5.0 TD% so again after the poor first 2 seasons his 3rd season was just a bit better than could be expected most years, and a 3.3 INT% but he was a player who was usually like 0 - 1% point better than league average in that, expect for the years he wasn't, and when he wasn't he was 1 to 2% points worse. Yeah he threw a lot, but that was really more a function of playing, at the time, forever. Now 19 years as a starting QB doesn't seem quite as odd.

I not only dug more into the 1996 GB comp because of how good that team started on defense being similar (and because I know GB), but because of the other similarities of a new coach and GM turning a team around, a young QB with flaws working with a coach who doesn't put up with that crap and coaching it out of him. I really like this Buffalo team, and I yes I think a shot at the SB that GB won in 96 is possible. Sure it would get the QB there one year earlier which is why 96 isn't in the "table" I made, but that was more to just track the QB progression. Favres was a bit different but he did jump a lot from where he had been and then basically sustained that for a while, with some down years and some other brilliant years for a long career.

Also that 96 GB team started 3-1 though they lost game 4, not game 1. But they put 34, 39, 42, 21 points. And gave up 3, 13, 10, 30. Not quite the same as BUF and 16, 35, 43, 40 and giving up 23, 0, 21, 0. But not hugely different either.

The comp is out there statistically and by my eyeball test and I think you should feel good about Buffalos chances. I don't think it's a mirage.

Points: 0

#35 by Pat // Oct 06, 2021 - 9:11am

Yeah. New England, 2019. It's mentioned in the article. They're really, really close analogs. Both #1, both almost entirely defense.

Obviously they still finished #1 in defensive DVOA, but not in weighted DVOA. In week 4 the team had a defensive DVOA of -45% or something like that, finished with an overall of -25%, but a weighted of -17%. Very strong fade over the year. 

Points: 0

#11 by jheidelberg // Oct 05, 2021 - 6:08pm

They have played the easiest schedule to date and their future schedule is also the easiest.  Once they play the Chiefs next week, their future schedule will be even more so the easiest.

Their schedule includes their pathetic division, the pathetic AFC South, and thus, despite having to play KC, NO, and TB, they still have the easiest schedule.

This looks like a Patriots schedule of the Brady era.


Points: 0

#13 by jheidelberg // Oct 05, 2021 - 6:16pm

The top defensive DVOA games ever includes two Super Bowls, incredible!  

There have been only 38 Super Bowls played in the DVOA era.  I am estimating that we are approaching 10,000 games played in the Super Bowl era.

Points: 0

#40 by Pat // Oct 06, 2021 - 10:44am


The top defensive DVOA games ever includes two Super Bowls, incredible!  

It's not that surprising. Opponent adjustments. Consider a regular season game: your opponent distribution is distributed across the league, so the adjustments will have the full range span (up and down). Now consider the Super Bowl - your opponent distribution is skewed much higher, so your opponent adjustment range is almost all positive. 

Basically if "Super Bowl-like matchups" only occur around ~10 times or less (sounds about right) in a season, it's what you'd expect.

Points: 0

#48 by jheidelberg // Oct 06, 2021 - 1:49pm

Opponent adjustments on chart: NYG 8th, NE 11th, these were hardly powerhouse offenses.  These were truly tremendous defensive efforts by the Ravens and Besrs.

Points: 0

#50 by Pat // Oct 06, 2021 - 2:29pm

Opponent adjustments on chart: NYG 8th, NE 11th, these were hardly powerhouse offenses. 

Adjustments aren't done by totals or by ordinal rankings, they're done on a play basis. The Giants in 2000 had a pass offense of 27.2%, which was ordinally 8, but basically in the same "upper tier" as the others It was their rush offense that was poor, and they barely ran that game. Same thing in '85. Patriots barely rushed, and their passing offense was 11.6%.

Weighting by their run/pass distribution (to approximate how total DVOA is actually calculated, which is a pure average of play results), the Giants would've had an 'expected' offensive DVOA of 19%, 4th in the league. Patriots would've been around 6-7th.

These were truly tremendous defensive efforts by the Ravens and Besrs.

Yes, of course they were, that's not what I'm saying. Obviously it's a tremendous defensive effort when you shut down offenses that much! But there's only so much you can shut down another team. You can't intercept someone if they keep handing off!

There's a second effect that boosts Super Bowls a bit, too: teams shut down later as well, because there's no "long game" to play.

Points: 0

#53 by jheidelberg // Oct 06, 2021 - 3:48pm

1.  You have moved the offenses up the ladder, still neither were dominating offenses.

2.  The assumption that you need a huge opponent adjustment to make this list is false, look at Buffalo vs Houston this past Sunday

3.Teams shut down later in the Super Bowl???   Any player would keep playing if the odds were 1000-1 of winning.  Even garbage time, the thrill of a Super Bowl TD would be enough for any player I must assume.

NE did not shut down vs ATL nor did SF vs Baltimore.  When does the alleged shutdown occur?

There is always a long game in the NFL, the Chiefs continued to let Mahomes take a beating, going for the miracle comeback, risking the torn ACL that would have ruined this season.

I was simply point out how odd it is to have 2 SB games on this list, not trying to give a reason as to why this occurred.

Maybe a statistical anomaly, two all time great defensive games in 38, 10 others in nearly 10,000 other games, this is massively statistically significant I presume.

Maybe there is an explanation other than statistical anomaly, but I choose to not buy any of your reasons.

Points: 0

#54 by Pat // Oct 06, 2021 - 4:15pm

The assumption that you need a huge opponent adjustment to make this list is false, look at Buffalo vs Houston this past Sunday

"If A then B" does not imply "if not A then not B."

Teams shut down later in the Super Bowl???   Any player would keep playing if the odds were 1000-1 of winning.  Even garbage time, the thrill of a Super Bowl TD would be enough for any player I must assume.

Yeah, you read that wrong. The "later" meant "later than in normal games," most often "not at all." As in, during the regular season during dominant victories, at a point teams just concede and you've got garbage time (so the defensive dominance doesn't show up as much). That doesn't happen in the Super Bowl, like you're saying.

Points: 0

#55 by BJR // Oct 06, 2021 - 5:06pm

Reading this I was curious how last year's Bucs Super Bowl defensive performance stacked up, given they held an absolutely monstrous offence without a touchdown. The answer is -54% DVOA for the game, so a long way off either of those Ravens/Bears performances. They did, however, give up 6.2 yards per carry (on 17 carries) to a previously mediocre rush offence, so I assume this is what is holding them 'down'. The pass defense is probably comparable, given the strength of the opponent.

I would also be interested in the Seattle rating vs. Denver in 2013 but I can only seem to access single game DVOAs for regular season games (I had to dig back for the article for last season's game). 

Points: 0

#18 by Lost Ti-Cats Fan // Oct 05, 2021 - 7:05pm

I'd be curious to know whether the inverse situation - teams projected to have an average DAVE, that ended up with a significantly worse DVOA after 4 weeks - similarly were more likely to stay at a low performance level than a 30/70 blend of DVOA and DAVE would predict?

Intuitively, it seems likely to me that a significant performance variance is more likely to be - well, significant - as a signal than more modest performance variance.  That is, the greater the difference between actual early performance and DAVE, the more likely it is that DAVE got things wrong. 

It wasn't apparent to me, though, from the table of comparable teams to this year's Cards, whether the week 4 blend of DVOA/DAVE was more or less accurate than week 4 DVOA alone at projecting DVOA for weeks 5 to 16?  In other words, I couldn't tell how much the variation in final DVOA was a result of DVOA that was already "baked in" from performance for weeks 1-4.  Maybe it's not that simple because of opponent adjustments, which take time to crystalize.  But if a team you predicted to be average has a really good first 25% of a season, then even if they're average the rest of the way, they're going to look better than average at the end because of that early performance.

Points: 0

#21 by MJK // Oct 05, 2021 - 8:41pm

Curious... what was the single game VOA splits for NE-TB?  TB strung together a lot of successful running plays, which VOA should like, but struggled to convert 3rd downs and high leverage situations.  NE had a few good strung together drives, but didn't run at all and so necessarily threw a lot of incompletes or short completions.  So I imagine that VOA probably thought TB played way better on average, and NE was "lucky" to be in it right at the end.  

On the other hand, I bet the passing split favors NE... to my eye, Jones outplayed Brady.  But I have subjective biases.  I'm curious what the numbers say...

Points: 0

#22 by MJK // Oct 05, 2021 - 8:43pm

... Edit, I just realized that as a FO+ subscriber, I can access the single game DVOA.  Looks like it thinks NE outplayed TB by quite a bit, especially on defense, and that TB's special teams was terrible.  

I'm still curious about the pass versus run VOA splits.  Is that in a table somewhere, or can you tell us?



Points: 0

#42 by Ambientdonkey // Oct 06, 2021 - 10:59am

I didn't know I could do that and would have never thought about it if you hadn't brought it up.

Points: 0

#24 by TomC // Oct 05, 2021 - 9:34pm

Does anyone think that KC will score fewer than 28 points this Sunday?

(And, in case it's not obvious from that question, I think the 2021 Bills D fails the eye test. Reminds me of the 2012 Bears, who also beat up on awful teams and ran up ridiculous turnover numbers but---at least in this Bear fan's opinion---weren't really that good.)

Points: 0

#26 by IlluminatusUIUC // Oct 06, 2021 - 12:21am

In reply to by TomC

For what it's worth, the Bills held the Chiefs to 26 in their regular season meeting last year. 

Points: 0

#31 by Tutenkharnage // Oct 06, 2021 - 7:05am

In reply to by TomC

The Chiefs gave up 29 to the Browns in week 1 and have given up at least 30 points every week since, including to the Eagles. Does anyone think they’re going to hold Buffalo under 31?

Now, as another person replied, what I described happened last year—twice, in fact—so it’s certainly possible. Sean McDermott and Brian Daboll might turtle again. The Bills might end up letting CEH run all over them. Or they might consistently get past a disappointing OL with just four rushers and generate enough pressure on Mahomes to win.

Points: 0

#38 by TomC // Oct 06, 2021 - 9:46am

Oh, I am not in any way saying the Chiefs are guaranteed to win. I'm just saying that if Buffalo had an all-time defense, they should be able to hold even the Chiefs to a modest number of points, and I don't think they are going to. I'm expecting a score like 41-38, and I think it's almost a toss-up which team ends up with the 41.

Points: 0

#43 by RickD // Oct 06, 2021 - 11:59am

The Chiefs' offense is very, very good.  Right now they're O/D split is mind-boggling: #1 offensive, #32 defense.  That terrible defense is why I think the Bills will win this game, but I have no feel for how much the Chiefs will score.  

Over-under is 56-56.5.  So Vegas, at least, thinks it'll be a high-scoring game.

Points: 0

#52 by TomC // Oct 06, 2021 - 3:24pm

If I were a gambler, I'd be all over that over. The Bills might almost get there by themselves.

Points: 0

#67 by Tutenkharnage // Oct 08, 2021 - 10:06am

I'm not convinced at all the Bills are anything other than what Vegas is implying: a 3-to-2 underdog to win the game. But in both games last year, it felt as if the Bills were the only one of these teams with questions to answer. (And their answers were terrible!) This year, by contrast, the Chiefs feel as if they have some questions to answer, too.

  1. Can their defense stop anyone?
  2. Can their offensive line hold up?

If the Bills can consistently generate pressure with their front four throughout the game, I'm going to feel pretty good about their chances of winning. It won't be a given, even with the pressure, but I'll end up feeling they're at least 50/50. I think there's a good chance the Chiefs' answers to those two questions are "no," in which case the Bills will be in great shape as long as (1) McDermott doesn't get conservative and (2) Allen doesn't freeze up.

Points: 0

#32 by BigBen07 // Oct 06, 2021 - 8:01am

Mahomes is unsurprisingly leading the best offense in the NFL, but is saddled with the WORST defense in the NFL. I guess he can't will them to succeed like TB12 could.

Points: 0

#44 by RickD // Oct 06, 2021 - 12:00pm

Still have sad memories of Super Bowl XLVI.

Points: 0

#51 by Cheesehead_Canuck // Oct 06, 2021 - 3:04pm

In reply to by RickD

Ah yes, one of the few years in a 21 season and counting career that Brady didn't have a above average to outstanding defense behind him.

Points: 0

#59 by greybeard // Oct 06, 2021 - 7:02pm

They spend a lot more money on offense (even with Mahomes cap hit being really low for this year, which starts accelerating next year) than defense.

Points: 0

#61 by greybeard // Oct 06, 2021 - 7:13pm

I was totally wrong. I read the data incorrectly. They do spend a lot more on defense than offense. Which makes this even worse. Having said that it seems like 65% of the money that goes to defense is spent on 5 players. Which might explain why they have a lot of weak links.

Points: 0

#68 by bedfordp // Oct 09, 2021 - 4:37am

Giving up 146 yards per game and 8 TD on the season not great and no real effort to sort that out in the offseason 

Points: 0

Save 10%
& Support Aaron
Support Football Outsiders' independent media and Aaron Schatz. Use promo code SCHATZ to save 10% on any FO+ membership and give half the cost of your membership to tip Aaron.