Writers of Pro Football Prospectus 2008

Most Recent FO Features


» Four Downs: NFC West

Our offseason Four Downs series ends with a look at the NFC West's biggest remaining holes and their most notable UDFA signings. The Rams and 49ers have to kick-start their passing games, Arizona's offense lacks a big dimension, and the Seahawks continue to rely on Russell Wilson's magic tricks.

02 Dec 2014

Week 13 DVOA Ratings

by Aaron Schatz

Now that the Arizona Cardinals have been nice enough to fall back to earth, the Football Outsiders DVOA ratings are looking a lot more like the actual NFL standings. Three of the five current 9-3 teams are our DVOA top three, and all three teams -- Denver, New England, and Green Bay -- see very little change from last week's ratings. The fourth 9-3 team, Philadelphia, is now seventh in total DVOA but fifth in weighted DVOA. And then the Cardinals... well, things haven't changed much in our numbers, which don't see their two losses as being worth much less than some of their early-season wins.

After the Packers beat the Patriots, I noted on Twitter that I expected both teams to come out of the game with positive DVOA ratings, and that's in fact what happened. The Patriots actually ended up with a higher DVOA for the game, despite the fact that the Packers gained more yards per play, 6.8 to 5.9. Slight differences add up. The Patriots had a slightly higher success rate on offense (44.4 percent vs. 42.6 percent). Mason Crosby loses more value for missing a 40-yard field goal than Stephen Gostkowski does for missing a 47-yard field goal. The Packers also get dinged for Aaron Rodgers' fumble on a sack (which he recovered himself) and Micah Hyde's muffed punt return. The numbers end up looking like this, with and without the opponent adjustments:

GB 28.9% 11.3% -9.6% 8.0%
NE 36.2% -0.6% -4.9% 31.9%
GB 27.0% 29.4% -9.6% -12.0%
NE 34.5% 21.7% -4.9% 7.9%

Both teams see their overall DVOA ratings drop by less than one percentage point this week; the Patriots fall despite having the slightly higher DVOA in this game because of changes in schedule adjustments for their previous opponents. But both teams also move up in weighted DVOA because Week 1 -- when the Packers and Patriots both lost -- has now dropped to only 20 percent strength in the formula. The gap between the Patriots and Broncos is nearly 10 percentage points of total DVOA, but is less than two percentage points in weighted DVOA. Meanwhile, the Packers lead over the rest of the NFC becomes bigger when we look at weighted DVOA, and the second-best team switches from Seattle in total DVOA to Philadelphia in weighted DVOA because we're discounting that wacky Week 1 Jacksonville game where the Eagles were competely outplayed for a half.

The Patriots have been the best team in the league since Week 7. This is, of course, completely arbitrary endpoint-setting, but at least it's arbitrary endpoint-setting that has nothing to do with the Patriots. It comes from something I put together for the ESPN playoff odds report today, looking at how the Saints have improved since their Week 6 bye. Anyway, as long as I ran these numbers, I figured folks would find them interesting. Here's a look at teams with an overall DVOA difference of at least 20% between Weeks 1-6 and Weeks 7-13:

Week 1-6 DVOA vs. Week 7-13 DVOA, Biggest Change
Weeks 1-6 Weeks 7-13
NE 4.0% 0.7% 2.9% 6.2% 15 32.8% -2.3% 8.8% 43.9% 1 37.6%
NO 15.6% 30.1% -0.6% -15.0% 25 18.1% 1.5% 2.5% 19.2% 7 34.2%
TB -24.8% 14.3% -5.6% -44.6% 32 -24.9% -13.5% -1.1% -12.5% 23 32.1%
PIT 9.6% 19.6% 0.6% -9.4% 21 25.1% 3.1% -1.4% 20.7% 6 30.1%
MIN -22.8% 4.1% 1.0% -26.0% 30 -3.6% -0.9% 5.9% 3.2% 16 29.1%
JAC -29.6% 9.3% -4.5% -43.5% 31 -25.8% -12.7% -2.6% -15.8% 25 27.7%
STL -3.3% 12.9% -4.3% -20.4% 26 -13.3% -11.3% 6.4% 4.4% 13 24.8%
x x x x x x x x x x x x
CHI 9.7% -2.7% -7.2% 5.1% 16 -7.1% 18.7% -0.9% -26.7% 28 -31.8%
CLE 18.2% 7.5% 0.4% 11.1% 8 -21.1% -5.8% -1.8% -17.1% 26 -28.2%
CAR 7.2% 14.9% -2.5% -10.2% 23 -26.4% -3.0% -10.2% -33.5% 32 -23.3%
WAS 3.8% 4.7% -9.3% -10.2% 22 -21.6% 11.1% -0.6% -33.4% 31 -23.2%
CIN 15.1% 0.3% 0.1% 14.9% 5 -15.1% 1.0% 7.8% -8.3% 22 -23.2%
NYG -1.2% -1.4% -3.4% -3.1% 19 -11.5% 16.0% 1.8% -25.6% 27 -22.5%
DAL 17.6% 1.1% -2.3% 14.2% 6 3.6% 12.1% 2.3% -6.2% 21 -20.5%

Yes, that's Carolina with the worst special teams of Weeks 7-13. And, of course, that's heavily just Week 13, where the Panthers somehow allowed two blocked punts to be returned for touchdowns. Carolina had -50.9% special teams DVOA for this game, and our system estimates that special teams cost the Panthers -15.9 points compared to an average performance. That's not quite enough to make me dig out the list of the worst special teams games ever, but I'm sure it would at least make the bottom 25. And that gets me to this question which came up in the comment thread for Quick Reads today:

Thunderbolt of ... : Once a punt is blocked, does DVOA reward/penalize special teams for recovering the ball and/or returning it for a TD? I'm not sure, but my guess is that DVOA treats a blocked punt like a fumble, where credit is given for the block but not for the recovery. But I'm not sure this is as accurate in the case of a blocked punt; a clean block will almost always be recovered by the defense, while a partial block isn't as good.

The answer to this question is that right now, I've got it in the system where the punting team gets penalized for the blocked punt based on where the ball ends up after a recovery and return, but the punt return team doesn't get any credit for what happens after the block. Honestly, that's pretty silly and is an example of one of those small things in our ratings that I probably set ten years ago and don't make sense given how we handle other things now. One of my goals for the offseason is to overhaul the special teams ratings to really segregate these special, "non-repeatable" but valuable plays like punt blocks, maybe even creating two different ratings for special teams -- one that is only for looking backwards at how well a team played that day, the other that only measures elements of special teams that are likely to tell us something about that team going forward. When I do this, I'll fix the punt blocks so they make more sense.

* * * * *

Once again in 2014, we have teamed up with EA Sports to bring Football Outsiders-branded player content to Madden 15 Ultimate Team. Each week, we'll be picking out a handful of players who starred in that week's games. Some of them will be well-known players who stood out in DVOA and DYAR. Others will be under-the-radar players who only stood out with advanced stats. We'll announce the players each Tuesday in the DVOA commentary article, and the players will be available in Madden Ultimate Team packs the following weekend. We will also tweet out images of these players from the @fboutsiders Twitter account on most Fridays. One player each week will only be available for 24 hours from the point these players enter packs on Friday.

The Football Outsiders stars for Week 13 are:

  • WR DeAndre Hopkins, HOU (24-HOUR HERO): 133 DYAR, the No. 5 best WR game since 1989 (9-for-9, 238 yards, 2 TD).
  • RT Lane Johnson, PHI: Allowed no sacks, one hurry; Eagles RB gained 118 yards on 10 carries to the right side.
  • DT Akeem Spence, TB: 4 TFL, 2 QB hits, sack.
  • CB Desmond Trufant, ATL: Allowed just three catches for 40 yards against Arizona.
  • LOLB Jason Worilds, PIT: 10 combined tackles, all of which were Stops, including a sack, run TFL, and pass reception TFL. Tackles allowed an average gain of just 1.2 yards.

* * * * *

All stats pages are now updated with Week 13 information -- or will be in the next few minutes -- including FO Premium, snap counts, and playoff odds. You can also read the new weekly playoff odds report on ESPN Insider to get more commentary on the current playoff odds.

* * * * *

These are the Football Outsiders team efficiency ratings through 13 weeks of 2014, 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 strength of schedule and 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. WEIGHTED DVOA represents an attempt to figure out how a team is playing right now, as opposed to over the season as a whole, by making recent games more important than earlier games. As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.

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 DEN 34.6% 1 33.0% 1 9-3 23.8% 2 -15.3% 4 -4.5% 27
2 NE 25.3% 2 31.1% 2 9-3 18.7% 3 -0.8% 12 5.8% 3
3 GB 23.9% 3 25.2% 3 9-3 25.4% 1 0.5% 16 -1.0% 19
4 BAL 21.6% 4 24.4% 4 7-5 10.7% 9 -2.2% 9 8.6% 2
5 SEA 21.4% 6 16.4% 7 8-4 12.6% 7 -9.7% 6 -0.9% 18
6 MIA 16.4% 5 16.2% 8 7-5 10.6% 10 -9.6% 7 -3.8% 25
7 PHI 16.0% 8 20.0% 5 9-3 -1.4% 15 -8.6% 8 8.8% 1
8 KC 13.8% 7 18.0% 6 7-5 8.3% 11 0.1% 14 5.6% 6
9 BUF 9.4% 9 10.6% 9 7-5 -12.4% 27 -16.0% 3 5.8% 4
10 IND 9.1% 12 9.1% 11 8-4 4.6% 13 1.2% 19 5.7% 5
11 DET 7.2% 14 5.8% 13 8-4 -6.0% 20 -18.9% 1 -5.7% 31
12 SF 6.8% 11 4.9% 15 7-5 -4.7% 18 -16.5% 2 -5.0% 30
13 PIT 5.6% 13 5.6% 14 7-5 17.6% 4 11.5% 29 -0.4% 14
14 NO 5.3% 18 9.8% 10 5-7 17.1% 5 12.9% 31 1.2% 11
15 DAL 4.0% 10 5.9% 12 8-4 11.1% 8 7.1% 24 0.0% 13
16 SD 1.8% 16 2.2% 16 8-4 12.8% 6 10.2% 28 -0.8% 16
17 CIN 1.4% 15 -3.0% 19 8-3-1 -2.5% 16 0.7% 17 4.6% 7
18 ARI -1.8% 17 -1.9% 18 9-3 -9.9% 24 -10.9% 5 -2.8% 22
19 ATL -3.1% 20 -6.9% 22 5-7 7.7% 12 15.2% 32 4.4% 8
20 STL -4.9% 24 1.8% 17 5-7 -8.5% 22 -1.7% 10 1.9% 10
21 CLE -5.6% 19 -6.3% 21 7-5 -4.9% 19 -0.2% 13 -0.9% 17
22 HOU -8.2% 21 -6.2% 20 6-6 -3.1% 17 0.3% 15 -4.8% 28
23 CHI -10.7% 22 -14.5% 24 5-7 1.9% 14 8.6% 27 -4.1% 26
24 MIN -11.7% 23 -13.2% 23 5-7 -13.7% 28 1.5% 20 3.4% 9
25 NYG -14.6% 25 -15.3% 25 3-9 -6.3% 21 7.5% 25 -0.8% 15
26 NYJ -18.4% 28 -18.0% 26 2-10 -15.4% 29 1.7% 21 -1.3% 20
27 WAS -21.4% 27 -22.9% 27 3-9 -8.7% 23 7.8% 26 -4.9% 29
28 CAR -22.4% 29 -25.8% 28 3-8-1 -10.0% 25 6.1% 23 -6.3% 32
29 TEN -26.2% 30 -31.0% 32 2-10 -12.1% 26 11.8% 30 -2.2% 21
30 OAK -28.5% 26 -28.5% 30 1-11 -24.6% 30 4.5% 22 0.5% 12
31 TB -29.4% 32 -26.5% 29 2-10 -24.8% 31 1.2% 18 -3.3% 23
32 JAC -29.9% 31 -29.2% 31 2-10 -27.7% 32 -1.3% 11 -3.6% 24
  • 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).

1 DEN 34.6% 9-3 31.1% 10.5 1 4.6% 4 -4.0% 21 11.1% 7
2 NE 25.3% 9-3 22.6% 7.8 7 3.9% 6 2.3% 12 18.8% 25
3 GB 23.9% 9-3 24.9% 8.2 3 0.5% 13 -4.0% 20 14.4% 12
4 BAL 21.6% 7-5 26.5% 8.3 2 -4.7% 25 -6.8% 25 8.5% 5
5 SEA 21.4% 8-4 19.3% 7.9 6 -0.7% 21 4.0% 7 16.9% 19
6 MIA 16.4% 7-5 19.0% 8.0 5 3.1% 7 4.2% 6 18.4% 24
7 PHI 16.0% 9-3 17.0% 8.0 4 -7.1% 30 -2.6% 19 14.8% 15
8 KC 13.8% 7-5 10.7% 7.4 8 6.0% 3 -5.7% 24 23.1% 29
9 BUF 9.4% 7-5 9.9% 6.6 12 0.7% 12 13.8% 3 5.4% 1
10 IND 9.1% 8-4 13.8% 7.3 9 -2.1% 22 -9.0% 29 12.5% 9
11 DET 7.2% 8-4 9.2% 6.9 10 -0.2% 19 -7.0% 26 8.0% 4
12 SF 6.8% 7-5 2.3% 6.8 11 3.1% 8 -1.8% 18 11.5% 8
13 PIT 5.6% 7-5 6.8% 6.0 18 -7.3% 31 3.4% 9 15.0% 17
14 NO 5.3% 5-7 4.7% 6.3 16 -0.1% 17 -16.4% 32 18.8% 26
15 DAL 4.0% 8-4 5.5% 6.6 14 -6.0% 27 -1.7% 17 18.2% 22
16 SD 1.8% 8-4 5.0% 6.6 15 0.4% 14 20.1% 1 18.3% 23
17 CIN 1.4% 8-3-1 4.6% 6.6 13 -3.5% 23 10.1% 4 23.2% 30
18 ARI -1.8% 9-3 -1.1% 6.0 17 1.6% 11 9.3% 5 6.1% 2
19 ATL -3.1% 5-7 3.4% 5.6 20 -7.5% 32 3.1% 10 14.7% 13
20 STL -4.9% 5-7 -5.0% 4.8 23 2.8% 9 -4.1% 22 29.2% 32
21 CLE -5.6% 7-5 3.0% 5.6 19 -6.4% 28 2.4% 11 13.6% 10
22 HOU -8.2% 6-6 0.7% 4.4 25 -6.4% 29 -7.3% 27 6.3% 3
23 CHI -10.7% 5-7 -14.0% 5.4 21 2.3% 10 1.2% 13 11.0% 6
24 MIN -11.7% 5-7 -11.5% 5.3 22 0.3% 16 -1.4% 16 14.9% 16
25 NYG -14.6% 3-9 -21.1% 4.0 27 0.4% 15 -9.1% 30 16.3% 18
26 NYJ -18.4% 2-10 -24.0% 4.4 24 9.0% 1 0.9% 14 18.2% 21
27 WAS -21.4% 3-9 -17.7% 3.6 28 -5.4% 26 0.1% 15 20.3% 27
28 CAR -22.4% 3-8-1 -27.6% 4.1 26 4.0% 5 -8.2% 28 14.1% 11
29 TEN -26.2% 2-10 -24.6% 3.1 30 -0.1% 18 -13.5% 31 17.9% 20
30 OAK -28.5% 1-11 -36.6% 3.0 31 6.4% 2 16.2% 2 14.8% 14
31 TB -29.4% 2-10 -21.3% 3.2 29 -4.1% 24 3.5% 8 23.9% 31
32 JAC -29.9% 2-10 -26.5% 2.4 32 -0.4% 20 -5.2% 23 21.1% 28

Posted by: Aaron Schatz on 02 Dec 2014

182 comments, Last at 06 Dec 2014, 7:37pm by LionInAZ


by ryanwanger :: Tue, 12/02/2014 - 7:55pm

I'm surprised the DVOA difference was so large for NE vs GB.

Maybe this is my Green Bay homerism, but if that game were played again, and each team were to play as well as they did before, I would take the Packers every time. They left a LOT of points on the board. Yes, yes, NE had good red zone defense, but you really can't *plan* to let the number 2 offense into the red zone on every drive and then hope to suddenly play better and stop them.

I'm also surprised specifically that holding the #3 offense in the league to 21 points is considered well below average (11.3%).

With the 23.9% DVOA advantage, are we to believe that the Patriots will win that game most times, playing the way the did on Sunday?

by oaktoon :: Tue, 12/02/2014 - 8:00pm

(Still biting my tongue) I'm happy with the final score.. I hope Belichick comes to the same conclusion as DVOA, but I suspect his metrics will come to a slightly different conclusion and lead to a very different game plan if there's a rematch.

by oaktoon :: Tue, 12/02/2014 - 8:02pm

and to your point, play it again and Patriots win once, packers win by less than 5 once or twice, packers win by 5-10 five or six times, packers win by 2TDs or more once or twice

by Perfundle :: Tue, 12/02/2014 - 8:54pm

Yes, yes, NE had good red zone defense, but you really can't *plan* to let the number 2 offense into the red zone on every drive and then hope to suddenly play better and stop them.

But equally, you also shouldn't plan on drives continually getting the chance to stall in the red zone. Green Bay faced an unusually large number of third downs in the game, and converted a higher-than-expected percentage: 10-16 for 62.5%, compared to 46.8% for the first 11 games; since they failed all four times inside the red zone, that meant they were an incredible 10-12 outside of it. Facing a lot of third downs is generally not a sustainable model going forward, and Green Bay had been very good at avoiding them previously, having the fifth-lowest third down play percentage.

by NYMike :: Tue, 12/02/2014 - 10:25pm

I think this is right. Because GB was so good on third down, on both sides of the ball, I think it masked from viewers (and I include myself) what DVOA was saying. I think, just remembering, and not looking it up, that NE tended to be successful on first down, and then the GB defense came up with big plays on second and third to stop them, whereas GB often started a series by running for two yards on first down. That seemed to happen over and over, but Rodgers and Co were able to overcome it, especially in the first half, so it ended up not mattering. But it would matter to DVOA.

by Anon Ymous :: Tue, 12/02/2014 - 10:38pm

Yeah, after the initial jolt, the DVOA ratings are starting to make more sense. I'm not sure I agree with them, but I can understand them. ;-)

I do lean in the direction of DVOA in that I think I might favor NE in a rematch. Aside from the fact that it wouldn't be in Lambeau, I feel like we saw a closer approximation to GB's best game than we did NE's. It is completely anecdotal, and I'm sure someone is going to scold me for not having 12 spreadsheets of data to support this, but it seems like the spread usually narrows or even reverses the second time around.

by chemical burn :: Tue, 12/02/2014 - 10:46pm

Really? GB's special teams had a ton of easily avoidable gaffes. In close games, special teams play ends up being a big factor. But I agree with you insofar as "has any team ever gone up against NE multiple times in a season and won all the match-ups?" I think coaching becomes the biggest factor in any rematch. However, I don't think a rematch is happening. If Seattle blows out Philadelphia this weekend, I thinks it's a Superbowl rematch all the way...

by Anon Ymous :: Tue, 12/02/2014 - 11:14pm

NYJ, Baltimore, and NYG are recent examples of teams beating NE more than once. Even in those examples, the defense generally plays a good deal better the second time around.

I understand GB wasn't perfect, which is why I said it was closer to and not their actual best. Has any team ever played their absolute best?

by dmstorm22 :: Tue, 12/02/2014 - 11:58pm

You're also missing the '05 Broncos and '06 Colts. In fact, every team to beat New England in teh playoffs had played them that regular season. ('07 Giants, '09 Ravens, '13 Broncos lost their games)

by Anon Ymous :: Wed, 12/03/2014 - 8:45am

My list wasn't meant to be all inclusive, that's why I said "recent". Your latter list is more tangential because the OP was looking for teams that beat NE multiple times, not those that faced them twice and won the second time around.

by chemical burn :: Wed, 12/03/2014 - 12:08am

I really did mean sweeping, not beating two out of three times. NYG I had forgotten about. Didn't realize Baltimore had beaten them either of those seasons before the playoffs. If GB beats them again, it will add them to the list of Superbowl winner who went through them twice, so that's fitting I guess!

As far as teams playing their absolute best, that not what I meant even - there was room for improvement in simple ways for GB. Not sending kickoffs short and out of bounds is a simple thing to do, so while I could've suggested "well, Aaron Rodgers could have scrambled such and such situation or completed an open pass he didn't see," I didn't suggest that they could have improved where they were indeed excellent. Their running game has definitely looked better at other point than it did on Sunday, but I didn't mention that either - I saw one area where they almost certainly would be improved in a rematch.

They COULD be better in other ways in a rematch, but they almost definitely WOULD be better on kickoffs. Also, their obvious redzone stumbles are an area where the would vs. could is debatable, but would be surprising to be repeated - either by failing to get into the redzone as frequently or their failure to capitalize on the end of drives where they moving the ball well. So, saying that's as good as GB could have been seems dubious to me.

I agree that New England seemed to have more options overall for improvement than GB, so maybe that's all you meant. I certainly think NE wouldn't be as poorly coached next time out - they employed many puzzling strategies and underutilized effective ones. That's why I brought up teams rarely sweeping Belichick in a season - relying on him being out-coached twice seems more unlikely than for a second time Aaron Rodgers playing like the MVP and their pass defense being ok but not excellent (which at 14th in DVOA, seems about right.) I'm not sure that any team in the league but Seattle is capable of shutting down Rodgers/Nelson/Cobb, let alone a defense like NE's that the team's fans are really excited for but DVOA (and my own eyes) think is a very pedestrian unit. I think they were lucky to hold GB to under 30 and most of that was from redzone excellence.

So, no, I don't agree with the idea that GB had topped off. I don't think NE topped off by any stretch of the imagination either, though.

by Anon Ymous :: Wed, 12/03/2014 - 8:45am

I'm right there with you. Simply by virtue of settling for a single FG, let alone four, it can be assumed GB didn't hit their apex. I just think NE was further away than GB was, a purely subjective opinion, I realize.

Regarding special teams, NE was pretty poor in that aspect themselves. Beyond the missed FG, every KO was returnable, the coverage units were only OK and Allen had a couple below average punts. It was the worst that group has looked all year, by a healthy margin. I'm not familiar enough with GB to know how much their performance diverged from the norm (or how mch they might have had to do with NE's struggles in coverage), but it should be noted that NE offset GB's issues somewhat.

by nuclearbdgr :: Wed, 12/03/2014 - 1:20pm

I think that kickoffs in 10-20 degree weather are generally returnable. That is one reason Crosby doesn't rank higher in terms of number of touchbacks - kicking a cold ball is hard (literally).

by Dave :: Wed, 12/03/2014 - 12:48am

Seattle came pretty close in February.

by Perfundle :: Wed, 12/03/2014 - 1:00am

Racking over 100 yards of penalties is a sign that they didn't play their best, and Lynch didn't play well in that game either. I would say that they were closer to their best against New Orleans than against Denver.

by chemical burn :: Wed, 12/03/2014 - 1:09am

Yeah, the New Orleans game is the one where I felt like "wow. there is no way to beat this team." The Denver game was just one of those things where some fluke-ish stuff went their way early and then shit got out of hand quick.

by dmstorm22 :: Wed, 12/03/2014 - 1:15am

I'm assuming you are both talking about the regular season game against the Saints? If so, yes, that was a frightening performance.

In the playoff game, hard to say they were close to their best when Wilson goes 9-18 and misses routine slants.

by Perfundle :: Wed, 12/03/2014 - 2:40am

Yeah, the regular season one.

I just realized that Lynch didn't play well in that one either, but I think that was more because they were stacking the box and sold out to stop Beastquake from happening again, whereas against the Broncos it was more about Knighton owning Unger again and again. Also, Wilson and the defense probably had their best games since 2012 as far as technique is concerned against the Saints. The defense shut down Peyton too, but that was more because he really wasn't that accurate and, after a careful review of their play, they probably did figure out several of Manning's signals considering their uncanny knowledge of some of the routes. Brees was actually pretty accurate throwing the ball and they managed to temporarily shake off the coverage with some well-designed routes several times, but the anticipation, recovery and pass disruption by the secondary was amazing that day.

by dmstorm22 :: Wed, 12/03/2014 - 1:16am

Deleted for double post.

by chemical burn :: Wed, 12/03/2014 - 2:38am

The one where they held Brees to like 150 yards. Pretty sure it was the regular season, but now you have me doubting myself.

EDIT: Yes, it was the regular season game. Although Lynch was very pedestrian in that one.

by Led :: Wed, 12/03/2014 - 8:36am

GB also had, I think, 3 pass deflections by the front 7, and NE had the good fortune that all three deflected passes happened to land out of reach of the GB defenders. A little worse luck for NE (or better luck for GB) and one or two of those is intercepted. Although DVOA doesn't reflect it, I think that turnover "luck" balances (if not outweighs) the non-muffed punt and Rodgers "fumble."

by Led :: Wed, 12/03/2014 - 11:39am

Double post.

by badger25 :: Wed, 12/03/2014 - 10:39am

NYMike: Eddie Lacy carried the ball 21 times for 98 yards with a long gain of 12, so I'm not sure when all those two yard runs you're referring to happened. Both teams ran the ball effectively, but Green Bay averaged two more yards per pass, which is a pretty significant advantage with neither team turning the ball over. Plus, New England's defense is rated higher than Green Bay's, so shouldn't that help the Packers' offensive DVOA, relative to the Pats'? The DVOA percentages don't make sense to me, even after reading the explanation, but I'm probably not going to lose any sleep over it.

by NYMike :: Wed, 12/03/2014 - 11:59am

Well, you made me look it up. Here's GB on first down in the first half:

Lacy 13
Lacy 24
Lacy 2
Cobb 2
Pass Inc
Lacy 2
Lacy 1
Starks 2
Lacy 3
Pass Boykin 6
Lacy -2
Lacy 2
Pass Inc
Pass Nelson 45

I think I remembered pretty accurately. Add to the fact that I didn't see the first two plays on this list because the Ravens/Chargers game hadn't ended yet ...

Anyway, in a half where GB scored 5 times in 5 possessions, they had 4 successful first down plays in 14 tries. I think we've discovered why DVOA didn't like the Packers as much as we fans did.

by oaktoon :: Wed, 12/03/2014 - 12:29pm

don't disagree with your point, but you missed some stuff.

The first play of the game was a 12 yard pass to Lacy. You omitted it.

There was another 12 yard pass on 1st down you missed-- to Cobb, I believe.

The 32 yard TD pass to Rodgers was on 1st down.

And the Nelson TD pass was a 3rd and 2, actually.... preceded by a 8 yd 1st and 10 pass to Nelson that you also missed

Overall (I chart GB 1st down performance regularly) they had 18 1st down plays for 121 yards-- nearly 7 yds per play-- which is way better than average for a NFL team... But I agree there were too many 2 yard runs-- McCarthy has a different philosophy than a lot of us-- it can be frustrating to watch at times-- they believe in "explosive" plays and, while they don't say this explicitly, they use bland 1st and 10 running plays to setup the defenses for those big plays. And again, despite playing at Lambeau, McCarthy-Rodgers has outscored every other NFL coach-QB pairing in history, save Peyton-Brees, so they must be doing something right.

by badger25 :: Thu, 12/04/2014 - 12:38am

NYMike: The original stat line I saw on Lacy was wrong (or maybe I misread it). His long run was 24, as you said, not 12, which of course means that he had a number of shorter runs. So, I stand corrected on that. That said, I still don't think the Packers' third down conversion rate was all that unusual, and I still contend that, all other things being equal, which they pretty much were on Sunday, if Green Bay has a positive two yards per pass differential and no picks, it wins this game almost every time.

by Perfundle :: Thu, 12/04/2014 - 12:45am

The Packers' number of third down attempts was definitely unusual. In the first 11 games, they had 121 offensive drives with 126 non-kneeldown third down attempts, which is about one third down per drive. Against the Patriots they had 9 drives and 16 third down attempts, which is almost twice as many third downs per drive, and it's that relative inefficiency on the first two downs that's depressing Green Bay's DVOA, not their conversion rate on third down.

by badger25 :: Thu, 12/04/2014 - 1:23am

Well that makes sense, but I guess what I'm saying is that I don't think that SHOULD depress the Packers' offensive DVOA to the point where it's lower than New England's, when Green Bay threw the ball with significantly greater efficiency. Rodgers averaged 9.7 yards per throw (not including sacks), which tends to make up for those unsuccessful early down plays. In other words, at least in this case, I think DVOA is wrong.

by Julio :: Thu, 12/04/2014 - 11:21am

DVOA will always be wrong for a variety of reasons, one of them being
not separating home and away performance. The entire argument about who is
better, Brady or Rodgers, is a waste of time imo. Rodgers was 10% better than
Brady at GB, Brady would have been 10% better than Rodgers at Gillette.


by Anon Ymous :: Tue, 12/02/2014 - 9:04pm

I have to agree. I thought it was remarkable that NE even had a chance to win at the end, as it seemed to me that this was a 34-17 blowout cloaked in a one score contest.

GB's defensive rating is probably the reverse of the Jets offense from a few years ago. Their poor drives were so short that they had a disproportionately low impact on DVOA as a per play metric. Perfectly reasonable, but the fact that NE left the field so quickly was a big factor in the final result of this game.

I was disappointed in NE's approach to this game. I don't buy the absurd argument that NE held things back for the eventual rematch, but I thought NE's coaching was less than impressive. On two separate occasions, I could tell immediately from the formation that they were iso-ing Ninkovich. If I could, surely NE could and both times Nink was burned for big gains. I can't help but wonder if NE would adjust - or simply call a TO - if that situation comes up again.

NE's time management at the end was also questionable. There was no reason whatsoever to think GB would struggle to score given three minutes and several time outs, so it seemed appropriate to play for the final score. NE supported this idea by going for on 4th down at midfield with plenty of time left. But targeting Gronk in the EZ, albeit with a favorable match up, went against this.

And why kick the FG? Sure, 4th and 18 is very unlikely, but how much more likely was stopping GB from getting a single first down? And then going down and scoring with less than two minutes and no TOs? Going for it also puts you in position to win on that drive instead of so many different things having to fall right. Even if you fail in your attempt, the only difference between that and a made FG is you'll need a TD on your subsequent drive instead of another figgie. Once you realize that even a 4th and 18 attempt was probably the better option, why not play for both downs on 3rd and 9 instead of taking such a deep drop when you haven't protected the QB well all game?

These are the types of questions I find myself asking after watching other games, not a Belichick coached team. Add in that this was one of McDaniels' semi-annual Vizzini (https://www.youtube.com/watch?v=U_eZmEiyTo0 Anyone willing to tell me how to link this url to "Vizzini"?) game, and I couldn't help feeling like this was a C- effort.

To this Pats fan, GB looked like a substantially better team (for 60 minutes, at least) and were certainly better at creating mismatches.

by Ezra Johnson :: Tue, 12/02/2014 - 9:05pm

On those plays with mismatches they might just hope the pass rush gets there before Rodgers can get the ball out. I think defenses are loath to call timeouts in general except in critical situations, anyway.

by PaddyPat :: Tue, 12/02/2014 - 9:14pm

I think DVOA looks at Green Bay's 2 fumbles and feels that the Patriots were unlucky in the game. I'm not sure how true that is. The Patriots' red zone offense was generally incredible in the game, and the Packers' offense was unsustainably successful on 3d down. Whatever one thinks about it, and I'm not saying that I intuitively agree, DVOA is indicating that if this exact game were to be replayed, Green Bay would regress on 3d down conversions, meaning shorter drives, the Patriots would probably recover a fumble, and it would be Green Bay playing catch-up at the end.

by PaddyPat :: Tue, 12/02/2014 - 9:16pm

I might note that this general statistical suggestion fails to take into account game plan and personnel realities, such as the specific nature of the New England pass rush in this game, which seemed far more intent on containment than pressure.

by Anon Ymous :: Tue, 12/02/2014 - 9:24pm

Nice post. I had completely forgotten about Rodgers' fumble, which I think is probably being more heavily weighted by DVOA than NE's likelihood of recovering it.

by oaktoon :: Tue, 12/02/2014 - 9:33pm

Hyde's fumble was of course not a fumble... Oh well, when the robots take over the world (As Stephen Hawking fears) this is the stuff we will be bothered by.. :)

by mehllageman56 :: Tue, 12/02/2014 - 10:44pm

The Packers looked better on Sunday, but trying to isolate Ninkovich might not work as well when Chandler Jones comes back. My problem with the Pats game plan is that they didn't pound the ball on the ground enough.

by techvet :: Wed, 12/03/2014 - 1:21am

From the other side: if Nick Perry is completely healthy, does Matthews play more in the middle?

by thebuch :: Wed, 12/03/2014 - 10:53am

Re: The field goal

The argument is that they'd have to stop GB to get the ball back. The Packers still had all three time outs and a conversion would have put them within the 10. Their best case scenario would be to score with over two minutes left, and they'd have to stop Rodgers from just getting into field goal range, which they hadn't been successful in all night. I'd argue their best bet was kicking the field goal and on side kicking it, to about the 50, so that if they could get a stop the Packers would be outside field goal range and kick it away, and they'd only need a field goal to win.

by Anon Ymous :: Wed, 12/03/2014 - 11:51am

I have to admit that I hadn't considered the onside kick, which is a serious oversight on my part. That may tilt the scales a bit in the FG direction, though I still wish they played for the final drive. Instead of targeting Gronk, for instance, Vereen was wide open for a 6-7 yard gain on 2nd and 9. Keep the clock and sticks moving, score with little time and/or burn Green Bay's timeouts.

by oaktoon :: Tue, 12/02/2014 - 7:58pm

I apologized for getting out of hand earlier so I am biting my tongue about DVOA's judgment that the Patriots were the "better" team... But 31.9% to 8.0% is a pretty wide gap, no?

So a few entirely substantive questions:

1) How can making 4 of 5 FGs but missing a 40 yarder ding the Packers more than making 0 of 1 FG but missing a 47 yarder? And if the failure to convert red zone opportunities is already counted against the Packers, isn't this double-counting?

2) Nobody turned the ball over. Rodgers "fumble" traveled about 6 inches. I could argue that a couple of Brady's tipped passes came a lot closer to disaster-- an INT-- than Rodgers dropping a ball at his feet. Hyde's muff was, according to the announcers, a ball that had already been touched by New England so his only risk was possessing the ball, then fumbling it. The "muff" was a non-play...

3) GB's defense-- which allowed 21 pts and 350 yards-- is rated much lower than NE's defense-- which allowed 26 pts and 470 yards. OK..... I'm listening. Home field effect? That I can buy, but both teams were rated 1-2 in offense going into this game....

Let's do it again in Phoenix!!

by Aaron Schatz :: Tue, 12/02/2014 - 8:06pm

All legit points... I thought I had wrote the result was surprising, but I guess somewhere in my fiddling with sentence structure, the word "surprising" came out. I certainly agree that my eyes told me the Packers played slightly better, and the DVOA result surprised me. And in Glendale, I think it's a very even pick 'em.

by Anon Ymous :: Tue, 12/02/2014 - 8:30pm

FWIW, I enjoy these DVOA oddities and the conversation they prompt, even if I may be particularly vocal in my disagreement.

by MJK :: Wed, 12/03/2014 - 12:32am

I suspect that the 3rd down arguments are right. I felt like the Patriots played better on 1st and 2nd down than the Packers, and worse on 3rd down. Even if 3rd down is weighted more heavily by DVOA (I don't think it is???), you still have more than twice as many 1st and 2nd downs than you do 3rd downs. So a team that consistently gains 4 yards, then 3 yards, then fails to convert on 3rd and 2 will look better to DVOA than a team that consistently runs for 2 yards, then takes 3 yard sack, then converts a 3rd and 11 with a 15 yard pass to their #3 WR.

by nat :: Wed, 12/03/2014 - 8:29am

Aaron, is this a case of VOA giving the trailing team easier baselines?

But here is another idea.

Total DVOA gives equal weight to offense and defense. But this game didn't happen that way. New England's defense was on the field for 25% more plays than it's offense. And the defense was not doing very well, allowing the Packers to move the chains 80% of the time. Your eyes give weight to O and D based on the play counts they saw. Total DVOA imagines an average future game.

VOA's idea of a 8-12% win for the Patriots is equivalent to a real squeaker, maybe a field goal advantage or less. Re-weight it for the actual play counts, and it will agree with your eyeballs. Or imagine the Packers magically deprived of 14 plays. The result might agree with VOA then.

by oaktoon :: Wed, 12/03/2014 - 10:59am

OK, Bear with me folks-- this is done in the belief that a) something is haywire in DVOA and b) maybe this will help Aaron fix it when he has time in the off-season.

I think any fair reading of the second half of the NE-GB game would say the teams were very even, with a slight edge going to the Pats because They scored a TD and the Packers didn't (thank you, Devante Adams).

NE gained 141 yds; GB 129 (before the final kneel downs)

Both teams had two non-scoring drives to opn the half (Crosby missed a FG in one), but GB's were better. Pats went three and out, gaining 7 yards; then got one first down and had to punt, gaining 13 yds. Packers gained 32 yards and missed a FG, converting one 3rd and 3 and another 2nd and 10; then gained 25 yards and had to punt, making a 1st down on a 1st and 10, converting a 3rd and 7, and, yes, seeing Rodgers get sacked where he dropped the ball at his feet.

If the "fumble" is enough to shift the scales in NE's direction, OK-- but should it? I'd argue that the Packers making 4 FDs to NE's 1 and actually having a makeable FG gives them a slight nod.

Then Brady went 78 yards for a score-- very efficient, converted a 3rd and 1, and another 2nd and 6-- the TD was a 2nd and 10 pass from the red zone. Rodgers then drove 65 yards, just as efficiently-- converted one 3rd and 6, one 2nd and 5, the TD was a 2nd and 5 pass from the red zone (OOPS--- Devante Adams dropped it!!)

OK-- obvious NE advantage-- but basically it was the drop. Other than that it was as even as one could get.

Brady then goes 43 yards in a less efficient drive-- had to convert a 4th and 3 and a 3rd and 7, before the missed hook-up with Gronk, the 3rd down sack (which was a huge play in that it forced Belichick to give up on a TD) and the missed FG.

Packers then went to 4 minute offense-- run, run 3rd down conversion to Cobb. Game Over

SO because of the Adams drop I'd give the Patriots a slight edge in points (7-3), yards (141-129), and overall production. BUT-- basically in the context of the game and the outcome, consider that both teams had two productive possessions and two non-productive possessions since Rodgers did what he needed to do to salt the game away and we'll never know what would have happened had Gronkowski caught that ball or Belichick had gone for it and converted on 4th and 18. I think the final GB possession ought to carry a lot more weight (or any team in a similar end-of-game situation) than I am guessing DVOA gives it.

Stay tuned for the first half, because I think it is absolutely definitive

by oaktoon :: Wed, 12/03/2014 - 11:23am

The first half, by any fair reading of the facts, was not close.

The Packers had an extra possession-- and we all know what they did with it.

GB's first possession-- 58 yards. Opened with 3 consecutive plays that got first downs, then a 3rd and 8 Def. Holding conversion, and a 3 and out with a FG.

NE's first possession-- one 1st down, and a failed 3rd and 2. 58 yards to 20.

GB's second possession-- 66 yards. OK-- here may be where DVOA rears its head-- only once did the Packers have a productive 1st down play--it got a 1st down; but they converted a 2nd and 8 with a 33 yd pass, Rodgers scrambled for 17 on a 2nd and 18 and they then converted-- and then they went 3 and out in the red zone. FG

NE's second possession-- was a 3 and out that got them 9 yards.

we're at 124 yds to 29. Two productive possessions, and two failed possessions.

GB's third possession-- was an 85 yard scoring drive in a heartbeat. 2 yd run, 6 yd pass, 3rd and 2 conversion with a 45 yd pass (adams), 32 yard TD pass (Rodgers said he wanted to throw to Nelson-- pulled ball down at last instant and then found the other Rodgers). I think DVOA-- or any of us-- really likes that drive.

NE's third possession-- ditto. 3 yd run 29 yd pass, 23 yd pass, 12 run, 6 run-- 73 yds TD.

OK we're at 13-7. 209 yds to 102. 3 straight productive possessions vs. 2 failed/1 productive.

GB's 4th possession-- 57 yards and a failure in red zone. OK-- Aaron Rodgers, who we all know is pretty darn good at this QB thing but DVOA can't recognize from the man in the moon, converted 3 straight 3rd down plays. a 3rd and 5 pass for 33 yds, a 3rd and 2 pass for 8 yards, a 3rd and 12 pass to Cobb for... 12 yards. And then they stalled and kicked a FG.

NE's 4th possession. 80 yards and another TD-- Brady converted 2nd and 10, 2nd and 10 again, 2nd and 5, then 3rd and 3, and finally they ran it in on a 3rd and goal from the 2.

it's now 16-14. Slight edge for GB in score. In yards it's 266 to 182. In productive possessions it's 4 to 2. Brady failed to convert 2 3rd downs, and succeeded twice. Rodgers failed to convert 2 3rd downs and succeeded five times. I don't think NE was the better team at this point-- I could listen to an argument that said it was close-- but one team had basically pushed the other all over the field, while the other had failed twice, but then recovered with two very successful possessions. If you are telling me that Micah Hyde's "muff"-- which occurred after NE's first possession and was basically a free play-- was sufficient to tip the scales hugely in NE's favor, something is wrong. If you are telling me that Rodgers going 5-7 on 3rd down while Brady was 2-4 was sufficient to tip the scales hugely in NE's favor, then something is wrong.

Because the close game to this point became a little less than that when Rodgers went 81 yards in 5 plays to make the score 23-14, the yards 347-182, and the possession count 5 productive for GB to 2 for NE.

The 2nd half was close, and maybe pipped by the Patriots due mainly to Adams' drop. He's a Packer-- his fault-- should count against his team. The first half wasn't close.

I see nothing in this data to ever conclude that when the #2 DVOA team played the #3 DVOA team-- the #2 team was better to the tune of 32 to 8. It simply defies logic-- UNLESS-- there was a home-field blowout quotient built in, and the mere fact that the Pats played the Packers close in Lambeau whereas nobody else had since the 1st half vs the Jets tipped the scales this strongly.

by Pat :: Wed, 12/03/2014 - 11:25am

I see nothing in this data to ever conclude that when the #2 DVOA team played the #3 DVOA team-- the #2 team was better to the tune of 32 to 8.

That's because you're looking at the game with opponent adjustments. "Your eyes" or "the facts" don't see opponent adjustments (since you're not looking at the other games) - so your eyes see VOA. Without opponent adjustments, the game was NE 34.5%, GB 27.0%: or converted to rough points, NE +1.5 points.

Why the opponent adjustments make such a big effect for GB versus NE, I'm not sure. But that could be Green Bay targeting a weakness in New England much more than vice versa.

by oaktoon :: Wed, 12/03/2014 - 11:34am

I'm smiling now... NE has been the highest-rated team by DVOA for most of the season... SO by what "opponent adjustment" did they benefit by scoring fewer pts and gaining fewer yards than the weaker team they just played?

It can only be a Lambeau effect-- DVOA punished the Packers (and rewarded the Patriots) for blowing out the Eagles and Bears.

And as for the non-opponent adjustment conclusion-- again-- read the pretty precise summary of the game I just made and tell me with a straight face the Patriots were the better team....

by Anon Ymous :: Wed, 12/03/2014 - 11:54am

NE hasn't been the highest rated team by DVOA a single week this year, not sure where you are getting that from. If anything, their rating was suppressed until their dominant performances accumulated enough to dwarf their terrible start.

The odd gap in DVOA for the Green Bay game has been explained several times on this page. I may not agree with it - hell, even Aaron admited he expected a different result - but it is perfectly understandable, so why are you still going on about this?

by oaktoon :: Wed, 12/03/2014 - 11:58am

OK-- I stand corrected-- Aaron said NE has been the clear best team over the past 6-7 weeks.... Point still stands-- they were higher-rated than the Packers coming into this game.

As to the "oddness" I am trying to shed some further light on it, because the explanations simply don't add up. And hoping it will help Aaron when he tweaks the machinery-- a Micah Hyde "muff" or Aaron Rodgers dropping the ball at his feet or being more proficient on 3rd down should not, can not, should never produce a wide margin for the Patriots given the actual story of this game.

So I told the story as a way to help. I think it's illuminating. It was, after all, to this point, the biggest game of the year. One team outplayed the other-- and DVOA got it wrong.


by NYMike :: Wed, 12/03/2014 - 12:04pm

Oaktoon: Please check my post 103 to see why DVOA got this "wrong." BTW, I'm a Packer owner. Sometimes, we don't remember things quite the way they happened because the end result was so positive. But in this game, at least in the first half, GB stunk on first down.

The claim DVOA is making is that a team playing so often behind the chains, or behind schedule if you prefer, will not be successful in the long run.

by oaktoon :: Wed, 12/03/2014 - 12:35pm

See my corrections that added 3 successful first half 1st down plays you left out, and traded one Nelson 8 yard pass which was on 1st down that you left out for another-- the TD-- which was actually on 3rd and 2.

In the long run, the Packers under Rodgers have been either the best, or one of the 3 best, offense(s) in the game for the past 7 years. Sometimes they are efficient on 1st down, and sometimes they're not-- and a lot of times it doesn't correlate to their overall performance, whatever DVOA thinks.

Again, while there may be outlier aspects to how DVOA rates this game, in no possible scenario should it conclude NE was the better team by the margin it did.

by Anon Ymous :: Wed, 12/03/2014 - 12:20pm

DVOA really dislikes turnovers and it considers fumble recovery luck, so that is going to have a much bigger impact on they system that it does for you and I. And I'm OK with that. There are a ton of turnover items that could be added in if Aaron had the time or the interest in getting subjective enough. Is a pick after a WR tips the ball predictive of future interceptions? How about a dropped pick by a defender? A fumble that quickly rolls out of bounds may not have impacted the game, but it clearly was a bad play, right?

As for the game itself, DVOA straddling the line between descriptive and predictive will create oddities like this from time to time. Even on NE's non-scoring drives, they had a lot of successful plays. On GB's drives they had a lot of unsuccessful plays, particularly on first down. Historically, that is not a formula for winning, but GB was so dominant on 3rd down that it didn't matter.

If NE and Green Bay meet again, and GB success spreads more evenly - more good first downs, but fewer 3rd down conversions - with NE winning 30-24, will DVOA still have gotten it wrong?

Again, even Aaron admits that this probably isn't an entirely accurate assessment of how that individual game went, but that isn't all DVOA is trying to do. I'm all for debate and critique, but I'd encourage you to strengthen your understanding of DVOA beforehand... as well as adding "constructive" to your criticism.

by oaktoon :: Wed, 12/03/2014 - 12:38pm

I understand it as well as possible for those of us who didn't write the program. I'm trying to improve it. I still haven't gotten an explanation that can justify the distorted results other than a huge home field allowance-- and it that's what it was, then I can live with it, I guess-- except then maybe GB should have climbed higher in the previous weeks.

by Pat :: Wed, 12/03/2014 - 1:12pm

Opponent adjustments aren't "global" - they don't just look at the overall team and say "oh, you played awesome against a good team so you get a huge boost." They're conditional - so you can actually get a bigger effect than you might expect depending on how the game goes (runs vs. passes, plays on offense, etc.).

This might seem weird, because some of those are choices of the coach - whether to run or to pass, obviously, but more subtly even the number of plays on offense/defense are also in some sense "coach's choice." So that means that Green Bay could get a penalty because they targeted a weakness of New England, only did a little above average based on New England's baseline for similar plays. Of course, that targeting won them the game - so it's not like you're saying that Green Bay was bad.

This isn't a 'shortcoming' of DVOA, it's an 'interesting feature' - if there are teams that are consistently better at tactically targeting opponent's weaknesses, they'll win more games than you'd expect from DVOA. Great for them. I don't think any teams/coaches have actually shown that they have that ability (over everyone else) and it's sustainable, so that ends up being part of the "randomness" that makes it a game.

And as for the non-opponent adjustment conclusion-- again-- read the pretty precise summary of the game I just made and tell me with a straight face the Patriots were the better team....

The only place where "better" exists is in the convoluted mind of fans. I watched the game, I've actually seen it again on replay, and yeah, I have no problem with the idea that on a per play basis, New England did very slightly better.

And honestly, the 'very slightly better' here is probably entirely due to the muffed punt being considered a muffed punt, rather than an illegal touch penalty, although I don't know that the illegal touch penalty was actually *called*. Take that away and I'd bet GB comes out with a higher VOA.

I think Green Bay had a much better game, tactically, which is what led them to a win. Heck, I think New England attempting the field goal at the end of the game was a mistake. What are we saying today if New England goes for it there, and gets the touchdown? Are all of our arguments suddenly reversed?

by oaktoon :: Wed, 12/03/2014 - 1:24pm

On a "per play" basis, NE did better?

Huh? Packers a) gained more yards per play; b) combined their plays into 9 possessions which produced 475 yds, 2TDS, 4 FGs, 1 Missed FG, 1 Punt, and I game-killing first down-- as opposed to 8 NE possessions which produced 323 yds, 3 TDS, 1 Missed FG, and 4 punts;

I am not trying to be obtuse-- if a muffed punt (which wasn't) and a QB drop of the ball at his feet and a missed 40 yd FG (when the kicker in question also made 4) vs a missed 47 yard FG (when the kicker in question made no others) and a few 3rd and 2 conversions-- if all that not only outweighs all the other demonstrated advantages GB had-- but outweighs it to the tune of 32 to 8? I mean, really?? That's all I'm saying. Aaron will fix it in the off-season, and I don't believe this big of a distortion will ever occur again...

by Pat :: Thu, 12/04/2014 - 11:57am

but outweighs it to the tune of 32 to 8? I mean, really

It's not 32 to 8! Why do you keep saying that? That's DVOA. The game does not 'see' DVOA. The game sees VOA. And the VOA difference was 34.5 to 27. DVOA corrects for the opponent. The game doesn't. If you want to argue 34.5 to 27 is wrong, that's fine. Arguing 32 to 8 is wrong is just misunderstanding an opponent-adjusted stat.

Correcting for the opponent is like handicapping golf. You know Green Bay's offense is way better than New England's defense, so you "spot" them a bunch of efficiency. Football is not handicapped.

DVOA never lines up with the game results unless the teams are near 0%.

And yes, most of that difference is in special teams, around 4.7%, or almost 60% of the difference. Missing a 40 yard FG is bad - those are typically ~80% made. 47 yard field goals are more like 65% made. And you can keep saying that the muff wasn't a muff, but the officials didn't call the illegal touch, and it's not in the play by play that way. That's life.

by Will Allen :: Wed, 12/03/2014 - 1:26pm

Your last sentence gets to the heart of the matter. It amazes me that people expect 130 plays in a football game to produce a metric that doesn't have oddities.

by thebuch :: Wed, 12/03/2014 - 1:37pm

If NE gets a TD, Rodgers has over two minutes and three time outs to get into field goal range, something he was successful at all day. Also, he'd have four downs to do it instead of 3.

4th and 18 in that situation, kick the field goal and go for the onside to about the 50 yard line. If you stop them, you'd still get the ball back, but that's the most likely way of keeping the ball out of Aaron Rodgers hands, and as a Packers fan, my biggest concern.

by Pat :: Thu, 12/04/2014 - 12:08pm

It wasn't 4th and goal, it was 4th and 18, so the clock isn't something you can predict, if they would get the first down.

but that's the most likely way of keeping the ball out of Aaron Rodgers hands, and as a Packers fan, my biggest concern.

No way. Go for it. You want to keep it out of Green Bay's hands? Kicking the FG is about ~65%. Onsides kick is about 15%. So you'll keep it out of his hands 9.75% of the time.

You think if you gave the Patriots offense 10 tries on 4th and 18 they wouldn't convert once?

by oaktoon :: Wed, 12/03/2014 - 11:30am

One more thing that I wonder if DVOA measures. We all understand the logic behind the "defer" decision at the coin toss--- the possessions will even out anyway, and the kicking team gets to lead off the 2nd half (and sometimes also gets the final possession of the 1st half and thus gets to "double-up"-- Rodgers talks about this a lot).

But GB did not defer-- they took the ball. And because they scored at the death of the first half with the Nelson catch and run, they basically got an extra possession in this game-- since we have to count that final game-clinching possession though it only gained 13 yards. GB had 9 possessions to NE's 8. That fact alone is a pretty damn important deal-- with offenses this good, who probably average close over 3 pts per possession if not more, that's a big edge.

by Pat :: Wed, 12/03/2014 - 1:18pm

No, it doesn't measure it, and honestly you wouldn't want it to. That's tactics (clock management/play selection/etc.) - and is more related to winning games than outperforming the opponent as a team.

You would probably want a second metric for that, and it'd be a lot harder to come up with. My guess is that it wouldn't be predictive, mainly retrodictive (explanatory). So it would tell you why a team won the game (X outcoached Y that game) but because coach's decisions really only decide games a small fraction of the time, you wouldn't have enough measurement points to get anything useful. That's just my guess, though.

(I mean, here, we're talking about Mike McCarthy outcoaching Bill Belichick. That is not the order I would naively expect to have those two.)

It's like fumble recovery: getting fumbles better than ~half the time is super important to winning the game, but no one appears to be able to do it, sustainably, better than average.

So if you had a coach who was super-awesome at clock management and could consistently get an extra possession, yeah, of course you'd want to measure it. But I don't think a coach like that exists. So in the end, if you get one, that's just part of what makes it a game.

by oaktoon :: Wed, 12/03/2014 - 1:29pm

Don't get me wrong-- I'm not saying it is a repeatable skill-- but does everything DVOA measures have to be? In the cases where it occurs it is very advantageous.

As to your McCarthy shot, at many moments we Packer fans agree with you!! and BB is what he is-- one of the greatest coaches ever. Yet as a play-caller and offensive strategist McCarthy is not only one of the best now-- but ascending into the all-time ranks-- that much is becoming more clear with each season. Sure he has a great QB-- but so does Payton, so did Walsh, so does Belichick.

by Pat :: Thu, 12/04/2014 - 12:17pm

Don't get me wrong-- I'm not saying it is a repeatable skill-- but does everything DVOA measures have to be?

That's... basically the entire idea behind a predictive metric. DVOA's not entirely predictive, but trying to do a completely predictive metric for football would be impossible. Teams can change radically from week to week just because of injury.

so did Walsh, so does Belichick.

Belichick isn't an offensive guy - he's a defensive guy. And yes, I'd take Belichick's defensive strategy/tactics over McCarthy's offensive strategy/tactics right now, and I think most NFL fans would as well. Maybe not Packers fans. In that game? Nope. But that happens. That's the point.

by tuluse :: Wed, 12/03/2014 - 1:38pm

That last paragraph makes me think. In baseball studies have concluded that while there is no such thing as clutch player (player who get better in leverage/pressure situations) there may be such a thing as a choking player.

Time management in football seems like that to me. No coach is significantly better than average, but some are significantly worse than average. Which would explain the Andy Reid Eagles often doing worse than DVOA would expect them to.

by bmay :: Wed, 12/03/2014 - 5:27pm

"In baseball studies have concluded that while there is no such thing as clutch player (player who get better in leverage/pressure situations) there may be such a thing as a choking player."

I would like a source for that claim because I have yet to see research that supports it.

by justanothersteve :: Wed, 12/03/2014 - 12:06pm

You're not even bothering to understand how DVOA works. Each play has its own value. Quick example. Play sequence #1 of first down for 5 yards, second down for 4 yards, and third down no gain has two successful plays and one unsuccessful play. Play sequence #2 of first down for 2 yards, second down a sack for -5 yards, and third down a pass for 13 yards and a first down. Play sequence #1 has two successful plays and one unsuccessful play. Play sequence #2 has two unsuccessful play and one successful play. I don't know what the calculated values would be, but each play is considered on its own without regard to how the drive ended. Play sequence #1 probably ends up with a slightly higher value than #2, even though #2 was a successful series of downs over #1. While this may be an issue on an individual comparison, the idea is that over the long run these success rates even out. This is why some have pointed out on discussions of the Packers-Pats game that GB's third down success rate was unsustainable as it was better than the average for the Packers. (Though I'm fairly certain Rodgers third down success rate is among the best in the league.)

Occasionally, a team may lose even though DVOA says they had a better game. It's similar to a baseball game where one team outhits the other by a ratio of 8:2 but gets outscored 2-1 because the two hits were a single followed by a homer. Normally, you win if you outhit the other team by that margin, but sometimes you just don't. If you regularly got outhit by a 4:1 margin, you will occasionally win but you will more likely lose. DVOA tries to bring that logic to football. It doesn't necessarily play out that way on the field, but it works out that way in the long run.

Click on the links in the About section for more info. Especially http://www.footballoutsiders.com/info/methods

by oaktoon :: Wed, 12/03/2014 - 12:46pm

I can understand the circling of the wagons that is going on in response to this native American. I truly can. I've been in that position myself. I'll repeat myself-- I get all the little pieces that could have thrown DVOA off-- I don't need to be condescended to, I'm not stupid. But the actual story of this game-- as I told to be very constructive a) doesn't add up; and b) tells me DVOA has some pretty serious flaws that should be fixed.

GB moved the ball the entire first half-- they scored all 5 times, including a TD drive that was pretty spotless and then a lightning quick drive at the end... NE moved the ball twice-- two efficient TD drives, and failed utterly the other two times... In no universe was this a close result, whether Rodgers converted 3rd down plays or not-- 3 of them were 3rd and 2, by the way and a 4th 3rd and 5-- which are hardly "unsustainable"... The second half was clearly a closer result, with the edge going to NE.

So fairly wide margin GB first half vs. close margin NE second half= 32-8 NE conclusion???

We'll just call it the Micah Hyde "muff" game....

by Anon Ymous :: Wed, 12/03/2014 - 12:59pm

There is no "circling of wagons" and your continued comments demonstrate that you don't understand DVOA as much as you claim to. I'm sorry if you find that belittling, but it's the truth.

As for critiquing in an effort to improve the metric, thus far all I've seen is complaining. Other than eliminating one of the fumbles, what specific recommendations do you have to make DVOA more accurate? Should 3rd down be weighted more heavily? Should all turnovers get a subjective once over?

If you really want to help, that's the kind of post that needs to be made. You've stated your case that DVOA wasn't perfectly descriptive of that game - an unnecessary case that not a single person disagrees with - now what does Aaron do from here?

by oaktoon :: Wed, 12/03/2014 - 1:54pm

OK, here goes.

1. A "muffed" punt return on a free play not only shouldn't be weighed as a negative, but probably ought to be viewed as a positive if it gained yards for the return team;

2. A "fumble" where the QB drops the ball at his feet while falling should not be judged the same as a strip sack or a QB getting whacked from the blind side and seeing the ball fly 10-20 feet in some direction. Particularly since you can argue that is part skill, as Rodgers accepts sacks but is more likely (not always, of course) to protect the ball better than your average Bear (or Cutler...)

3. A team that misses a 40 yd FG but whose kicker makes 4 others should not be docked compared to a team that misses a 47 yd FG but attempts no others. (I assume this happened because Aaron said it did in his original comment about DVOA and this game)

4. A team that converts several 3rd and short situations (they were mainly 3rd and 2) should not be viewed/judged as performing an "unsustainable" result, not in this day and age and partic. when it involves one of the two or three best offenses in the league-- and particularly in contrast to another team that couldn't even gain the first downs to produce 3rd down situations for much of this game

And finally, from a strict common-sense standpoint, since I am not in control of the DVOA joystick (and I don't view what I did in telling the story of this game as a "complaint"-- but in fact as an illustration designed to help the powers that be fix the program), a team that clearly outplays its opponent-- and by a wide margin-- in one half, and performs close to that opponent in the other half-- should never be judged to have been the clearly inferior team-- by a wide margin-- not no way, not no how. And if and when the program is fixed, I don't expect we'll ever see this great a distortion in DVOA's judgment again...

by LyleNM :: Wed, 12/03/2014 - 2:08pm

Anybody else out there OK with the FOMBC being invoked?

by Pat :: Thu, 12/04/2014 - 12:24pm

... Maybe? It's not like Green Bay's being seriously slighted. They're 3rd, for crying out loud, and basically even with New England. In a matchup on a neutral field, you'd almost expect a coin flip.

If it were me, though, and Green Bay - New England played in the Super Bowl in two weeks from now, I'd probably bet on the Patriots, so if the FOMBC expectation would be NE over GB, then yeah, it works.

by cjfarls :: Wed, 12/03/2014 - 2:18pm

Many these things can not be determined by the PBP data... if not, they are useless suggestions. Aaron has no ability to subjectively make those adjustments to the PBP reports for every game.

Things like #3 and #4: Again, you show you have no understanding of how DVOA works. Plays are independent events, adjusted by the opposing team... what a team does repeatedly or in the past has NO VALUE in judging the VOA impact of the subsequent play. That VOA value is then adjusted by the OPPOSING TEAM, not by what the performing team has done in the past.

by Anon Ymous :: Wed, 12/03/2014 - 2:41pm

I agree that the fumble that wasn't should be removed, but I don't about the other. It was clearly a drop and we see fumbles all the time come out at the last second and end up in the opponent's hands. That is asking for too much subjectivity IMO.

As for the 3rd downs, I just went through and here are the yards to go on every third down

8, 8, 1, 9, 2, 5, 12, 6, 2, 3, 17, 7, 21, 6, 5, 4

Only 4 times out of 16 was Green Bay in a 3rd and 3 or less. 10 times out of 16 it was 3rd and 6+ to go. That's not quite the on-target offense you portray in your comments, it means Green Bay - despite being so efficient overall - was not nearly as efficient on a per play level. As good as they are, Green Bay is "only" converting 47% of their 3rd downs for the season, meaning they had two additional scoring drives over their average on the year, let alone the NFL average.

As I've said several times already, the issue is DVOA trying to be both descriptive and predictive. It is simply pointing out that 63% success on 3rd down (with 50% on 3rd and long) isn't sustainable, particularly when the offense had so many unsuccessful plays on first and second down.

Personally, I find that more interesting than something that simply tells me what I already knew, though that doesn't stop me from voicing disagreement every now and again myself. ;-)

by Perfundle :: Wed, 12/03/2014 - 2:56pm

3. A team that misses a 40 yd FG but whose kicker makes 4 others should not be docked compared to a team that misses a 47 yd FG but attempts no others.

A team that makes four field goals the longest of which was 35 yards should hardly see their rating move up; everybody should be expected to make those field goals.

by justanothersteve :: Wed, 12/03/2014 - 5:15pm

1. I agree on this. I'm not sure most players knew about the no penalty for returning after a kicking team touch rule until this year and this may be the first instance where it was scored a muff, so I imagine a change to the formula to account for this should not cause much grief.

2. While in theory I agree, I think this would become too subjective and I'd leave it as it is. I've seen enough football to know that even when you'd expect the guy falling on the ball to easily recover it, many times it just doesn't happen.

3. The made FGs were all fairly short and don't improve DVOA much. A missed 40 yard FG should hurt more than a missed 47 yard FG. I also suspect the muff and the out-of-bounds kickoff were bigger docks on the value.

4. They weren't mainly 3rd and 2. There were conversions on third and 8, 5, 12, 7, and 6 yards which meant one or two unsuccessful plays before this.

Rodgers was also sacked three times vs one for Brady. DVOA penalizes sacks. The formula also doesn't like big plays as much as people do. You even commented on how Green Bay's offense is geared towards explosive plays. (I've also heard this before from McCarthy, so I agree with you here.) DVOA is not the be all/end all of formulas. It's a very good attempt to bring baseball sabermetrics to football and it's better to tweak any formula than radically change it just because of one game. I don't always agree with it either. Relax and accept it for what it is.

by PatsFan :: Wed, 12/03/2014 - 5:49pm

Don't you mean R-E-L-A-X?

by Pat :: Thu, 12/04/2014 - 12:50pm

It's actually been fixed in the play by play now - it's just listed as being downed by NE, which to be honest I don't get - that's not what happened, since I think NE touched it closer to the goal line.

It's just a weird situation, period - I'm not sure the illegal touch ever gets called unless it matters, so it's not even recorded anywhere.

by Travis :: Thu, 12/04/2014 - 4:51pm

It was originally scored as a 41-yard punt fumbled at the GB13; it was corrected to a 37-yard punt downed at the GB17. The video here shows this to be an accurate (if incomplete) description of what happened.

The closest thing that covers this situation in the scoring manual mentions not charging fumbles that occur beyond the spot of penalty enforcement; if you view NE downing the ball as an illegal touching penalty, the non-fumble scoring makes sense.

by Pat :: Fri, 12/05/2014 - 10:33am

Oops, I didn't realize that GB's recovery was further downfield than New England.

Downing the ball is an illegal touch penalty, so it makes sense. Honestly, the "illegal touch is a penalty but we're not going to waste time calling it except in the rare case where it becomes important, like offsetting penalties" bit is probably one of the more confusing things in football, so it's not surprising that it's goofy.

I'm still waiting for the insane play sequence where a punt is touched by the punting team, recovered by the return team, returned nearly for a touchdown before it's stripped, recovered by the original punting team, then someone from the return team horse-collar tackles him down. Offsetting penalties mean a re-kick, and then on the re-kick the punting team fakes the punt and scores a touchdown.

It would make me laugh watching announcers' heads explode trying to explain it.

by DisplacedPackerFan :: Tue, 12/02/2014 - 8:01pm

I get that the official play by play calls it a muff by Hyde, but a NE player had touched the ball so it was a free play when he went after it. That should not be a negative against GB, it's an NFL play by play issue.

by nat :: Wed, 12/03/2014 - 10:30am

The ball was spotted where it went out of bounds, so the illegal touch penalty wasn't applied. DVOA ignores unaccepted penalties, as it should.

And sure, muffing the ball was harmless, even helpful, in this case. But unless it was intentional (which would be a foul for "batting") it can only be predictive of future ball security problems.

So it's not a play by play issue. Nor is it a DVOA problem, really. It's just a smart and safe play, poorly executed, that worked out well. DVOA sees the poorly executed part, but not the safe part.

by Flounder :: Wed, 12/03/2014 - 11:38am

But it was a "safe" play because the execution was irrelevant. It was "safe" because it could not result in a turnover.

However, it wasn't a "safe" play in the sense that a "muff" was likely given that Hyde was attempting to field a bouncing ball on the run, a difficult play to make that, in a "non-safe" situation, Hyde would not have attempted in the first place.

So the "poorly executed" part is not predictive of ball security, and hence not useful and counter-productive for DVOA, unless you assume that Hyde would attempt to make such a play a "non-safe" situation, which, to me, is not a reasonable assumption.

It's a very uncommon scenario that neither the play by play nor DVOA are dealing with properly.

by turbohappy :: Wed, 12/03/2014 - 3:13pm

I agree. Think of a more common scenario. Someone is offsides, quarterback sees the flag and takes a chance deep. You either get the completion or the penalty...not going to be a negative against your offensive DVOA either way.

by Ezra Johnson :: Tue, 12/02/2014 - 8:19pm

32 to 8, seriously? That seems like a lot more than "little differences" adding up. Does DVO correlate with points 1:1, such that a 24-6 NE victory could have resulted? I hope not.

by jebmak :: Tue, 12/02/2014 - 8:42pm

Nope. It is how much better the team played than an average team. NE was 32% better than an average team would have been vs GB, GB 8% better vs NE.

Here it is in much more detail, if you are interested:

by Anon Ymous :: Tue, 12/02/2014 - 8:53pm

True, but NE has a similar edge in VOA as well, so it thinks NE was a good deal better even without opponent adjustments.

by chemical burn :: Tue, 12/02/2014 - 8:57pm

It's so strange. I watched the game and felt like NE was very, very lucky to be in it - and I was rooting for GB to lose. At the end, it felt like "wow, NE is somehow going to pull this out" more than the score "correcting" to catch up with what was happening on the field. I'm sorta shocked by these numbers.

by Anon Ymous :: Tue, 12/02/2014 - 9:12pm

Yeah, I have to assume it is the Jets' 3-and-out effect for NE's offense and that GB wasn't quite as efficient on first and second down as they were on 3rd, accumulating a disproportionate amount of low value plays.

IMO, it is one of those times when DVOA straddling the line between being descriptive and predictive creates an odd result.

by Pat :: Thu, 12/04/2014 - 1:07pm

Huh? NE has a minor edge in VOA, not a similar one - 24% vs. 8%, and most of that from special teams (~5%), which obviously changes when the muff gets removed. So it puts them basically identical.

Which sounds about right to me. I mean, Green Bay basically won the entire game due to that drive at the end of the first half, and the fact that they had enough time there to do that was entirely preventable by New England. So bully for winning the game, but relying on beating the Patriots with mostly field goals is probably not a great strategy.

by Anon Ymous :: Fri, 12/05/2014 - 12:20pm

You might want to read the chart again, Pat. NE has a nearly 20% edge over GB in VOA (-12% to 7.9%)

by oaktoon :: Tue, 12/02/2014 - 8:57pm

OK-- the average team held New England to 380 yards and 23 pts? Sure... maybe at halftime. :) The average team gained 430 yards and scored 22 pts vs NE? The latter is possible-- in the context of getting blown out by NE and a lot of garbage time production....

The only conceivable explanation that makes sense is that the Lambeau blowouts vs. Philly, Chicago, Carolina and Minnesota (and the 2nd half vs the Jets) set DVOA to penalize the Packers (and reward the Patriots) for anything short of a blowout in this game.. We can argue until the cows come home whether or not that is anything close to an appropriate metric--- but i can't imagine any other possible rationale for the distortion of this conclusion...

This is why the "eye test" for something like TCU vs. Baylor (and I believe the committee is doing the absolutely right thing by separating them and thus rendering the head-to-head result meaningless. TCU is obviously a better team) will still matter. If the damn metrics go against our "lying eyes", then what do we do other than throw up our hands.. I can absolutely understand if DVOA said NE was slightly better than GB in this game-- but this gap is simply befuddling...

by chemical burn :: Tue, 12/02/2014 - 9:04pm

FO, more than a lot of advanced stats, despises turnovers. I knew it wouldn't like McCoy's rushing day versus the Cowboys because of his fumbles and it sounds like the muff and Rodgers fumble, which didn't set off alarms in anyone watching the game (because they truly didn't matter) made DVOA say, "GB was mere luck away from two turnovers, which are the worst thing that can happen."

by Perfundle :: Tue, 12/02/2014 - 9:42pm

I didn't watch the entire game so I missed his fumble, but did Rodgers simply drop the ball by himself and then cover it up?

Assuming that's what happened, then the issue is that such plays, as well as botched handoffs between the QB and the running back, generally get recovered by the offense, and are quite different from strip sacks which probably get recovered by the defense the majority of the time, but DVOA treats them the same, because they all go down as strip sacks on the play-by-play.

by oaktoon :: Tue, 12/02/2014 - 9:45pm

exactly what happened... Artificial intelligence-- beware!!!

by tuluse :: Wed, 12/03/2014 - 11:53am

It's bad to drop the ball. DVOA can't know how close defenders were to Rodgers when he did so. I doubt Rodgers himself felt, well this is a good time to drop the ball on the turf since I can recover it easily.

by Perfundle :: Wed, 12/03/2014 - 2:13pm

Certainly true, but one difference is that the defense gets credit for forcing a fumble, when in reality they did nothing of the sort.

by JoeyHarringtonsPiano :: Wed, 12/03/2014 - 2:17pm

I recall Aaron saying once that for fumbled snaps the offense gets dinged for fumbling, but the defense doesn't get credit for forcing a fumble. Correct me if I'm wrong, but this may be a similar situation.

by Perfundle :: Wed, 12/03/2014 - 2:31pm

I imagine that is the case, However, Rodgers didn't go down as having a fumbled snap, but as having been strip sacked.

(3:52) (Shotgun) 12-A.Rodgers sacked at GB 41 for -6 yards (91-J.Collins). FUMBLES (91-J.Collins), and recovers at GB 41. 12-A.Rodgers to GB 41 for no gain (54-D.Hightower).

by Will Allen :: Wed, 12/03/2014 - 2:21pm

Over a couple thousand plays, that stuff gets put into the right perspective. Over 130, not so much. I really thinks people's expectations for metrics in a single game get out of whack.

by Led :: Wed, 12/03/2014 - 3:26pm

Absolutely. The FO metrics are a tool that is most useful for analyzing teams over the long term. For any given game, however, the VOA (or DVOA) results are less informative and should be supplemented liberally with intelligent analysis based on watching the game itself. As with any tool, you have to know how to use it. A hammer is extremely useful but it makes a sh***y screwdriver.

by young curmudgeon :: Wed, 12/03/2014 - 12:41am

"rendering the head to head result meaningless"? Is the point of the game to win or to look good? "TCU is obviously a better team"? Football is a more enjoyable sport than, say, figure skating in large part because there is an objective, quantifiable result at the end of the contest. Subjective, 'judged' sports are simply not as interesting. The optimal way to decide who is "the better team" is to play the game. Baylor and TCU played. Baylor won. If they both end up with identical records, Baylor is the conference champion and should be the team under consideration for the playoff.

Yes, yes, I know that games have fluky outcomes, that the "better team" doesn't always win, that plenty of subjectivity is involved in the final result (interference calls, etc.) No, I didn't see the Baylor-TCU game, so I don't know if there were weird circumstances that allowed Baylor to win. No, I don't have a rooting interest at all. My point is simply that the result of the game is the most important thing about the game, if you are judging success and failure. New England was unquestionably a better team than the Giants a few years ago...ask a Patriots fan if that therefore counts as a Super Bowl victory.

by oaktoon :: Wed, 12/03/2014 - 9:31am

well, not to turn this into a CFP forum, the counter-argument involves a Minnesota win (somewhere between 25-30 team in the country) vs a Buffalo win (109th in latest computer composite--Baylor played worst OOC schedule in country), a Win over West Virginia on the road compared to a loss vs West Virginia, and better results for TCU in 4 other common opponent games, compared to 2 for Baylor. And in the last chance to see the two teams play, Team A beat a team by 55 pts Less than Team B had done earlier in the year, and Team B beat a team by 17 pts more than Team A had done earlier in the year. All of that vs. a head to head game in Waco? I think the committee is on very firm ground--remember FSU-Notre Dame in the 1990s--- both unbeaten, ND wins close in South Bend, loses to Boston College the next week, also in South Bend... FSU wins national championship....

by young curmudgeon :: Wed, 12/03/2014 - 3:11pm

Surprise! A counter-argument from oaktoon! (Just ribbing you, I appreciate your responding with a thoughtful post and, unlike some on the board, I've taken the back-and-forth about the GB-NE game to be a legitimate effort to understand rather than just a fan's complaining.) To answer your question, yes, other things being relatively equal, I'd always prefer the head to head result. And you wouldn't want to hang your hat on the comparative results vrs. WVU, as the Mountaineers proved to be generous hosts by turning the ball over five times to TCU and still only lost on a last second field goal. Comparative results in general add many elements of subjectivity (although one could also describe it as "reasoned judgement")--"well, when you beat them, their left guard was injured, but when they beat us, the holder for our field goal kicker had a hangnail," etc. OTOH, I agree that teams should not be rewarded for playing Cupcake State, Western Downtrodden, and Lowe Tech for their out of conference games. I guess it comes down to how you interpret "other things being relatively equal," and I just think that the actual result of the actual game should carry a large amount of weight. I don't think it's a gross injustice if the committee selects TCU, but I can see how a Baylor fan would.

by Eddo :: Wed, 12/03/2014 - 6:06pm

"I just think that the actual result of the actual game should carry a large amount of weight."

I don't disagree, so let's look at the actual results.

Baylor won. By three. At home. That doesn't tell us very much; it's basically a draw.

TCU did blow a 21-point lead in the fourth quarter. That's bad for TCU.

But Baylor was down by 21 in the fourth quarter! That's bad for Baylor.

So I'm comfortable - as a fan of neither team, who generally roots for both, as traditional underdogs - weighing the rest of their seasons more strongly than their head-to-head matchups. And over the rest of their seasons, TCU is undefeated, while Baylor has a loss.

by young curmudgeon :: Wed, 12/03/2014 - 9:26pm

Fair enough, I guess it depends on how much evidence you feel is necessary to explain away the result on the scoreboard. You and oaktoon think there's enough, I'm in favor of weighing the head to head result heavily, but you raise good points and, as I said, I don't think it's a huge injustice, just unfortunate. However, if TCU wins the national championship (and in truth they would be my preference among the remaining reasonable contenders), I wouldn't mind if one of those billboards appeared outside of Waco reading "Baylor 61, National Champions 58."

by RickD :: Tue, 12/02/2014 - 9:39pm

Well, there were two fumbles. Green Bay recovered both of them. If you believe in "fumble luck", that should only happen ~1/4 of the time. A turnover in a game played as tightly as this one was could have made a huge difference.

by oaktoon :: Tue, 12/02/2014 - 9:47pm

Just have to do this one more time-- Hyde's play was not a fumble, or a muff. New England had already touched the ball on the punt. The only risk he was taking was possessing the ball, then getting hit for a fumble... He never possessed the ball....

by Perfundle :: Tue, 12/02/2014 - 9:57pm

Then the fault is in the play-by-play for recording it as a muff, and not as a zero-yard penalty on New England.

by chemical burn :: Tue, 12/02/2014 - 10:12pm

Can someone at FO see how not recording it as a fumble affects DVOA? I wonder if there's a compounding effect where where it sees GB recovering a single fumble as lucky and recovering two fumbles as extremely lucky some order beyond that.

by justanothersteve :: Tue, 12/02/2014 - 11:12pm

Yes, the muff definitely needs to be fixed. Hopefully this can be changed by the game trackers as it's obviously an error by the NFL statisticians.

by RickD :: Tue, 12/02/2014 - 11:21pm

I was explaining the disparity in the DVOA ratings. DVOA "thinks" it was a fumble.

by chemical burn :: Tue, 12/02/2014 - 9:02pm

Wait - what the hell? Am I understanding this right? Philadelphia is ranked first in special team and first in punt return WITHOUT their blocks factored in? Are field goal and extra point blocks also handled similarly? Because, holy smokes, Philly special teams are being UNDERRATED?!

FO really should come up with a better way to handle "non-repeatable" special teams stuff like the Eagles punt blocks and Indy's surprise onside kicks. These are very valuable, smart plays that are clearly intentional and influencing the outcome of games. Having two teams in one season that the system just decides to ignore is... not optimal.

by Perfundle :: Tue, 12/02/2014 - 9:11pm

Blocks are factored in. The punt return team doesn't get any credit for what happens after the block.

by chemical burn :: Tue, 12/02/2014 - 9:23pm

It's not clear - the guy in the question makes an assumption that isn't directly confirmed. What you (and he) are saying sounds right, though...

by RickD :: Tue, 12/02/2014 - 9:33pm

I'm reading "after the block", not "after the kick". So either it's been edited or you misread it.

It would make no sense to not credit the special teams for blocking a kick.

by chemical burn :: Tue, 12/02/2014 - 9:39pm

Unless they're so rare or erratic as to be non-predictive, like surprise onside kicks. How many punts get blocked in an average year? 2? The Eagles have 3 already this season, plus two blocked extra points and at least two deflected field goals. How often are extra points blocked?

by RickD :: Tue, 12/02/2014 - 10:00pm

Being rare isn't the same as being erratic. Blocking kicks is clearly an example of ST skill.

by chemical burn :: Tue, 12/02/2014 - 10:21pm

I agree with you - there are plenty of things in DVOA's calculations that I don't agree with or think are weakly reasoned. That's why I'm curious how this is being handled. It's not 100% clear from how it's written above. I think they are counted in the team's ST Punt Return number upon re-reading the question, but it's not clear.

Also, that the Eagles have an enormous "Hidden" number on their ST page makes me feel like some of what they're doing is not clearly being attributed to them as skill - they have a #1 ranked 20.9 "Hidden" score. There's only one other team above 11 points (yards? not sure how to characterize the number.) I suspect that at least their blocked extra points and deflected field goals are not being attributed to their ST ranking, even if the blocked punts are.

by techvet :: Wed, 12/03/2014 - 1:23am

Bizarre game of the year is where the Eagles score three non-offensive TDs and yet score no offensive TDs and lose to the Niners (is my memory correct?).

by Bobman :: Tue, 12/02/2014 - 9:11pm

Some of these don't quite pass the eyeball test for me (and maybe that's because they're quite bunched up in the second tier) but BAL as the #4 team with a #9 O and D? Miami as #6 overall with #10 O? (The #7 D I buy) Philly seems kind of low and KC kind of high. I think I like IND over BUF eight times out of ten (their DVOA is pretty close, but still....) Actually, ARI and BUF look extremely similar except for special teams with a pretty big delta in DVOA--if they played ten times (and Fitz returns from injury, which they keep saying is any day now), I don't get the feeling BUF wins the majority of the games.

Is the data more bunched up this year compared to prior years (meaning that clusters of 5-6 teams are all closer than usual and the 1-32 rankings are pretty fluid)?

by chemical burn :: Tue, 12/02/2014 - 9:20pm

Yeah, no great teams this year, not even any great individual units. Denver's number feels even too high to me.

(Also, Philly is not too low, they turn the ball over too much and their secondary is horrible. McCoy's fumble on their own 12 this week, score 23-7 easily could've let the Cowboys back in the game. Teams like GB and Denver are always going to blow them out through the air.)

by bubqr :: Tue, 12/02/2014 - 11:18pm

Sadly, I agree - I don't see how the Eagles can win vs GB, DEN, NE, and even NO. It seems that they get absolutely torched every time they face a top QB.

by chemical burn :: Wed, 12/03/2014 - 12:14am

Ha - I've been hesitant to add New Orleans when I list the teams I absolutely think the Eagles have no shot against because it seems so absurd and I don't want to undermine my point... but yeah, I don't see them beating New Orleans either. I could see it being a game if I squint, whereas with GB, Denver and NE, I just don't see how they could avoid a blowout, let alone win. I think the scenario has to start with "Their QB is injured on the first play from scrimmage, a sack-fumble returned for a TD by the Eagles." So, as I never root for injuries, I therefore cannot even root for the Eagles to win those match-ups...

by Perfundle :: Tue, 12/02/2014 - 9:31pm

Some of these don't quite pass the eyeball test for me (and maybe that's because they're quite bunched up in the second tier) but BAL as the #4 team with a #9 O and D? Miami as #6 overall with #10 O?

Well, they match up pretty well with traditional stats: Baltimore is 6th in scoring offense and 7th in scoring defense, and Miami is 10th in scoring offense.

by chemical burn :: Tue, 12/02/2014 - 9:36pm

Eh, I think it's bunching - #2-#5 it thinks are more or less the same, #6-#8 as well and then that run of teams that everyone agree are just ok or somehow flawed #9 down through #15.

by Noah of Arkadia :: Wed, 12/03/2014 - 2:12am

If we're talking about offense, it looks like #1-2, #3-5, #6-8, and then maybe #9-11. There's a nice chunk of DVOA real estate between #9 and #15.

As to the other points, Kansas has faced the toughest schedule of defenses. Miami has played the 4th toughest. Both seem to suffer more from a lack of sexiness than a lack of efficiency. Philly, meanwhile, has faced the 21st, but they're pretty sexy. Higher than #7 overall with Sanchez at QB, though? I just can't see it. If Foles were around, sure. Baltimore is the one I don't get, either.

Who, me?

by chemical burn :: Wed, 12/03/2014 - 2:36am

KC definitely suffers from having an offense that operates almost exclusively even through the air at the line of scrimmage. It's such an anomaly in the modern NFL that I don't know how to judge it. If your QB almost never throws the ball more than 12 yards downfield, can it actually be any good? It seems impossible, but they've made it work.

Philly's offense is a weird thing to judge with DVOA. They were a mess to start the season with massive o-line injuries and huge problems with their #2-#4 wr's. The o-line has stabilized allowing their RB to function and Matthews developing has upgraded their #2 wr position (and #3 wr slot in the sense that having a viable second option takes pressure/focus off of the #3.) Their underperforming TE has been more or less benched for a solid veteran who can be relied upon to run the correct routes (and, incidentally, blocks much better helping with the RB's productivity and pass protection.)

They're really a different team on offense since the Texans game and the mediocrity of the new QB masks how much has changed. They went from being below average in one way to maybe just about average in a completely different way. I'm not sure how DVOA should handle them, but one of DVOA's biggest holes is accounting for injuries. Again, it's stuck between descriptive and predictive - in the case of Philly, I think their season before the Texans game has close to zero predictive value. So how to rate them, I just don't know. "Maclin is really good" is the only thing that has remained consistent all year...

by herewegobrownie... :: Tue, 12/02/2014 - 10:26pm

Biggest surprise is Oakland averaging one more win. You're telling me that there are nearly as many scenarios in which they win 2+ remaining games as 0 games? ("Nearly" because I have to think 3 or 4 remaining wins is essentially 0 chance but not quite.) I'd assume the model has no way to account for Osweiler possibly being in all of week 17, and I'd think Buffalo would still be a clear favorite Wk16 despite it being at the Black Hole.

by TecmoBoso :: Wed, 12/03/2014 - 3:14pm

The Ravens and Eagles I see as sort of the same, two good teams who have played pretty easy schedules (Ravens best win @NO? Philly's @DAL?) and thus, DVOA and the public might have them a tad overrated. I actually think DVOA is on the money with Miami, a team where the schedule has made them look 'worse' than they are imo.

by ElJefe :: Wed, 12/03/2014 - 3:42pm

The Eagles have had a really weird schedule. They haven't yet played any of their division opponents twice, so we can just list their twelve opponents by perceived quality:

1. GB
2. Ari
3. Ind
4. SF
5. Dal
6. Hou
7. StL
8. NYG
9. Was
10. Car
11. Ten
12. Jax

Opponents 1-6 is the Eagles road schedule, 7-12 is the Eagles home schedule to date. In week 14 the Eagles finally play a home game against a team with a winning record. All five previous games vs. +.500 teams have been on the road.

I'm not sure I see this as an unusually easy schedule; I guess you'd choose the AFC South as the easiest non-conference but the NFC West + GB is the most difficult in-conference. Seems fairly average other than there being two bad teams in the NFC East. The other thing is that there isn't much of a middle class among these teams (Hou & StL), everyone else is either playoff-worthy or awful.

Overeducated Layabout

by Perfundle :: Wed, 12/03/2014 - 4:07pm

How is that not an unusually easy schedule? They've played 5 of the bottom 8 teams by DVOA and only one in the top 8. Playing Green Bay is canceled out by playing Carolina. And if you want to call St. Louis and Houston middle class then you have to call Indianapolis, San Francisco, Arizona and Dallas that too.

by chemical burn :: Wed, 12/03/2014 - 4:06pm

It's also a weird schedule because as far as the good teams they've faced, they have played a slate of teams that the public perceives as being very good, but that DVOA doesn't particularly like: SF, Arizona, Dallas and Indy. The only team DVOA loves that they've played is the team that blew them out of the water. So, I kinda disagree about the "middle class" comment - they've actually played the NFL's middle class, just that's not the perception.

by TecmoBoso :: Wed, 12/03/2014 - 7:03pm

I don't think it's been a weird schedule. It's been an easy schedule as the strength of schedule rankings back up (30th). And as I said, they really don't have a good win. The Dallas game probably is the best win (unlike the Indy game where they were fortunate to win). Meanwhile, they lost to @SF, @Ari, and @GB (three of the top four teams on their schedule). Then you look at the list after the top six and it's pretty poor, especially the bottom five. Like the Ravens, I'd be much higher on the Eagles if they had a good win (which may come this weekend against Seattle).

by chemical burn :: Wed, 12/03/2014 - 10:06pm

Oh - you're misinterpreting me. I'm an Eagles fan and I think they actually kind of stink. I think their schedule is causing them to be over-rated because of the gulf in perception between the public imagination (Dallas, SF, Indy and Arizona are good) versus DVOA's more realistic opinion of those teams works in the Eagles' favor. They beat Dallas and Indy and were a combined 3 feet from beating SF and Arizona. I think if they beat Seattle this week, it will mask their massive problems even more: Seattle is the best match-up for them as far as the elite teams are concerned. Even if they beat Seattle, they still won't be in the class of GB, NE and Denver even though a lot folks will think just that...

by Sporran :: Thu, 12/04/2014 - 8:35pm

Typical Negadelphian -- if the team isn't in the upper tier, they "stink".

"Stink" is an adjective that should only be applied to teams that are clearly below average, which the Eagles are not.

by chemical burn :: Thu, 12/04/2014 - 8:58pm

Sorry, a more accurate thing to say would be that they have units and key players who stink. And stink badly enough that I can't take them seriously as a title contender. So, "kinda stink." They definitely have good aspects, but don't kid yourself about just how awful their awful aspects are. Also, I'm at least an atypical Negladelphian in that I love Nick Foles and am dubious of Chip Kelly...

by Sporran :: Thu, 12/04/2014 - 10:20pm

There's no shame in being a Negadelphian -- if it's warranted (Phillies). With the Eagles -- their weighted FO ranking seems about right to me -- even considering their bad/questionable aspects (as I said, they are not in the top tier). This puts their chance of winning the Superbowl at about 10%. It's hard to see a reasonable revisionist history scenario where their chances would be much better than that by now. Mistakes have been made, of course, but it's not reasonable to expect them to be perfect.

I guess it depends on what your definition of "Championship Contender" is. There is a reasonable scenario where they win the Superbowl -- and it isn't far-fetched. That fits my definition of "contender". It is extremely unlikely that they win a championship (90% odds against), but if you have enough years like this, eventually things will fall into place.

by Alexander :: Tue, 12/02/2014 - 9:11pm

Da Bears is clearly ranked too high because Jay Cutler sucks. Da Coach is way better than this. Y U so Biasd?

by Anon Ymous :: Tue, 12/02/2014 - 9:40pm

Aaron, can you explain why DVOA has Denver as 6% more likely to make the playoffs than NE and 15% more likely to win their division? They have the same record, virtually identical weighted DVOAs and similar division competition. NE's future schedule is more difficult, but they also have an extra game advantage on their closest foe. Yes, that difference might not be important considering how daunting SD's schedule is, but KC's slate is the easiest of all, by a substantial margin.

Denver with an edge is certainly reasonable, but the size of it seems disproportionately large. Is it the future schedule? Or the fact that Denver has the H2H tiebreaker over their most likely competition whereas NE has a loss? Something else I'm missing?

by Perfundle :: Tue, 12/02/2014 - 10:14pm

They don't have similar division competition if you dive into the details. For New England, Miami and Buffalo are both above 10% in wDVOA, and both get another crack at beating New England. If Miami wins out and New England loses one more game, Miami wins the division. For Denver, San Diego is low enough in wDVOA that the predictor doesn't give them much of a chance to pull even with Denver despite being only one game back, especially since they have the hardest schedule remaining compared to 6th-hardest and 3rd-hardest for Miami and Buffalo, respectively (and Miami's schedule is easier than DVOA has them down for because they have three home games). As for Kansas City, Denver holds a massive advantage over them because they swept them, and even in a three-way tie with them and San Diego would win the tiebreaker, so for Denver to lose hold of their division they would need to lose at least three more games or San Diego needs to win out, and both are exceedingly unlikely.

by Anon Ymous :: Tue, 12/02/2014 - 10:24pm

That makes sense. So, if Denver still had to face KC instead of Cincy, for instance, the odds would probably be similar. Their 4-0 division record compared to NE's 2-1 is pretty much the difference.

by RickD :: Tue, 12/02/2014 - 10:26pm

Good summary. The key is that Miami has beaten NE and neither SD nor KC has beaten Denver.

by big10freak :: Tue, 12/02/2014 - 9:41pm

Green Bay's special teams keep taking steps backwards in performance. I hope they can turn it around.

by NYMike :: Tue, 12/02/2014 - 10:27pm

I think a lot if it this week was Hyde's "muff" and Crosby's lousy kick-offs, that were either short or went out of bounds.

by chemical burn :: Tue, 12/02/2014 - 10:36pm

It was overlooked because it was a blowout, but their ST in the Eagles game were a disaster. I've never seen multiple missed extra points in a game with a healthy kicker.

by techvet :: Wed, 12/03/2014 - 1:27am

Don't forget Micah Hyde's 75-yard punt return for a TD. They also forced an Eagles fumble on a KO (Eagles recovered). It wasn't all bad.


by chemical burn :: Wed, 12/03/2014 - 2:04am

Well, to the extent that anything went badly for them in that game, the ST were bizarre and bad.

by DisplacedPackerFan :: Tue, 12/02/2014 - 10:52pm

GB has a couple of play by play stupidities that hurt special teams now. It lists Hyde with a muff in this game, it also doesn't give Boykin credit for a blocked punt when he McKee the back before the punter a few weeks back. That was listed as a fumble for the other team. That being said the special teams have had some bad plays in the last couple of weeks. Crosby kicking out of bounds, the blocked punts, extra points, and field goals. Lots of inconsistency with the special teams.

by kingfisher413 :: Wed, 12/03/2014 - 1:12am

Hyde's muffed punt was on a free play, no?

Also, what did DVOA say about Miami's partial punt block against the Jets?


by mitch :: Wed, 12/03/2014 - 9:32am

What my power ratings showed was GB winning by more than the final score, not the NE , as DVOA would have you believe.

I think NE is the team that got lucky in this game, and DVOA will give you the wrong perception of the 2 teams.

by JoeyHarringtonsPiano :: Wed, 12/03/2014 - 9:45am

Detroit is still #1 in defensive DVOA, but by far less of a margin than earlier in the season. Is this due to opponent adjustments, or decline in recent play? My eyes tell me it's the latter, because Atlanta, Arizona, and New England figured out that the 3rd string nickel corner is the achilles heel of this defense and attacked it vigorously.

by justanothersteve :: Wed, 12/03/2014 - 11:42am

This will probably also be the plan week 17 for Green Bay. Anybody who is still playing FFL then might consider a flyer on Devante Adams that week.

by IntoTheVoid :: Wed, 12/03/2014 - 10:04am

I'm wondering about Philadelphia's numbers. According to the DVOA, they are completely average on offense, but at the same time, they are only a mere touchdown from being the highest scoring team in the league. How can they be scoring as much as offensive powerhouses like GB and Denver if they are mediocre on offense? Conversely, they rank 8th in DVOA defensively, but are tied for 19 in points allowed. Could all this have to do with Chip Kelly's coaching, that more points are scored in his games on average than would be scored in a game with two other equally effective teams? And if so, is there any way to illustrate why with mathematics?

by jacobk :: Wed, 12/03/2014 - 10:27am

Their special teams have been absurdly good. Return touchdowns + short fields will let you get more points per DVOA percentage point than average.

by Eddo :: Wed, 12/03/2014 - 1:51pm

"According to the DVOA, they are completely average on offense, but at the same time, they are only a mere touchdown from being the highest scoring team in the league. How can they be scoring as much as offensive powerhouses like GB and Denver if they are mediocre on offense?"

The Eagles' defense and special teams are directly responsible for 70 points, nearly six points per game.

If you take away seven points for every non-offensive touchdown, the Eagles fall to 25.4 points per game, closer to the Giants than the Broncos and about as close to the Chiefs as the Packers.

Total points is rarely a good way to judge an offense.


And, looking at FO's drive stats, the Eagles are fourteenth in points per drive. Their 1.99 points per drive sits between Cincinnati (1.89) and Atlanta (2.09).

(EDITed to correct drive stats to be offense-only, not net.)

by Pat :: Thu, 12/04/2014 - 1:24pm

And this, really, is one of the biggest advantages of DVOA in my mind. If you look around at other ratings, the Eagles frequently come out with a high ranking on offense in any game-based metric (PFR's Simple Rating System, AFA's Win Points Added, etc.), but that's really misleading whenever a team has significant special teams (or defensive scoring, but that's rare), and manipulates pace like the Eagles do, since you're comparing that team's performance against others at a fast pace to the other team's performance against others at a normal pace.

I always thought it would be interesting as a trivial thing to 'rescore' games, where offenses score points, defenses *take away* those points, and special teams non-FG/PAT points which get added at the end.

So, for instance, the Eagles-49ers game instead of being 26-21 San Francisco would've been 19-0 normal points, 0-14 bonus points. Makes that game look a lot different.

by crw78 :: Wed, 12/03/2014 - 10:45am

They've scored a ton of points on kick/punt/int/fumble returns, significantly more than anyone else in the league, so their points scored is not a reflection of the quality of their offense. On defense, they give up more yards and points than they otherwise would because the offense moves so quickly, therefore the defense faces more possessions than a normal team. I believe Philly has the lowest TOP in the league.

by thebuch :: Wed, 12/03/2014 - 11:10am

Their offense is also hurt by a lot of turnovers (a MUCH higher rate than those offensive powerhouses), while their defense gets a lot of turnovers. DVOA loves turnovers on defense and hates them on offense.

by tuluse :: Wed, 12/03/2014 - 12:07pm

Philadelphia is 14th in points per drive. They have by far the fastest pace in the NFL creating more drives per game than other teams.



by chemical burn :: Wed, 12/03/2014 - 2:08pm

I agree with everything everyone else is saying and I'd also throw in, "look at their individual player pages." They have several players with significant roles in their offense with very bad DVOA: McCoy, Cooper, Huff and Celek jump out. There are mitigating factors for all those players's situations, but the badness of their numbers shows you why their offensive DVOA is low: they have a bunch of guys seeing significant time who have been performing poorly. And I think anyone who has watched this team would agree that those guys have not been productive - it's not controversial to say McCoy and Cooper have been major disappointments. If you have several key players who have been very bad on a per-play basis, your offense has likely been bad on a per-play basis.

I don't mean this as an explanation of why the offense has been shaky, just proof of it.

(Also, have you watched this offense? Until the o-line looked healthy in around week 11 or so, they were a total mess...)

by Nevic :: Wed, 12/03/2014 - 12:15pm

All the talk about the GB fumbles* reminds me of a DVOA question I have wondered about for some time: How does DVOA handle plays, like the Rodgers fumble where he dropped the ball as he was being tackled and immediately fell on it to recover it, if it does not show up in the box score as a fumble. In many cases the officials don't throw the blue bag and unless there is a replay from a different angle, I'm not sure that ever gets recorded as a fumble, when it actually was one by the rules of the game. Is this recorded as a fumble in DVOA, or not, or up to how the game charter records it? Thanks.

by Eddo :: Wed, 12/03/2014 - 1:51pm

If it doesn't show up in the play-by-play, DVOA won't consider it a fumble.

There are *some* manual adjustments to the play-by-play that Aaron makes, such as end-of-half/game interceptions on hail maries.

by nat :: Wed, 12/03/2014 - 4:55pm

Some lessons to be learned about DVOA from the GB-NE game.

DVOA and VOA are per-play stats. Sometimes being better per-play doesn't translate into being better per set-of-downs or per-possession. And sometimes being better per-possession doesn't result in better results in a game.

For example: New England was better at getting first downs per play. It was exactly as good at moving the chains (DSR in drive stats), succeeding 80% of the time. It nevertheless had fewer first downs per possession and in the game.

DVOA and VOA come almost completely from the play-by-play. It's a fumble if the play-by-play says so. Unaccepted penalties don't count. DVOA doesn't know about "free plays". That "free" return that gets fumbled or muffed out of bounds for a net gain in yards? DVOA sees it as a return that gets fumbled out of bounds after a gain.

DVOA treats fumbles as predictive and recoveries as random.

VOA values offensive fumbles based on who actually recovered. It still treats special teams fumble recoveries as random.

VOA probably doesn't value first downs versus yardage versus fumbles vs interceptions the same as you do. You are probably wrong. For decades, football statisticians focused on yardage and points as the only team stats worth comparing. We have left that dark age. Get over it.

Usually, if VOA disagrees with a game's score, you can figure out what VOA was seeing. In this game's case, it's better first downs per play, no fumbles or muffs, and a better ability to get into the end zone being valued over more yards per play (which VOA sees) and simply having more plays (which VOA ignores).

Total VOA and Total DVOA are constructed assuming an equal number of offensive and defensive snaps. That can be wildly off in a single game. This means Total DVOA is more about predicting future play than it is about describing the current game.

by tuluse :: Wed, 12/03/2014 - 7:21pm

Good post.

I would be interested in seeing (ODVOA * OPLAYS) + (DDVOA * DPLAYS) + (SDVOA * SPLAYS). Seems like it would be more enlightening in single games than the current method.

by dank067 :: Thu, 12/04/2014 - 12:13pm

You made me interested too so I just tried this quick with one adjustment, multiplying VOA/DVOA by the % of total plays that unit was on the field rather than just total plays so that the final result can be still expressed as a percentage. I estimated that there were 135 total plays in the game, 65 plays for GBO/NED, 49 plays for NEO/GBD, and 21 special teams plays, and so I multiplied GB's OVOA of 27.0% by 48.1% of the total plays in the game that GB's offense was on the field, etc. In the end I got "possession-adjusted" VOA Scores of NE 1.3%, GB 0.84%, and DVOA Scores of NE 12.6%, GB 8.3%.

I don't have a great statistics background so maybe someone else could chime in as to whether what I did here was valid. One thing in particular that probably throws this off is that I think Aaron has said in the past that the special teams numbers are already adjusted to reflect the fact that there are fewer special teams plays in a game.

by Grendel13G :: Wed, 12/03/2014 - 6:46pm

This is good stuff.

by Vandal :: Wed, 12/03/2014 - 6:35pm

One thing that struck me as odd, and important during the GB-NE game, was the Fake 12 man penalty at 2:35 left in the 2nd half.

The whistle/restart prevented NE from letting the clock run down to 2:00 before starting the next play. As a result, an additional 40 seconds was left on the clock for Rodgers on the next possession. And Jordy took one to the house... (I know GB could have taken a time out, but frankly I won't ever give McCarthy the benefit of the doubt on clock management).

Just a bizarre, that an undone penalty call prevents the clock from running the full 40 seconds...

by SFC B :: Thu, 12/04/2014 - 3:48am

Wasn't GB on offense when it happened?

by Vandal :: Thu, 12/04/2014 - 12:18pm

No, New England was. Penalty was "12 men in the huddle". Which was reversed.

by dank067 :: Thu, 12/04/2014 - 12:42pm


It happened in the first quarter and GB was on offense. Unless there was another screw up later that I don't remember?

by Flounder :: Thu, 12/04/2014 - 1:26pm

GB definitely had the ball. I remember thinking it was so weird because the penalty was called before the snap, and when it's on the defense that penalty is almost always after the ball is snapped.

The whole thing was really bizarre.

by dank067 :: Thu, 12/04/2014 - 2:51pm

*Edit, misunderstood you at first.

But yeah, as of I believe 2012, if the refs recognize 12 men and the snap is "imminent" it can be called as a pre-snap penalty on the defense.

by Vandal :: Thu, 12/04/2014 - 8:08pm

You are referring to a Different penalty. Which was also reversed.

Correction: It was an Offensive Pass Interference flag, which was called, then picked up. So it stopped the clock and made the maximum run off at 25 seconds, rather than 40. Preventing the Pats from going down to the 2 min warning on the next play.

by Ranccor :: Fri, 12/05/2014 - 12:54am

I'm a bit surprised by the Colts offensive rating. Their passing game has been pretty incredible, while the running game seems to (mostly) get the job done. Is there huge gulf in the split between the running and passing game? Or perhaps, a few stinkers (most notably the NE game) dragging them down? If not, why does FO rate them as basically a slightly above average offense (+4.3%) while the general perception of them is as a very dangerous offensive team and conventional stats have them as top 3 in many offensive categories?

Disclaimer - I'm a Colts homer, so my perception could be skewered.

by Perfundle :: Fri, 12/05/2014 - 3:46pm

I think the Colts are a better team than DVOA has them down for, but not because DVOA is wrong. It's because they're woefully misusing the talent on their team, and in particular Richardson should not be getting many (read: any) rushes. Luck has been great this year, and non-Richardson backs are doing pretty well even with their fumbling issues, so they should stop letting their pride get in the way of actually winning games.

by chemical burn :: Fri, 12/05/2014 - 9:38pm

Colts run more plays on offense than any team in the league - that means their conventional stats will be high, whereas DVOA is a per-play stat. It's a similar effect to what's discussed with the Eagles above. (Also, I believe the Colts turn the ball over a fair amount, don't they? Luck's 11 interceptions are not a disaster, but are not great. Again - that's discussed in the Eagles thread. They're similar in why DVOA dislikes them...)

by Perfundle :: Fri, 12/05/2014 - 10:43pm

They're average with respect to interceptions, but horrible with fumbles. They've had 24 fumbles, which is the third-highest fumbling rate per play.

by mitch :: Sat, 12/06/2014 - 10:00am

Crosby gets a bigger deduction because he missed a shorter FG ?

But wouldn't GB get more credit for their offense getting deeper into NE territory and getting a closer FG attempt ?

That extra credit should out-wiegh the bigger deduction or at a minimum even things out.

So by giving Crosby an extra deduction DVOA is saying the Pats did a better job on their drive then the Packers did while getting a closer FG attempt.

How is that possible ? How can that be credible ?

A team gets rewarded for attempting a longer FG ?

by chemical burn :: Sat, 12/06/2014 - 4:15pm

Descriptive of a single game, it's not that great. Predictive of what might happen in the future, it's perfectly logical.

by tuluse :: Sat, 12/06/2014 - 6:17pm

DVOA grades on a per play basis, not on a per drive basis.

by LionInAZ :: Sat, 12/06/2014 - 7:37pm

itYou're thinking about it incorrectly. GB is being penalized for missing a relatively easy FG. NE is penalized less for missing a longer FG. It's about the percentages of making each. No one is being "rewarded" for missing FGs, as you put it.