Writers of Pro Football Prospectus 2008

18 Jan 2016

Week 19 DVOA Ratings

by Aaron Schatz

Once again, it is time for postseason DVOA ratings. As always, the following rules apply:

  • All 32 teams are ranked, whether they made the playoffs or not.
  • Teams are ranked in order of weighted DVOA, not total season DVOA. Since weighted DVOA is meant to lower the strength of older games, these ratings do not include Weeks 1-5, and Weeks 6-11 are somewhat discounted. "Last week" here refers to last week's rank in weighted DVOA, not total season DVOA.
  • Teams are treated as having a bye week in any week where they did not play. Since most teams haven't played in two weeks, that means some of the ratings for non-playoff teams can start getting a little unreliable. Really, this is only to be used for playoff teams, the other teams are just there for ranking comparison purposes.
  • DVOA, as always takes a long-term view of an NFL team's performance. The games of the last two weeks are just two games among many, so teams may be listed below other teams that they have beaten in the playoffs.

Since we've been writing a lot again these last couple weeks about the question of weighted DVOA vs. total DVOA, and which is more accurate for predicting the postseason, this year I'm including total DVOA in the postseason DVOA ratings table. The total DVOA listed below adds the eight postseason games in with all 256 regular-season games. Obviously, adding eight games into a sample of 264 games doesn't change things much.

Of course, if you scroll down the page to the full ratings, you will notice something really strange: the top three teams are all teams that lost this weekend, not teams that won. There are a handful of reasons why this is the case.

First, remember what I just said about taking a long-term view of an NFL team's performance. Kansas City's loss to New England was just one game; it adds information to what we knew about a Kansas City team that won 11 straight games, but it doesn't mean those games didn't happen. Likewise, New England's strong performance against the excellent Kansas City defense was definitely a positive indicator for the Patriots offense going forward, but it doesn't mean the offense didn't struggle in December. We know the return of Julian Edelman was probably a big part of the improvement, but DVOA only measures the plays on the field. We don't adjust it for every single injury; that's what common sense is for, and why we write commentary rather than having a website that consists of nothing but tables of numbers.

Second, weighted DVOA ratings can sometimes seem a little screwy because of the weights we use and the specific timing of when games drop in importance. (That's something we test every couple years, so there's always a chance we'll change that weight system in the future.) For example, the last game of Kansas City's 1-4 start drops out of weighted DVOA this week. (Yes, it was actually a 1-5 start, but I've written a few times about how the Chiefs actually had a really high DVOA in Week 6 despite losing to Minnesota.) There's also a drop in weight for Week 11, which happened to be the first of Carolina's real blowout wins (44-16 over Washington) and the first week Brock Osweiler started at quarterback for Denver and started to pull the Broncos' offensive DVOA out of last place.

Third, this week's games did not come out with DVOA ratings that necessarily matched the final scores. On the field, all four games were close, six or seven points. However, when we break down the play-by-play, DVOA scores three of the games as so close that the team with the better DVOA (or VOA without opponent adjustments) actually lost, while the Kansas City-New England game "should have" been more of a blowout.

Here are the one-game DVOA ratings for the divisional round:


DVOA (with opponent adjustments)
TEAM TOT OFF DEF ST
KC -10% 5% 24% 10%
NE 74% 60% -10% 3%
GB 39% 7% -37% -4%
ARI -10% -18% -3% 5%
SEA 53% 47% -3% 4%
CAR 33% 30% -7% -4%
PIT 32% 41% -1% -10%
DEN 30% 3% -6% 20%
VOA (no opponent adjustments)
TEAM TOT OFF DEF ST
KC -30% -1% 38% 10%
NE 55% 48% -4% 3%
GB -2% -14% -17% -4%
ARI -12% -23% -6% 5%
SEA 15% 18% 7% 4%
CAR -4% 14% 15% -4%
PIT 13% 14% -10% -10%
DEN 8% -5% 7% 20%

I know I'm going to hear it from Carolina fans now. I hope they understand that the purpose of DVOA is not to enable me to say mean things about the Carolina Panthers. I'm sure that's the biggest DVOA shock among the four games, so let's take that one first. The most important thing to understand about this game is that Carolina got a 31-0 lead but did not actually win the game by that score. I know this seems like a simple statement but it's hard to really understand because the narrative in our minds is "Carolina blew Seattle out of the water from the first snap of the game." Which is true, but only for 30 minutes, and you have to play 60. Everything Seattle did in the second half is just as important as what Carolina did in the first half. This is the insane breakdown of DVOA in the first and second halves of that game:

Seattle and Carolina DVOA, 1H vs. 2H
TEAM TOT OFF DEF ST
SEA 1H -62% -41% 16% -5%
CAR 1H 140% 44% -88% 8%
SEA 2H 149% 110% -29% 9%
CAR 2H -59% 7% 54% -12%

In the end, this game ended up exactly what we would have expected beforehand: two really good teams playing a close game, won by the home team. It looked like a laugher for a while, but if you look at all 60 minutes, it wasn't. And remember, DVOA is primarily a measure of efficiency, and Seattle was actually much more efficient over the entire course of the game. The Seahawks averaged 6.2 yards per play compared to just 4.6 yards per play for the Panthers. That's somewhat balanced out by turnovers, because Russell Wilson threw two interceptions. Carolina didn't have any turnovers, but that's partly because the Panthers fumbled twice and recovered both of them. (One was a muffed punt, which is normally recovered by the return team, so that's not a big penalty in DVOA.)

Similar issues were at play in Pittsburgh's loss to Denver. The Steelers gained 6.7 yards per play compared to just 4.6 yards per play for the Broncos. Even more than in the Carolina-Seattle game, fumble luck ended up the differentiator. There were four fumbles, but three of them were plays normally recovered by the team that fumbles: an aborted snap by Denver and two muffed punts by Pittsburgh. The fourth was the Fitzgerald Toussaint fumble, and that was more of a 50-50 shot. It just happened to go straight into the arms of DeMarcus Ware, and that essentially turned into the difference between two evenly matched performances.

The final game along these lines was Green Bay-Arizona, where both teams averaged the same 5.8 yards per play. Carson Palmer threw two picks, Aaron Rodgers only one. Again, fumble recovery luck played a role, as there was only one fumble in the game, when Nick Perry sacked Carson Palmer, but Palmer recovered it himself. The Green Bay-Arizona game also has a much bigger gap in DVOA than in VOA because Green Bay was clearly the worst of the final eight teams over the course of the entire season.

Back to those Carolina fans who are so angry with Football Outsiders: the irony is that we now have Carolina is a very clear favorite to win Super Bowl 50. What I kept writing during the season was that the Panthers were not the historically dominant team that an almost-undefeated season would suggest, but they were clearly one of the very good teams in a year where there were a number of very good (but not historically great) teams. Well, the Panthers are now the standout team from that group of teams. The teams that dominated the second half of the season alongside the Panthers ended up wild cards and lost on the road this week. The teams that dominated the first half of the season alongside the Panthers generally faded in the second half of the year, which is why Arizona ended the regular season only slightly ahead of the Panthers and New England was behind them. The other team left, Denver, was generally behind Carolina all year because the Broncos were winning entirely with defense.

It doesn't matter anymore if Carolina was the best team over the course of the year, because the Panthers are the best team left and they get home field this week. What we have here is a really good team with a really good head coach, a lot of likeable players, and a fan base whose only professional championship is a Stanley Cup nine years ago won by a team that plays 100 miles from Charlotte. If I end up having to constantly spend the rest of my career having to explain that the 2015 Panthers, despite the 15-1 record, are a "best team of the year" like the 2008 Steelers instead of a "best team of all-time" like the 1985 Bears, it's not the worst thing in the world.

* * * * *

You will find DVOA matchup pages for the AFC and NFC Championship Games on the FO Premium page.

The following items will be updated later Monday night: snap counts for the divisional round, and scores in the Football Outsiders 2016 Playoff Fantasy Challenge. When those are updated, you will find the standings and all scores here. Note that there was an error with Alex Smith's score last week -- it was related to the errors that had Kansas City's DVOA against Houston listed too high -- and so that's been fixed in the scores now. If you can't find your team, you might be one of the teams that was entered without a name. In that case we should still have your e-mail to contact you if you end up winning.

We also have not finalized our Madden Ultimate Team content for next weekend, so I can't announce those players now. We'll announce that in a separate post on Tuesday.

* * * * *

To save people some time, we remind everyone to put their angry troll hatred into the official zlionsfan angry troll hatred Mad Libs form:

<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>

If you are new to our website, you can read the explanation of how DVOA is figured here. As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.

Teams in yellow are still alive in the playoffs. Teams in gray lost this past weekend.

TEAM WEI.
DVOA
LAST
WEEK
W-L WEI OFF
DVOA
OFF.
RANK
WEI DEF
DVOA
DEF.
RANK
WEI S.T.
DVOA
S.T.
RANK
TOTAL
DVOA
RANK
1 SEA 50.7% 1 11-7 26.9% 1 -22.0% 2 1.7% 15 37.8% 1
2 KC 39.6% 2 12-6 11.3% 7 -22.5% 1 5.9% 4 28.4% 2
3 PIT 30.3% 4 11-7 16.1% 2 -12.7% 6 1.5% 16 22.7% 7
4 CAR 29.4% 3 16-1 14.0% 5 -15.4% 5 0.1% 20 26.3% 4
5 DEN 21.5% 6 13-4 -0.5% 17 -20.8% 4 1.2% 19 18.6% 8
6 CIN 21.4% 5 12-5 5.7% 13 -12.3% 7 3.4% 10 28.1% 3
7 NE 18.8% 9 13-4 12.3% 6 -5.1% 14 1.5% 17 25.2% 6
8 MIN 16.9% 8 11-6 3.2% 14 -8.4% 10 5.3% 5 6.7% 11
9 DET 13.7% 11 7-9 10.5% 8 -0.7% 16 2.6% 12 1.1% 13
10 ARI 13.2% 7 14-3 8.4% 10 -10.1% 9 -5.3% 30 25.3% 5
11 NYJ 11.6% 10 10-6 2.1% 15 -10.5% 8 -1.0% 23 12.4% 10
12 WAS 7.4% 12 9-8 7.9% 11 5.5% 21 5.0% 6 -2.0% 15
13 GB 3.3% 14 11-7 -5.1% 19 -7.2% 11 1.2% 18 14.6% 9
14 BUF 3.3% 15 8-8 14.2% 4 13.4% 27 2.5% 13 2.7% 12
15 HOU 0.6% 19 9-8 -15.5% 28 -21.7% 3 -5.6% 31 -9.9% 21
16 CHI 0.4% 13 6-10 10.4% 9 13.0% 26 3.0% 11 -5.7% 18
TEAM WEI.
DVOA
LAST
WEEK
W-L WEI OFF
DVOA
OFF.
RANK
WEI DEF
DVOA
DEF.
RANK
WEI S.T.
DVOA
S.T.
RANK
TOTAL
DVOA
RANK
17 OAK -1.0% 16 7-9 -6.9% 20 -6.7% 13 -0.8% 22 0.0% 14
18 BAL -1.9% 17 5-11 -9.2% 22 2.8% 18 10.1% 1 -3.0% 17
19 TB -5.3% 18 6-10 6.1% 12 4.9% 20 -6.5% 32 -9.3% 20
20 STL -5.8% 20 7-9 -14.4% 26 -4.6% 15 4.0% 8 -2.2% 16
21 NYG -10.1% 21 6-10 -0.5% 16 15.8% 30 6.2% 2 -6.5% 19
22 IND -12.3% 22 8-8 -21.3% 32 -7.2% 12 1.9% 14 -12.9% 23
23 SD -13.8% 24 4-12 -3.9% 18 6.4% 23 -3.4% 28 -14.8% 24
24 DAL -14.2% 23 4-12 -18.4% 31 2.0% 17 6.1% 3 -18.3% 27
25 NO -17.4% 26 7-9 14.6% 3 28.9% 32 -3.1% 25 -18.7% 28
26 SF -18.1% 28 5-11 -11.5% 25 3.3% 19 -3.3% 26 -27.4% 32
27 JAC -18.3% 25 5-11 -8.4% 21 13.5% 29 3.7% 9 -16.0% 25
28 PHI -22.3% 29 7-9 -10.3% 24 16.3% 31 4.3% 7 -11.2% 22
29 MIA -23.6% 27 6-10 -9.7% 23 10.4% 25 -3.5% 29 -19.0% 29
30 ATL -23.8% 30 8-8 -15.7% 29 8.1% 24 0.0% 21 -16.1% 26
31 CLE -24.8% 31 3-13 -15.4% 27 6.4% 22 -3.0% 24 -23.0% 30
32 TEN -33.3% 32 3-13 -16.4% 30 13.5% 28 -3.4% 27 -26.5% 31

Posted by: Aaron Schatz on 18 Jan 2016

86 comments, Last at 24 Jan 2016, 3:13pm by Alternator

Comments

1
by Raiderfan :: Mon, 01/18/2016 - 6:58pm

"However, when we break down the play-by-play, DVOA scores three of the games as so close that the team with the better DVOA (or VOA without opponent adjustments) actually won, while the Kansas City-New England game "should have" been more of a blowout."

I think you meant to write "actually lost" not won.

5
by Aaron Schatz :: Mon, 01/18/2016 - 8:01pm

Oh, duh. I'll fix that.

63
by adamwickens :: Wed, 01/20/2016 - 11:37am

Your comment really confused me at first because I totally understood what you were saying. Then I realized that you aren't Raiderjoe.

2
by WeaponX :: Mon, 01/18/2016 - 7:22pm

Carolina was playing to kill the clock the entire 2nd half, not to impress a formula. It worked. I figure most Carolina fans are happy with the outcome.

Sometimes I even trip myself out.

22
by MJK :: Tue, 01/19/2016 - 1:53am

One fundamental weakness of DVOA, compared to, say, a win probability based model, is that it doesn't take time into account. Usually, this doesn't matter, but it does late in the 4th in close games, and it does for more of the 2nd half in blowouts.

To DVOA, a 1 yard rush on 1st down is always an unsuccessful play and a 5 yard reception or rush that goes out of bounds on 1st down is always a success. But if you are trying to bleed clock, in many cases the 1 yard rush that keeps the clock running actually increases your chance of winning more than the 5 yard gain that stops the clock.

Similarly, giving up a 15 yard play that ends in the middle of the field is always a failure of the defense to DVOA, but if the other team is out of timeouts or coached by Andy Reid late in the 4th, then it could be a success, in some game situation.

The other issue, at least on the defensive side, is that there is no correction for an opposing offense playing more aggressively when trailing big. A team that is trailing big will go for it on 4th down more, and generally call more aggressive plays, often abandoning the outdated but still pervasive notion of "establishing the run" and choosing more favorable pass-run ratios. However, DVOA compares the defense's performance in this situation to a baseline value of a defense facing an offense that is "normal"--i.e. inefficiently conservative.

37
by bravehoptoad :: Tue, 01/19/2016 - 12:52pm

One fundamental weakness of DVOA, compared to, say, a win probability based model, is that it doesn't take time into account.

How do you know that?

78
by Scott C :: Thu, 01/21/2016 - 1:18am

Probably the same way I do -- 10 + years reading this site and the comments.

DVOA does have adjustments for 'late and close' and blowout, if I recall correctly, but not drastically. The whole thing is a model that is tested against its accuracy of predicting _future games_. Tweaks that might help better explain why a team was good or bad in the past but do not help predict the future are not added to the formula. Its not a measurement of who was best, but who is most likely to win in the future.

I don't know how many years I've been around here, honestly. Long before DPAR became DYAR. Well before the falcons fans revolted one week because DVOA said they were not that good and just had great luck with fumbles and opponents missing field goals. (they proceeded to not have that luck and tank the rest of the year -- I'm not sure the falcons fans ever came back). Maybe 2004? In that time myself and many others that have been here a looooong time just know random things about DVOA.

28
by Pantherpryde :: Tue, 01/19/2016 - 5:32am

LMAO, Let's just use the CSA!!!! I have studied this sites analytics, and I see lots of flaws. I will not go in depth about other major indicators not being considered,but I will point out that Carolina's deep threat defense ranking is completely off!!! We have not faced a team with three great WR's along with a great QB. I am a huge panther fan, but I am going to Las Vegas and bet the house on Arizona!!! They will beat Carolina at least by ten points!!! This will never be a close game in any quarter!!!
Outsiders always falls back on the excuses the next week, but if they would of taken into consideration of the take aways and penalties, they should of known the Panthers would have won. Carolina will not have that luxury this week!!!!!!!

39
by Guest789 :: Tue, 01/19/2016 - 1:22pm

Remember kids, the more exclamation points you use, the more persuasive you are.

41
by Hurt Bones :: Tue, 01/19/2016 - 1:36pm

Just grateful he hasn't discovered caps lock.

55
by Bobman :: Wed, 01/20/2016 - 2:03am

+1

oops, I mean +1!!!

49
by Grendel13G :: Tue, 01/19/2016 - 4:43pm

What????? You're a genius!!!!!!!!

44
by SFC B :: Tue, 01/19/2016 - 3:40pm

Does this post count enough to invoke the FOMB curse?

46
by Eddo :: Tue, 01/19/2016 - 4:22pm

Would a post by a Panthers fan saying the Panthers aren't as good as FO is saying they are invoke the curse? Or is it an attempt to invoke a reverse curse? I'm so confused.

85
by Eleutheria :: Thu, 01/21/2016 - 6:30am

FOMB curse?

86
by Alternator :: Sun, 01/24/2016 - 3:13pm

Briefly - a fan complaining that FO is underrating their team because (reasons). It doesn't count if you make a solid case, though.

3
by panther91 :: Mon, 01/18/2016 - 7:25pm

Because DVOA meassures the efficency of moving the ball, it seems logic to me, that Carolina gets a bad DVOA. Carolina made a terrible job in moving the ball in the second half. It is another question, if it was a terrible job to win the game. The point is that carlina changed their play calling to conservatism at halftime both in defense and offense. So I would argue that a bad DVOA is right, but I am not so sure this rating should be used to predict the outcome of upcoming games. In a game were thepantehrs dont have a 31:0 lead they probably woulndt shift to clock milking mode and would make a better job at gaining yards per play.

4
by Karl Cuba :: Mon, 01/18/2016 - 7:42pm

I think this is where most people will end up after a 'common sense adjustment' (can we introduce CSA as a thing?).

27
by Pantherpryde :: Tue, 01/19/2016 - 5:29am

LMAO, Let's just use the CSA!!!! I have studied this sites analytics, and I see lots of flaws. I will not go in depth about other major indicators not being considered,but I will point out that Carolina's deep threat defense ranking is completely off!!! We have not faced a team with three great WR's along with a great QB. I am a huge panther fan, but I am going to Las Vegas and bet the house on Arizona!!! They will beat Carolina at least by ten points!!! This will never be a close game in any quarter!!!

Outsiders always falls back on the excuses the next week, but if they would of taken into consideration of the take aways and penalties, they should of known the Panthers would have won. Carolina will not have that luxury this week!!!!!!!

35
by Aaron Schatz :: Tue, 01/19/2016 - 11:31am

Somebody is not using the proper Mad Libs form.

40
by tuluse :: Tue, 01/19/2016 - 1:31pm

He's pretty close

The Panthers are clearly ranked too low because they have not faced a team with three great WR's along with a great QB. CSA is way better than this. I am a huge panther fan, but I am going to Las Vegas and bet the house on Arizona!!! They will beat Carolina at least by ten points!!! This will never be a close game in any quarter!!!
Outsiders always falls back on the excuses the next week, but if they would of taken into consideration of the take aways and penalties, they should of known the Panthers would have won. Carolina will not have that luxury this week!!!!!!!

48
by anotherpatsfan :: Tue, 01/19/2016 - 4:30pm

Carolina is clearly ranked way too low because Superman, duh! Back in the day counting the guns on Greg Hardy's bed was way better than this. GTFO with this nerd-geek magic formula negativity - game was over 20 minutes in and Seahag garbage time means nada!

or

Seattle is clearly ranked way too low because Beast Mode of Boom! Measuring how far Russell Wilson can walk on water is way better than this. NFL unfair to low seed west teams - real men can't play football during west coast brunch time smh fml.

or

New England is clearly ranked way too low because Patriot Way! Brofessor Gronk's Theorem of Spiked Football Deflation is way better than this. Brady can't throw the ball and catch it too! F'ing Welker...

These are sort of like haiku...

56
by Bobman :: Wed, 01/20/2016 - 2:10am

That was pretty sweet.
Been a while since I've seen a mass parody (or curb-stomping) of... how shall I describe them... logical-outlier comments. Or civil discourse-challenged comments. Or ones generated from other than planet earth.

I, of course, am betting not only my house, but my cars, and two of my three sons on the Packers, who REALLY won last weekend. Friggin' rigged coin toss. Once the NFL sees the error of its ways, they'll put them in their proper place in time for the NFCCG.

Damn, if they don't win, my wife may just declare football off-limits. Screw it--I like my odds.

59
by Duff Soviet Union :: Wed, 01/20/2016 - 2:47am

"Or ones generated from other than planet earth."

The first time through, I actually read that as "planet meth". Which is probably appropriate.

60
by tuluse :: Wed, 01/20/2016 - 3:18am

Hey bobman, awesome to see you commenting again

71
by Bobman :: Wed, 01/20/2016 - 5:55pm

Gracias. The Colts crappy year gave me ample time to not comment and actually get some work done while sitting at my desk.

80
by Scott C :: Thu, 01/21/2016 - 1:27am

Haha....

Yeah, charger fan here. (lower case intented) I read about 5x less football related stuff this year than usual.

61
by WeaponX :: Wed, 01/20/2016 - 11:17am

*standing slow clap*

Sometimes I even trip myself out.

6
by duh :: Mon, 01/18/2016 - 8:12pm

Yeah the Panthers clearly played a lot of we'll give you 15 yards for 20 seconds D in the 4th quarter particularly it seemed to me.

7
by Duff Soviet Union :: Mon, 01/18/2016 - 9:02pm

Forgive me if I don't find "you should count all the plays where my favorite team played well and ignore the ones where they didn't" to be a persuasive argument.

8
by duh :: Mon, 01/18/2016 - 9:07pm

Uh I'm not a Panthers fan, I just thought it was pretty obvious that the Panthers played a soft shell that kept everything in front of them and in bounds. Playing that kind of D gives up yards but eats clock.

23
by blarneyforbreakfast :: Tue, 01/19/2016 - 2:05am

I can see how DVOA might have gotten the wrong impression of winning teams. Pass plays are generally much "higher" than running plays, and a team that is protecting a lead will typically have far more run plays than usual. This is also mathematically a good idea for a team with a large lead--by using up clock you reduce the number of possessions the other team has to come back. We might think that Carolina would have benefitted from passing more, but opponents are also looking for the run, and it should count for something when teams are successful with the run in obvious running situations.

24
by Duff Soviet Union :: Tue, 01/19/2016 - 3:17am

"I can see how DVOA might have gotten the wrong impression of winning teams. Pass plays are generally much "higher" than running plays, and a team that is protecting a lead will typically have far more run plays than usual."

DVOA compares each play to a baseline of plays in the same situation (or very similar). So Seattle's offensive plays when down by 20+ are compared to the offensive plays of other teams down 20+. I'd say that would get rid of the "pass plays are higher than running plays" thing since everyone passes a lot in that situation.

I really think this game was just as simple as "Carolina totally outplayed Seattle in the first half and Seattle totally outplayed Carolina in the second half".

The only way Carolina gets penalised is because even by the standards of prevent defenses, theirs was pretty terrible. Their offensive DVOA for the second half was actually above average despite the 0 points which indicates that DVOA does see some value in simply not turning the ball over and grinding clock.

32
by Hoodie_Sleeves :: Tue, 01/19/2016 - 10:12am

". So Seattle's offensive plays when down by 20+ are compared to the offensive plays of other teams down 20+"

THat's a huge assumption - Aaron has refused to give us any of the criteria for how the bucketing actually works - so we have no idea what plays are actually being compared to. "Down 20 in the first half" could be the same bucket as "Down 20 with 2 minutes left" which are drastically different situations.

We don't know the yardage cutoffs, we don't know the points cutoffs, we don't know the time cutoffs.

10
by Karl Cuba :: Mon, 01/18/2016 - 9:09pm

You don't think that the score dictated play calling in that game at all?

13
by Duff Soviet Union :: Mon, 01/18/2016 - 9:14pm

It certainly dictated Seattle's pass heavy play calling, not so much Carolina's. They ran the ball a lot on first down, because that's what they like to do. They did that in the first half too. The difference was it was working then because Seattle's linemen looked like they were playing on rollerskates.

11
by Duff Soviet Union :: Mon, 01/18/2016 - 9:12pm

I'll also say that it didn't look to me like Carolina were noticeably more conservative in the second half, just much less effective as they stopped mauling Seattle on both lines of scrimmage.

And even if I were to grant that this was conservatism rather than ineffectiveness, why is this a good thing? Ask Mike McCarthy about the dangers of conservative playcalling when your highly regarded opponent is on the ropes. Those points allowed were not "garbage time", they mattered and if Thomas Davis had been Brandon Bostick, I think a lot of Panthers fans would be changing their tune.

14
by duh :: Mon, 01/18/2016 - 9:25pm

I agree that playing 'conservative' that way can certainly backfire and can be horribly enervating. But looking at PFR their win probability for the Panthers never drops below 93% in the 2nd half (and I think Burke had it never below 95%)

15
by Karl Cuba :: Mon, 01/18/2016 - 9:27pm

It isn't a good thing in terms of getting a nice DVOA score but running so much took time out of the game.

Maybe you don't agree but my impression was that the Panthers were trying to run the clock as a priority, much more than they would in a closer game.

Look at the other side, don't you think Seattle threw the ball much more than they usually would?

16
by dank067 :: Mon, 01/18/2016 - 9:31pm

The thing is that Carolina has already outright blown leads of 28 and 17 late in games this season. Add near-misses with leads of 23 and now 31, and this has happened in nearly a quarter of the games they played this year.

Could be a coincidence, they've closed other teams out and come back in games themselves. But whether it reflects a flaw in their pass defense, their coaching, their offensive playcalling, or any combination, I wouldn't be dismissive of it.

17
by Duff Soviet Union :: Mon, 01/18/2016 - 9:42pm

Exactly. It hasn't burned them yet, but if it happens again, it might.

18
by Karl Cuba :: Mon, 01/18/2016 - 10:30pm

Oh I'm not saying it might not be a problem, it could be their strategy in such situations is really bad, all I'm offering is that the DVOA imbalance does not in itself suggest that Seattle outplayed them.

At the same time the fumbles didn't hurt Carolina but are counted in the DVOA numbers but even then I think the Panthers were clearly the better team that day and would likely have a better DVOA if they'd been less conservative with a 31 point lead.

19
by dank067 :: Mon, 01/18/2016 - 10:55pm

I'm with you, Seattle got some tips/close conversions to go their way as they climbed back into the game as well.

I do want to say that the comebacks against Carolina suggest to me that a team willing to throw aggressively downfield early and often could have some big-time success against Carolina's secondary, and the Cardinals can definitely play that way. But we don't know exactly what Palmer's condition is right now, and that does leave you susceptible to Carolina's swarming front 7 and turnovers.

25
by Pantherpryde :: Tue, 01/19/2016 - 5:13am

I agree Duff, If they were going conservative, they would of taken out the starters!!! If you look at their trend, They always shoot out the gate in the first half and suck in the third quarter. Then respond in the forth if they are not more than two TD's behind!!!

50
by yayFootball :: Tue, 01/19/2016 - 5:55pm

How one factors conservative tactics into deciding which team played better overall is debatable, and no sane person would dispute that Seattle outplayed Carolina in the second half, but Carolina was clearly more conservative after getting the big lead. When they got the ball back with just over two minutes left in the first half, they ran three times (including on third-and-6)and punted. They obviously weren't trying to score on their last possession of the game either, instead kneeling to run out the clock. On the possession before that, the Panthers also dialed up a running play on third-and-7, indicating they were again primarily concerned with running time off the clock rather than advancing down the field. That may not sound like a lot of instances of conservatism, but it represents half of Carolina's possessions after they got the big lead and 30% of their total possessions. It also seemed like the Panthers weren't attacking down the field as much on their passing plays in the second half, but it wasn't clear whether that was due to conservative play-calling or good coverage.

29
by Dr. Bill :: Tue, 01/19/2016 - 9:41am

In the Marine Corps' warfighting doctrine, we call this "trading space for time."

30
by Dr. Bill :: Tue, 01/19/2016 - 9:53am

"Carolina made a terrible job in moving the ball in the second half. It is another question, if it was a terrible job to win the game."

I think this is an inherent limitation of DVOA--it measures one thing (play efficiency) as a proxy for something else (winning). It seems to be a decent proxy, in that play efficiency correlates pretty well with winning, and it's hard to be good at winning games without being good at plays. But we all recognize certain contexts under which seemingly inefficient plays move the team higher in win probability.

I don't understand the inner working of DVOA very well--could DVOA be adjusted to recognize "time efficiency" during 2-minute/4-minute drills based on lead? So that a play which gains a first down, but burns up 30 seconds when you're down by two scores with 3:00 to go, isn't a success? Or a defensive play that gives up 15 yards in the middle of the field, and keeps the clock running at 0:40, up by 7, is a defensive success?

33
by Will Allen :: Tue, 01/19/2016 - 10:58am

In the history of the league, there has been one comeback from a 32 point deficit, and two from a 28 point deficit. I don't know if there is much insight to be gained from the 2nd half metrics of such a game, on a macro level. On a closer look, however, the success Seattle had in the 2nd half throwing deep may tell us something about Carolina's dbs after their number 1 guy, given Arizona's success at throwing deep. This game is interesting to me, because I think both offenses match up well with the opposing defenses, especiay given Carolina's defensive injuries. Jared Allen hasn't been great this year, but he's been useful, and has allowed Carolina to have good rotation of pass rushers. If he can't go, it may really make 2nd half fatigue a factor. Carolina needs Newton to have one of his best games.

53
by Jerry :: Tue, 01/19/2016 - 7:19pm

FWIW, I suggested adding some kind of clock component to DVOA a decade ago, and it hasn't happened. I don't know if Aaron couldn't find a workable framework, or if he just didn't think it was worth pursuing. Regardless, it hasn't happened, so we can't quantify just how devastating Kansas City's lethargic pace was.

57
by Bobman :: Wed, 01/20/2016 - 2:21am

You say "lethargic" and Tanier says "Hallucinogenic."

My 15 year-old and I were bitching and screaming at the TV and my wife's saying "oh yeah, I guess I see what you mean...."

Reid truly needs to hire a clock management coach--a decade late for Eagles fans, of course. Like a WWII bombardier taking control of the plane on the final bomb run, when the games gets to 4 minutes left, the team should be controlled by somebody else. Reid and the OC/DC can talk to this guy, but make no actual calls.

KC's GM, make this happen, for all of our sakes.

68
by Aaron Schatz :: Wed, 01/20/2016 - 3:26pm

Percentage of suggested changes to DVOA that don't get added because I test them out and can't find a way to improve the formula: 30%

Percentage of suggested changes to DVOA that don't get added because I don't have the time to try them out because of other responsibilities, even in the offseason: 69%

Percentage of suggested changes to DVOA that don't get added because I don't feel they are worth pursuing: 1%, or less. Probably less.

69
by Will Allen :: Wed, 01/20/2016 - 3:30pm

SLACKER!!!!!!

72
by Bobman :: Wed, 01/20/2016 - 5:59pm

Aaron missed a great opportunity to have his percentages not foot to 100, thereby causing at least ten conspiracy-theory comments (THAT is why my team is under-valued!), 50 pointing out the error, and another 25 noting that it was probably intentional to make us crazy. Which it would have been.

74
by Hurt Bones :: Wed, 01/20/2016 - 8:05pm

I was thinking this exactly. The sort of thing I might do. Though I might have gone to four decimal places like 1.0507% to make it look like I have lots of time to figure how much time I don't have. Aaron's just to sensible.

79
by Scott C :: Thu, 01/21/2016 - 1:23am

Right, or the next time the other team actually completes the comeback.

Going into turtle mode ineffectively IS A PROBLEM.

If they went turtle mode and had several 5 to 9 play drives that moved the ball, made first downs, ate up clock, but did not score they would not have had such horrible DVOA in the second half.

Not being able to protect a lead is a problem and a weakness.

81
by yayFootball :: Thu, 01/21/2016 - 2:29am

Your description of an 'effective turtle mode' that would supposedly not produce a bad DVOA is an exact description of the Panthers' offensive performance in the second half. Prior to the last drive, where all they had to do was kneel, every single drive produced at least one first down and ran between 5 and 9 plays. The problem with Carolina's second-half performance was the defense/special teams on Seattle's first two possessions. They were allowed to score those first two touchdowns way too quickly.

82
by Duff Soviet Union :: Thu, 01/21/2016 - 3:17am

They actually didn't have a horrible DVOA in the second half. Not on offense anyway. Their second half offense had DVOA above 0, albeit only likely due to opponent adjustments despite scoring 0 points in the half. This should make a lie of the whole "DVOA doesn't care about game context" stuff. Pretty much every team with a big lead tries to play the same way Carolina tried to play in the second half. A few long, slow drives that don't have to score points to be successful. And that's what DVOA says: That Carolina's "prevent offense" was actually ok. Their entire problem in the second half was defense. And again, their offensive gameplan in the second half was no different to what it was in the first half, it was just less successful. Yeah, they ran a lot on 1st and 10 in the second half. They did that in the first half as well, it's just that they went from gaining 5 yards a carry to gaining two yards per carry.

83
by Duff Soviet Union :: Thu, 01/21/2016 - 3:43am

I'll also say that I'd have more patience for this "DVOA doesn't care about context" argument if someone could show evidence that it was actually a problem with DVOA beyond this one game.

For instance, as others have pointed out, Carolina has blown or almost blown a few leads this year.

In the game against the Giants, where they blew the lead and then won on a last second field goal, they had a DVOA of +15.8% despite downward opponent adjustments.

In the game against the Colts, where they blew a 17 point 4th quarter lead and won in OT, they had a +12.1% DVOA despite strong downward opponent adjustments.

In the game against the Packers where they turned a 23 point 4th quarter lead into a situation where Green Bay had the ball with a chance to tie with 3:30 to go, they had a +30.1% DVOA with an upward opponent adjustment.

Punishing teams for blowing leads and not rewarding them for getting them in the first place is not an issue.

If a play by play breakdown says that over the course of the full game, Seattle slightly outplayed Carolina and were unlucky to lose, well maybe they did.

84
by yayFootball :: Thu, 01/21/2016 - 5:21am

First off, Carolina's offensive strategy definitely changed. I don't why first down play selection is being used as a barometer of aggressiveness. For a run-heavy team like the Panthers situations that are neutral to favorable for the offense are a poor place to look for evidence of conservatism because the normal and conservative strategies are closely aligned. You need to look at situations that would normally be passing situations. Twice, the Panthers called runs for Tolbert on third-and-six and third-and-seven. That is clear evidence of a shift to a conservative strategy.

Second, the issue isn't how Carolina's performance is rated in a vacuum, it's how their performance is rated relative to Seattle's. You seem to be suggesting that when DVOA's assessment disagrees with an impression formed from watching the entire game, DVOA should be believed. That's crazy, especially for a metric you don't know every detail about. Reducing the data from a game into a single number requires you to throw away information. DVOA may generally do a good job of capturing the important information, but that doesn't guarantee that it will accurately represent any particular game.

9
by Pen :: Mon, 01/18/2016 - 9:07pm

If you take Seattle's first half splits for playoff games started before 1pm and those started at 1PM in the Pete Carroll era you find quite the amazing stat:
Seattle has outscored their opponents in the first half of 1pm games 106-68.
Seattle has been outscored by their opponents in the first half before 1pm 0-91.

Yes ZERO points in 5 games before 1pm and 106 pts in 7 games starting 1pm or later. While their offense is shut down in the morning, their defense is giving up 34% more points.

I wonder what all the other west coast teams look like? Have to do a study. Wonder if DVOA is available for that?

EDIT: Just looked it up. From 2010 to present other west coast teams only have 3 10AM playoff games total between them. Seems the NFL likes hitting Seattle with that more than the rest of the west coast. SF had two. SF was able to score 37 points in two first halves before 1pm. Interestingly, those three games saw the west coast team curb stomp their east coast opponents in the second half 44-0.

SF 13-10 vs Carolina 10-0 second half
24-14 vs Atlanta 14-0 second half
SD 7-10 vs Cinci 20-0 second half

12
by HPaddict :: Mon, 01/18/2016 - 9:13pm

Since 2010 the other west coast teams only have four games which fit your criteria. The only exception was Arizona-Carolina last year.

26
by Pantherpryde :: Tue, 01/19/2016 - 5:15am

Awesome Stat!!!

31
by Travis :: Tue, 01/19/2016 - 9:55am

SF-Atlanta was a 3:05 Eastern (12:05 Pacific) start.

20
by Libertyguy2 :: Tue, 01/19/2016 - 12:42am

There's a story going around that the Seahawks players had too small cleats and looked like they were on roller skates. This was fixed at halftime and the turnabout began. Any thoughts about this?

34
by Hoodie_Sleeves :: Tue, 01/19/2016 - 11:12am

Sounds like shitty coaching and game-prep to me if its true. Its not like the field is unveiled at game time.

36
by Will Allen :: Tue, 01/19/2016 - 11:34am

The cleats on a lot of guys were swapped after the 1st possession, and they still had their a$$es handed to them for the balance of the half.

58
by Bobman :: Wed, 01/20/2016 - 2:26am

Even worse, when the camera zoomed in on one O player having his cleats changed, one of the announcers said the D or ST had changed pre-game, but the Offense had not. So it's not like they didn't know....

Everybody seemed to be slipping on that field--not sure what next week will bring. I read that it was the rain/cold conditions causing a slick surface (either the local Seattle paper or WSJ), but that looked like the old stadium in Philly (no seams and broken knees, but everybody falling down).

43
by WeaponX :: Tue, 01/19/2016 - 2:10pm

They had players trading out cleats immediately. Funny excuse story though!
Sometimes I even trip myself out.

21
by jwkelly :: Tue, 01/19/2016 - 1:28am

It makes me crazy when teams open the game in the wrong sized cleats. How hard is it to have a couple of the scrubs run around the field during warm-ups and make a lot of hard cuts?

38
by tuluse :: Tue, 01/19/2016 - 12:54pm

I don't think there's anything wrong with DVOA's breakdown of the Seahawks-Panthers game.

All it can measure is how the teams played. The Panthers got a few fluky breaks in the first half (a pick return TD and getting set up in great field position by another pick).

When a team is just gifted essentially 10 points like that, that's what DVOA see. The team was gifted some points. That doesn't make the Panthers bad, but it does mean they got some early luck.

Now if a team is literally gifted 10 points, they don't have play as well as the other team. So I think DVOA is right, the Seahawks slightly outplayed the Panthers. It didn't matter because they didn't out play them enough to overcome the slight bad luck they had. If they hadn't had that particular bad luck, the Panthers would have played differently anyways.

65
by TomC :: Wed, 01/20/2016 - 2:52pm

Totally disagree that those plays were fluky and that those 10 points were a gift. In my opinion, the most significant factor in both picks was pressure (caused by good DL play and lousy OL play), and the only thing remotely lucky was the fact that one return went for a TD.

66
by tuluse :: Wed, 01/20/2016 - 2:55pm

I didn't explain myself well. The Panthers forcing a couple picks was not fluky. The spot on the field they happened, and the near guaranteed points, was.

67
by Will Allen :: Wed, 01/20/2016 - 3:20pm

Well, not entirely. If your offense, defense, and special teams play in such a way that more of the game is concentrated at one extreme end of the field, compared to the norm, then the effects of turnovers become magnified, both positive and negative. That isn't random.

This stuff gets complicated, doesn't it?

42
by aceofsween :: Tue, 01/19/2016 - 1:40pm

I think it's pretty easy to see what the Panther's gameplan was in the second half. And frankly (as a Panthers fan), I wasn't pleased with it because throughout the season we saw how utterly ineffective it was (cough, Giants, cough cough).

That being said, there are some things you can't deny going into the second half. The Panthers were up by 5 possessions, 4 TDs and a field goal. So Seattle needed at least 5 possessions to win the game, meaning that under most circumstances the Panthers would have 4 (Seattle received the kick off in the 2nd half), for a total of 9 possessions. Therefore, the Panthers needed to force an 8 possession game.

To me, it is clear that is how the Panthers were approaching this half. With only 30 minutes left to play, 8 possessions comes out to an average of just over 3 minutes 45 seconds each possession, or about 7 and a half minutes each time Seattle gets the ball. Burn through that amount, and the game is over.

Now, this philosophy works. But it should have been abandoned after the first set of downs for the Panthers. At that point, Seattle took only 1:33 to score a TD, and the Panthers had only burned off 3:07. Now, you're looking at an average of 4 minutes 13 seconds for each possession... The Panthers ultimately stuck to their game plan though, which probably cemented a victory after a nearly 6 minute drive at the beginning of the 4th quarter.

Would I have liked to see the Panthers ice the game with another 8 minute TD drive in the 3rd quarter? Of course. It didn't happen though, and while I find that utterly frustrating, the philosophy behind why the Panthers won is sound.

It's just... incredibly frustrating.

47
by Eddo :: Tue, 01/19/2016 - 4:23pm

This is a really good analysis, I think.

70
by nickd46 :: Wed, 01/20/2016 - 5:19pm

Seconded, I think it's a really good analysis of time in the game... to the extent where I think there's a job opening for you on the Chiefs' staff.

Do you think the Panthers went for a time based approach rather than a yardage based approach to winning the game based on:

A) their experience from previous games and opposing teams mounting comebacks against them

B) the aura of the Seahawks being a team who could win a game from 31-0 down

C) both

?

75
by aceofsween :: Wed, 01/20/2016 - 8:08pm

Honestly, I think they just play the statistics. The only way Seattle wins is to consistently score every 7 minutes or so. The easiest way to stop that, in their opinion, was to employ the strategy they did.

My guess is that they wanted to cede 10 yards or so on the edges and clamp down in the red zone. Russel Wilson made a little magic happen on two drives that would have otherwise been field goals, so it looks worse on the Panthers end. In particular, the pass where not only does Wilson just barely get the ball off before being sacked for a 10+ yard loss, but Josh Norman just barely can't get enough reach on the ball to deflect the pass from Kearse.

Holding them to a field goal was probably the goal all along.

51
by ChrisS :: Tue, 01/19/2016 - 6:17pm

I agree that slowing the game was the Panthers second half strategy. Limiting the game to 8 possessions would not be an automatic winning strategy if Seattle had any defensive TD's

73
by TomC :: Wed, 01/20/2016 - 6:34pm

It was only a 5-possession game if Seattle kicks the XP or misses the 2-pointer at least twice. I still maintain going for 2 every time would have increased Seattle's win %.

76
by aceofsween :: Wed, 01/20/2016 - 8:09pm

That's a given.

45
by big10freak :: Tue, 01/19/2016 - 3:48pm

In this case single game DVOA matches what was seen in the GB/AZ game. GB played really well on defense, mediocre on offense and awful on special teams (thanks Ginger Wolverine!!)

52
by poplar cove :: Tue, 01/19/2016 - 6:46pm

When discussing the yards per play numbers in articles like this can't someone take the time (only takes less than 3 minutes a game to do) and throw out all the kneel downs, red zone plays and garbage time plays where the defense is obviously playing more against the clock (e.g.- the Steelers last drive and Seahawks also).

Also I wish the whole yards per play talk would slow down in football analytics word. IMO it's a crap stat and is being looked at way too much. There's so many flaws in YPP like when a team runs the ball quite a bit more than their opponent in a game (I'd say at least 8 more times) than if that's the case they're going to most likely have worse yards per play numbers. We seen this in 3 of the 4 games this weekend and the differential was big in all of them.

YPP numbers are a lot less accurate when it comes to showing the winning team when the rushing attempt numbers have a large differential. This is the reason the Panthers numbers aren't as good as you'd expect in their game as they ran the ball 29 more times than Seattle. I'm sure there were a few out there but I don't recall seeing another game with that big of a differential between the teams this entire NFL season.

54
by Cythammer :: Tue, 01/19/2016 - 8:52pm

Watching the second half (I actually missed the first hour, only turning the game on when it was already 24-0 Panthers driving with nine minutes left in the second… Never been so surprised by a score.) I was wondering if this might end up happening. Seattle did do a great job with their first two drives, but after that you have to take everything they accomplished on both offense and defense with a grain of salt. They were moving the ball well, but not nearly as quickly as they should've been, which in context meant they weren't really doing well at all. Carolina was the opposite. DVOA probably doesn't love their 5 minute long drive at the beginning of the fourth quarter, but in the context of the game it was a very valuable, effective drive that came close to killing off Seattle's chances. I think is obviously an example of the Panthers' strategy, which was designed to maximize their chances of winning the game, warping their DVOA as a side effect.

62
by Kyndynos :: Wed, 01/20/2016 - 11:26am

On the playoff odds screen, it says that NE only gets a two percent boost for the return of Julian Edelman. Is that really all he's worth to the Patriot offense? Watching games this year, the eyeball test says that New England's offense tanked after he got hurt (although the O-line also started falling at this time). Can somebody help me understand this?

64
by BJR :: Wed, 01/20/2016 - 12:56pm

I don't know how the number is arrived at, but it sounds acceptable to me. Remember, Edelman only missed 7 weeks so it isn't a straight case of with or without him; DVOA already accounts for the fact he was there for a good chunk of the season.

But the main issue is not wanting to go too far down the line of attributing (what I assume are) subjective judgments on a players worth. We all assume, fairly safely, the Patriots offence is better with Edelman, but it becomes dangerous, within supposedly objective analysis, to guess by how much. What if he is only 70% healthy? And what about the Patriots other injuries? Frankly I'd rather the numbers weren't tinkered with at all (other than maybe for QBs) and we were left to apply our own critical thinking on how to interpret them.

77
by Edstein :: Wed, 01/20/2016 - 11:51pm

I know that I'm a bit late to the party here, and a lot of this has been covered to some extent already, but ...

I don't know why some people take the DVOA/DYAR ratings as anything like a definitive assessment of which team is - better - than another. Not this or any other analytical approach can do that, and never could. It's only about the odds (not the Vegas variety but statistical) of one team likely to come out ahead at the end of the game based on an arbitrarily (subjectively) chosen assortment of metrics.

The Packers were clearly the better team for much of the game, and even with adding a couple of phenomenal plays on top of that at crucial moments, were nevertheless outdone by a few phenomenal plays on the other side. But it took the other side doing enough things to remain close to begin with.

In the Seahawks-Panthers game, both halves of the game were measured the same, and "DVOA'd" accordingly, even though the two situations were entirely different. The DVOA/DYAR actually rewarded the Seahawks for being the faster rabbit in the second half more than it penalized them for being the much more correspondingly slower rabbit in the first half. Quite a bit much of an "A for effort" grading gone overboard, there.

When the situational game was even, both teams off the same starting blocks, the Panthers obliterated the Seahawks. When the Seahawks were put into position of having to catch up from such a great distance, in the second half, it is no wonder that their effort in that attempt might have graded a bit higher than the Panthers' attempt to keep them at bay in the same situational element, the latter using the clock and (not correspondingly/appropriately graded) important offensive first downs successfully to that end.

And all that before we even get to the mind numbing obsession with yards qua yards as a measure of offensive 'efficiency.' Yet again, the Panthers offense came out ahead in points per play, but are trumped by the obsessional 'metric' of yards per play by the analytic crowd.

It wobbles the mind.