Writers of Pro Football Prospectus 2008

Most Recent FO Features

houstonsanders.jpg

» Film Room: Justin Houston and Emmanuel Sanders

Cian Fahey breaks down one of the league's best defensive players, and the receiver who has made Denver even more dangerous.

10 Jan 2011

Week 18 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 1-4, and Weeks 5-10 are somewhat discounted.
  • Teams which did not play in the wild card round are treated as if they had a bye week. (That includes both the 20 non-playoff teams and the four teams with byes.)

Unlike last year, we have not made any adjustments based on teams sitting starters in Week 17, as there was very little of that going on this season.

The playoff odds report is updated through the wild card games, and you will find DVOA matchup pages for the four second round games on the FO Premium page. Please note that the playoff odds posted earlier this afternoon were based on some DVOA ratings with errors, so the new playoff odds are a bit different (and correct).

* * * * *

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.

TEAM WEI.
DVOA
LAST
WEEK
W-L WEI OFF
DVOA
OFF.
RANK
WEI DEF
DVOA
DEF.
RANK
WEI S.T.
DVOA
S.T.
RANK
1 NE 55.8% 1 14-2 49.8% 1 -1.0% 10 4.9% 6
2 PIT 40.6% 2 12-4 23.9% 3 -17.1% 1 -0.4% 18
3 BAL 30.4% 3 13-4 9.1% 12 -13.4% 3 7.9% 3
4 GB 27.9% 4 11-6 13.4% 8 -14.8% 2 -0.3% 17
5 SD 21.8% 5 9-7 20.6% 4 -3.1% 7 -2.0% 26
6 PHI 18.1% 6 10-7 17.6% 6 4.1% 16 4.5% 9
7 CHI 17.1% 11 11-5 -3.0% 21 -11.0% 4 9.1% 2
8 ATL 16.2% 9 13-3 8.5% 13 3.3% 14 11.1% 1
9 OAK 15.7% 12 8-8 6.0% 16 -4.7% 6 4.9% 7
10 NYJ 14.4% 7 12-5 -0.3% 19 -9.5% 5 5.3% 5
11 TB 12.0% 14 10-6 18.7% 5 5.7% 19 -1.0% 21
12 NO 8.9% 8 11-6 10.8% 11 1.0% 11 -0.9% 20
13 HOU 8.9% 13 6-10 23.9% 2 13.5% 29 -1.6% 24
14 NYG 8.3% 10 10-6 11.0% 10 -1.7% 9 -4.4% 31
15 DET 3.1% 16 6-10 8.2% 14 5.8% 20 0.7% 15
16 IND 0.4% 18 10-7 14.2% 7 8.1% 26 -5.7% 32
17 MIA -0.4% 15 7-9 -1.0% 20 -2.7% 8 -2.1% 27
18 TEN -1.9% 17 6-10 -3.5% 22 4.0% 15 5.7% 4
19 JAC -2.0% 19 8-8 12.8% 9 17.9% 30 3.1% 11
20 CIN -5.7% 21 4-12 7.4% 15 9.8% 27 -3.3% 29
21 CLE -6.4% 22 5-11 -3.8% 23 5.3% 18 2.7% 13
22 SF -9.5% 20 6-10 -3.8% 24 4.4% 17 -1.2% 22
23 DAL -10.2% 23 6-10 -5.2% 25 6.8% 22 1.8% 14
24 KC -11.9% 25 10-7 3.1% 18 11.0% 28 -4.0% 30
25 DEN -13.2% 24 4-12 5.7% 17 18.5% 31 -0.4% 19
26 BUF -16.5% 27 4-12 -12.1% 27 2.9% 12 -1.6% 25
27 MIN -17.0% 28 6-10 -12.6% 29 3.0% 13 -1.4% 23
28 STL -17.5% 26 7-9 -14.2% 30 7.7% 25 4.4% 10
29 WAS -22.1% 29 6-10 -12.4% 28 7.6% 24 -2.1% 28
30 SEA -29.0% 31 8-9 -9.1% 26 22.8% 32 2.8% 12
31 CAR -31.4% 30 2-14 -25.2% 31 6.2% 21 0.0% 16
32 ARI -33.2% 32 5-11 -30.7% 32 7.2% 23 4.7% 8

Here are the one-game DVOA ratings for the first round of the playoffs. The special teams ratings may surprise you, because some clear failures were balanced by quieter successes. The Eagles lost significant value for David Akers' missed field goals, but did very well with kickoffs, punts, and kick returns. The Saints lose value because of their silly short-kicks in the first half, but get a lot of value from Thomas Morstead's punts. And yes, the DVOA system does think the Colts outplayed the Jets in their ultra-close Saturday night contest.


DVOA (with opponent adjustments)
TEAM TOT OFF DEF ST
SEA 33% 50% 27% 10%
NO -11% 23% 40% 7%
NYJ -2% -4% 0% 1%
IND 23% 19% -3% 1%
BAL 92% 15% -78% 0%
KC -74% -61% 13% -1%
GB 36% 23% -11% 1%
PHI 16% 19% 5% 2%
VOAf (no opponent adjustments)
TEAM TOT OFF DEF ST
SEA 16% 43% 36% 10%
NO 15% 38% 30% 7%
NYJ -3% 0% 5% 1%
IND 12% 9% -2% 1%
BAL 92% 18% -74% 0%
KC -90% -72% 17% -1%
GB 22% 20% 0% 1%
PHI -5% 5% 12% 2%

Posted by: Aaron Schatz on 10 Jan 2011

89 comments, Last at 14 Jan 2011, 10:52am by Mr Shush

Comments

1
by MJK :: Mon, 01/10/2011 - 3:45pm

So if I understand correctly, for the single game table, VOAf tells us who outplayed who, and DVOA tells us who SHOULD have outplayed who.

So it tells us Seattle and NO essentially played to a draw, but that is impressive for Seattle and pathetic for NO. No surprise.

Indy slightly outplayed the Jets, but that's more impressive for Indy because DVOA expected them to do worse vs the Jets defense. (They still lost, though).

Baltimore crushed KC, which was not surprising.

And GB solidly outplayed Philly, which was impressive for GB, but Philly actually played better than might have been expected.

And the only game that was won by "luck" (i.e. team that played more poorly on average won) was IND-NYJ.

Did I get all that right?

21
by Kulko :: Mon, 01/10/2011 - 5:55pm

Thats how I read it to.

36
by Tom Gower :: Mon, 01/10/2011 - 10:15pm

I'd put it a little differently.

VOAf tells us how Team A and Team B performed just this past weekend, taking out the effect of fumble luck (the f).

DVOA tells us how Team A and Team B performed this past weekend, taking into account in Team A's rating that it was playing Team B and taking into account in Team B's rating that it was playing Team A.

Using NO-SEA as an example, VOAf tells us the two teams played pretty equivalent games, both excellent offensively and poor defensively.

When we look at DVOA, though, we see that the Saints' offensive performance was less impressive than it appeared, because the Seahawks' defense isn't very good, while the Seahawks' defensive performance was better than we thought it was, because the Saints' offense is good. A similar story is told on the other side of the ball, in that the Seahawks' offensive performance was better than we thought it was and the Saints' defensive performance was worse. What looked like and was an even game was actually a superior performance by the Seahawks.

Keep in mind that the VOAf to DVOA adjustments depend on the quality of the units they were competing against. It's straightforward in SEA-NO because the Seahawks' units were both below-average and the Saints' above-average. Compare IND-NYJ: both the Colts offense and the Jets defense are better by DVOA terms than VOAf because both those units played an above-average opposing unit.

42
by nat :: Tue, 01/11/2011 - 9:12am

No. No. No. I say again: No.

You cannot look at the total DVOA for that game and say that the Seahawks played a far superior game. DVOA says, in effect, that if you assume the Saints played their usual quality of game, the Seahawks would have had to play an excellent game to accomplish what they did. It says that the Seahawks did far better than an average team would have been expected to do against the Saints.

But when you ask the question "who played better that day" it doesn't matter one whit what an average team would have done on an average day, and you can't assume that either team is playing at their usual level. What matters is what actually happens on each play. You don't get any credit for exceeding expectations in the NFL.

This is the DVOA paradox in action. You can compare the offenses with DVOA, because you can do so assuming the defenses played with their usual quality. You can compare the defenses with DVOA, because you can assume the offenses played at their usual level. But if you try to combine these two numbers, the result is meaningless, because you are assuming that both teams are playing at their usual quality on both sides of the ball, which implies that the Saints played a vastly superior game, only to conclude the exact opposite.

The rule of thumb: Use DVOA to compare teams or units that didn't face each other. Use VOA or VOAf to compare two teams in the same game.

43
by Noah of Arkadia :: Tue, 01/11/2011 - 9:44am

All in all, I still think MJK got it right.

48
by nat :: Tue, 01/11/2011 - 10:23am

Yes, MJK got it right. Seattle and New Orleans played an equal game, which is well below expectations for the Saints, and would have been below expectations for an average team, too.

There's an interesting subtlety in the VOAf numbers for that game. VOAf says that Seattle's offense got slightly better results per play than the Saints, while at the same time insisting that the Saint's defense got slightly better results per play than the Seahawks. That seems odd, but it has an explanation.

In any game, there are a small number of plays that get different handling for evaluating the offense and defense facing each other. Some penalties are counted against the offense, but not for the defense, for example. The result is that different plays get included in the total, and different average results are used for judging the offense and defense on the same play. The same play can be considered a better than average result for both teams.

This game falls in that "too close to call" zone where per play success doesn't explain the result. In a close game, getting good results at the right time and in the right sequence can make all the difference.

Ain't football fun?

58
by Tom Gower :: Tue, 01/11/2011 - 1:48pm

1. I don't think we disagree substantively as much as you seem to think we do.
2. I think "who played better?" is a vague and imprecise question that VOAf and DVOA can usefully give us two different answers for, and that DVOA is analytically more useful for substantially all purposes other than "ignoring the effect of opponent quality, which team put up a higher level of performance in that particular game."

59
by nat :: Tue, 01/11/2011 - 2:30pm

I think we do disagree.

"Who played better?" is a clear question in English. It means which one team outplayed the other in absolute terms on that day according to the rules of the game of football, and in this case, according to their value per play compared to average in the same situations.

If I go into the boxing ring with the heavyweight champion of the world, and I last 6 rounds before getting beaten into unconsciousness, my "DVOA" is going to be much, much, much higher than his. I far exceeded expectations, perhaps even exceeding what an average heavyweight fighter would do. He on the other hand, was terrible, had a epically bad "DVOA", and should consider giving up the sport and selling waffle irons.

But he still fought better than I did.

65
by tuluse :: Tue, 01/11/2011 - 4:36pm

You are ignoring the degree of difficulty factor. If you go 6 rounds rounds against the heavy weight champion, he fought better than you, but it suggests you are a much better fighter than previously thought. Because what you did is very difficult. Which is what DVOA is measuring.

72
by nat :: Tue, 01/11/2011 - 5:38pm

I'm not ignoring degree of difficulty. It's just not important to the question "Who played better in that game?"

There's no contradiction between these statements:

"The Saints and Seahawks played almost equally well in that game."

"The Seahawks seem to be better than we thought, given how well they played against the Saints."

"The Saints seem to be worse than we thought, given how well they played against the Seahawks."

The first sentence is about who played better, and is best measured by VOAf when teams play each other. The other two are about expectations for future games, and are best measured with DVOA.

74
by Tom Gower :: Tue, 01/11/2011 - 7:03pm

Yeah, we do disagree to an extent, but I still think we disagree less than you seem to think we do. In any event, I think we both see where we disagree, and I'm inclined to leave it at that unless somebody sees a need for me to expound further.

78
by nat :: Tue, 01/11/2011 - 10:30pm

Nah, that's okay. We've covered it enough.

45
by Boo-urns (not verified) :: Tue, 01/11/2011 - 10:07am

Unless I'm missing something, Philly slightly underperformed their DVOA, which was 18.1% for the season. But yes, I basically read it as: VOAf tells us how well a team performed, in a vacuum, in one game, whereas DVOA tells us how well that team performed, given opponent adjustments. So KC's offense played horribly, which we saw, and that's reflected in their VOAf, but because they were playing the Ravens D, that actually means they played slightly less poorly.

So basically, DVOA is telling us that both the Packers and Seahawks are looking legit with robust DVOAs that are even better than the actual results we saw on the field. Of course, what DVOA seems to ignore is that the Saints traveled thousands of miles to a cold venue that is perhaps the best HFA in the league, and played a classic trap game.

So I will say definitively, the Bears have NOTHING to worry about. The Seahawks away are a far different team than at home, and the "disrespected underdog" card has now been safely discarded, on behalf of all future 7-9 divisional winners.

Meanwhile, I'd say the Falcons are going to have a dogfight on their hands, thanks to Mike Vick's mishaps.

46
by Boo-urns (not verified) :: Tue, 01/11/2011 - 10:10am

To correct myself, although DVOA doesn't tell us these things, the playoff odds report does. 92.8% chance of beating Seattle sounds about right.

2
by jimmyp (not verified) :: Mon, 01/10/2011 - 3:58pm

CHI went from 11.5% and 11th to 17.1% and 7th. How did this happen? that seems a little large of a change just due to opponent adjustments. I note that CHI lost 17-3 to NYG in week 4; was that game included in last week's Wtd DVOA?

6
by Eddo :: Mon, 01/10/2011 - 4:12pm

The game against the Giants fell off completely, and games against the Panthers, Seahawks, and Redskins are all part of the discounted time period (though I'm not quite sure if they're discounted any more than they were last week).

8
by DisplacedPackerFan :: Mon, 01/10/2011 - 4:20pm

I thought the discounting (since they discount 6 weeks worth of games before dropping them off) was something like Week 5 - 15%, Week 6 - 30%, Week 7 - 45%, Week 8 - 60%, Week 9 - 75%, Week 10 - 90%. That would make sense for what the metric of weighted DVOA is trying to represent, which is basically "How has the team played the last 8 weeks with how they were playing before that, counted less and less." Though perhaps every game does get the same discounting, but I thought it was a sliding scale.

3
by Ryan (not verified) :: Mon, 01/10/2011 - 4:00pm

NY vs. CHI was one of the worst games of the season for any team. That was Cutler's Sacktacular game of the year.

37
by Mr Shush :: Mon, 01/10/2011 - 11:53pm

You mean the one where he was concussed early on on an illegal helmet-to-helmet hit that wasn't called, then unsurprisingly didn't have a clue what was going on until he was pulled (at the half?)? Yeah, I'm going to say that's a situation where most teams would struggle.

40
by Athelas :: Tue, 01/11/2011 - 9:09am

Touchy!
I don't think he blamed Cutler--just clearly reminding us which game it was.

69
by chisox24 :: Tue, 01/11/2011 - 5:22pm

Actually, considering that Cutler's name is the only one mentioned, I pretty sure he is implying ownership, and thusly, blame.

70
by tuluse :: Tue, 01/11/2011 - 5:24pm

I read it as just reminding us that Cutler got murdered in that game, but tone is hard to convey in writing.

81
by Mr Shush :: Wed, 01/12/2011 - 3:59am

Yeah, sorry, didn't mean to come of so snappish. What I was really trying to say was that while yes, that was a horrible game for both Cutler and the Bears, it wasn't a terribly meaningful one, because it was the result of an injury. It would be a bit like holding the Rusty Smith game against the Titans.

84
by rdy4thefiesta :: Wed, 01/12/2011 - 1:03pm

I don't remember the illegal hit. My recollection of the game was that all the hits were clean. I went back and looked at a compilation of all the sacks on youtube and none of them looked dirty. The only one that was even close was the first sack, where Tuck jumps on Cutler as he's being sacked by someone else, but it looks like Tuck lands on Cutler's shoulder. I wasn't able to find any non-sack hits on Cutler, so maybe it happened there, but I don't remember it.

Regardless, you have a valid point. That game, and the Carolina game that followed, are almost irrelevant to how the Bears are playing now, just like the Patriots' loss to the Browns, or the Steelers' 4 games without Ben. I like weighted DVOA late in the season better than DVOA for this reason, and sometimes wish it was weighted even further toward more recent games. but there has to be a cutoff somewhere.

85
by Andrew Potter :: Wed, 01/12/2011 - 1:30pm

I don't think the Patriots' loss to the Browns is at all irrelevant to how they're playing just now.

86
by Mr Shush :: Wed, 01/12/2011 - 3:21pm

It was the third sack of the game, by Umenyiora, at the top of the second quarter. The play starts at about 0:32 in this NFL.com highlights video. It's not that obvious at full speed, but in slow motion (someone posted a much better video breakdown in a thread here shortly after the game in question) it's pretty clear that there's helmet-to-helmet contact. After the play, Cutler took an age to stand up, and then tried to walk back to the wrong sideline.

88
by Anonymous Joe (not verified) :: Fri, 01/14/2011 - 1:59am

Puleeze - nothing helmet to helmet about that one.

89
by Mr Shush :: Fri, 01/14/2011 - 10:52am

This video frame-by-frames the collision from a more helpful angle (around 6:20 in). It's tough to see in real time, but it was most definitely helmet-to-helmet. If you wind the video a bit, you can see Cutler struggling to get up, and discussion of the fact that this was the concussion play.

4
by RichC (not verified) :: Mon, 01/10/2011 - 4:06pm

Its amazing how much better the NE defense gets when you start pulling away weeks from the beginning of the year.

5
by Eddo :: Mon, 01/10/2011 - 4:10pm

Interesting (and promising) to see the Bears make such a big jump (11.5% to 17.1%).

It's not just the offense losing the Giants game completely and having the Panthers, Seahawks, and Redskins games discounted; while their weighted offensive DVOA did increase from -5.3% to -3.0%, the Bears' weighted defensive DVOA actually improved the same amount (-8.7% to -11.0%). And their weighted special teams DVOA improved from 8.2% to 9.1%.

I hope this trend continues.

7
by Eddo :: Mon, 01/10/2011 - 4:13pm

And wow - the playoff odds report gives the Bears a 92.1% chance of beating the Seahawks. That has to be one of the highest figures ever for a playoff game.

38
by Danish Denver-Fan :: Tue, 01/11/2011 - 4:04am

Never mind a divisional playoff game (ie, quarterfinals).

9
by cfn_ms :: Mon, 01/10/2011 - 4:23pm

Seattle is clearly ranked too low because they won a playoff game. Anyone who has them over Denver and Buffalo at least is way better than this. ROFL at playoff win, +450 vs nawleans = FUN!

10
by Dej (not verified) :: Mon, 01/10/2011 - 4:24pm

Does anyone know if the weather adjustments are directional for Special Teams DVOA? In the Philadelphia game it was pretty clear that kickoffs with the wind were likely to be touchbacks, and against the wind were not, and the same logic probably applies to punts and kicks, so I was wondering whether that gets factored in?

Following on, can someone enlighten me as to how Philadelphia "outplayed" Green Bay on Special Teams. Philadelphia missed two makeable field goals, averaged a good 40 yards on punts, and mediocre kick return averages of 18.3 yards on kicks and 7.5 yards on punts. Green Bay had decent to slightly below average numbers of 37 yards on punts, 15 yards on kickoff returns and 9 yards on punt returns. How those numbers come out worse than missing 2 makeable field goals is beyond me. Does Green Bay get penalized for Brandon Underwood kicking the ball and Philly recovering on a punt return? Overall I'm confused by the Special Teams numbers.

11
by Eddo :: Mon, 01/10/2011 - 4:38pm

The Underwood play is extremely negative. Had the punt simply been downed, the Packers would have had first-and-ten at their own 41. That's worth 1.51 expected points(*). Instead, the Eagles had first-and-ten at the Packer 41. That's worth 2.60 points, a 4.11-point swing!

By comparison, Akers's missed field goals were attempted from the Packer 23 and the Packer 16, worth 1.57 and 1.96 expected points, respectively, for a total net loss of 3.53 points. Which is less than the amount of expected points the punt muff cost the Packers.

(*) Based on Brian Burke's figures from Advanced NFL Stats.

EDIT: Also, I'm not clear on the rules, but Aikman kept insinuating that, since Underwood was blocked into the ball, it should not have been a fumble. I didn't think how the receiving team player touched the ball was relevant - that the onus was on the receiving team to stay the f away from the ball so that you can't get blocked into it. What is the actual rule? Travis?

14
by Travis :: Mon, 01/10/2011 - 4:56pm

Aikman was right, although it could be argued that Underwood was doing the blocking at the time (IIRC, he appeared to me to be going forward when he made contact with the ball, although he looked like he was passively being driven backwards before that). If Underwood was indeed blocking at the time, it was a valid fumble; if he was merely blocked into it, no.

Rules:

3-15-3-Note 1: "If a player is pushed or blocked into any kick or fumble or into a backward pass after it has struck the ground, and if such pushing or blocking is the primary factor that sends such a loose ball in touch, the impetus is by the pusher or blocker, and the pushed (blocked) player will not be considered to have touched the ball. See 9-2-4."

9-2-4: "There is no distinction between a player touching a ball or being touched by it, but a player is not considered to have touched the ball if he is blocked into it by an opponent, provided he is in a passive position and not blocking. A player who is engaged with and blocking his opponent when he contacts the ball is deemed to have touched the ball."

Note also that: (1) as far as I can tell, such a ruling is not reviewable, and (2) the Santonio Holmes "fumble" (had it actually occurred while he was jostling with various Colts) would not have been.

50
by rk (not verified) :: Tue, 01/11/2011 - 10:38am

If you watch the complete replay, it's pretty clear that GB got hosed. I can't tell who the Philly player is, but he is clearly blocking Underwood with the intent of getting him to touch the ball. The ball is rolling harmlessly on the ground, and the Eagle starts pushing Underwood while holding him pretty fiercely from about 3-4 yards away. Underwood avoids the ball, so the Eagle pulls him forward (very obviously holding now), and he steps on the ball as he falls on top of the Eagle. By this time, the play has been essentially over for several seconds. It looks like Underwood is blocking because he's moving forward, but that is only because he is being dragged in that direction.

49
by The Powers That Be :: Tue, 01/11/2011 - 10:29am

"By comparison, Akers's missed field goals were attempted from the Packer 23 and the Packer 16, worth 1.57 and 1.96 expected points, respectively, for a total net loss of 3.53 points. Which is less than the amount of expected points the punt muff cost the Packers."

Thanks for the link to the expected point tables - that's good stuff.

However, those aren't the relevant numbers in the paragraph above. We're not interested in the expected points associated with all 4th downs at this particular yard line. We're interested in the expected points for all field goal attempts from this yard line. I'm fairly certain the expected points associated with 41- and 34-yard FG attempts are much greater than 1.57 and 1.96 points.

25
by Jerry :: Mon, 01/10/2011 - 7:28pm

Weather adjustments are much less granular than you're looking for. From http://www.footballoutsiders.com/info/methods#specialteams:

"Kickoffs, punts, and field goals are then adjusted based on weather and altitude. It will surprise no one to learn that it is easier to kick the ball in Denver or a dome than it is to kick the ball in Buffalo in December. Because we do not yet have enough data to tailor our adjustments specifically to each stadium, each one is assigned to one of four categories: Cold, Warm, Dome, and Denver/Mexico. An additional adjustment drops the value of field goals in Florida and raises the value of punts in San Francisco."

12
by jhoff (not verified) :: Mon, 01/10/2011 - 4:40pm

i'm not shocked that dvoa has the colts out playing the jets (though i think its a good example of a team executing a winning performance that dvoa can't capture), but on special teams i'm a little confused. the jets averaged 10 more yards on kick returns, and weatherford placed his punts deep repeatedly. field position was probably the difference in the game.

16
by JSA (not verified) :: Mon, 01/10/2011 - 5:25pm

Weatherford actually placed his punts a little too deep, I think he had 4 touchbacks and only one punt inside the 20. That's unusualy for him. On the entire regular seasons he had only 4 tocuhbacks, and 42 punts inside the 20

26
by Dave :: Mon, 01/10/2011 - 7:48pm

Two of those were pretty much nothing but luck away from being downed at the one. I have a hard time blaming him so much as to make the ST grade for the teams even. It's pretty clear which team was better on special teams in that game.

Actually, several times a year I look at the total DVOA table and think "man, giving Special Teams 33% of the total score just seems so WRONG." But then the Colts go and do something worthy of that 32 next to their name and it kills their season and I can only shake my head.

28
by coboney :: Mon, 01/10/2011 - 8:16pm

Actually Special Teams isn't given 33%. The way it works (as I recall) is that DVOA judges it as 3/7 Offense, 3/7 Defense, 1/7 Special Teams but the reduction is already done to special teams before it gets here hence why its numbers are often so much closer to 0 then Offense and Defense.

29
by Dave :: Mon, 01/10/2011 - 8:41pm

I really should've known that.

31
by jebmak :: Mon, 01/10/2011 - 9:13pm

We're all disappointed in you Dave.

33
by Eddo :: Mon, 01/10/2011 - 9:30pm

Well, everyone except for The Anti-Dave.

44
by Noah of Arkadia :: Tue, 01/11/2011 - 9:51am

Or Coach Dave.

73
by leviramsey (not verified) :: Tue, 01/11/2011 - 5:50pm

The 3/7 offense, 3/7 defense, 1/7 special teams is a general rule of thumb derived from the fact that offensive DVOAs greater in absolute value than 30 (i.e. greater than +30 or less than -30), defensive DVOAs greater in absolute value than 30, and special teams DVOAs greater than 10 are all about equally rare. There's nothing special in the DVOA calculation that makes ST 1/3 of offense or defense: the reason why special teams DVOA is typically about a third of offense or defense is due to there being fewer special teams plays to consider.

Thinking a bit more about it, DVOA implies that on a per-play basis, special teams is on average more important than offense or defense: I suspect that in most games the number of special teams plays is well under one-seventh the total number of plays: about the only way to get that large a number of special teams snaps is to have a ton of return TDs (fumble, INT, or kick) and/or a lot of 1-5 play TD drives by the offenses (since a TD results in two special teams plays being run).

66
by zlionsfan :: Tue, 01/11/2011 - 4:42pm

Put another way, two of those punts were touchbacks unless the cover team managed to intervene, which they did not. It's true that DVOA doesn't distinguish between a punt that hits at the 15 and rolls into the end zone and a punt that lands in the stands behind the end zone, but that's one thing about modern punting vs. coffin-corner punting: a punt headed out of bounds along the sideline needs only to be aimed well and does not require coverage help.

This probably goes back to what we mean when we talk about DVOA. It isn't really Weatherford's DVOA; when we mention his DVOA, it's "the DVOA of the Jets' punting unit when Steve Weatherford is punting", and in that context, there's probably no question about why those punts counted against "him".

19
by biebs (not verified) :: Mon, 01/10/2011 - 5:45pm

You are wrong about Weatherford. He had 5 punts, and 4 punts that were touchbacks.

One was a great punt that should have been downed inside the 5, I don't remember the others.

It was actually a shockingly bad game for him based on his season. He had an incredible season with about 50% of his punts inside the 20 and 20 punts inside the 10.

13
by bravehoptoad :: Mon, 01/10/2011 - 4:45pm

Look at Oakland there at #9. I knew they'd won a few big games, but did not see that one coming.

39
by Danish Denver-Fan :: Tue, 01/11/2011 - 4:09am

Usualy I don't closeread the DVOA-table, and OAKs rating has completely snuck up on me. This changes my perception of the Raiders completely.

47
by Mr Shush :: Tue, 01/11/2011 - 10:16am

I don't think it should. I've watched quite a lot of the Raiders this year, for some reason, and I'm pretty damn sure DVOA is wrong about them. Maybe because SD managed to have two of its meltdown games against them. I don't know. But I'm pretty sure they're an average/below-average team that played a soft schedule. They didn't beat a single team outside of the West divisions. They lost two games against NFC West opponents, for goodness' sakes.

55
by bravehoptoad :: Tue, 01/11/2011 - 11:51am

You mean the same NFC West with the .643 winning percentage in the playoffs the past four years? You mean the same NFC West with the longest active streak of winning a playoff game each year of any division? You mean the same NFC West that's sent each of its teams to the Super Bowl since...lemme see...1995? ...and three of them since 2002?

Glory in our power, baby. Glory in it.

80
by Mr Shush :: Wed, 01/12/2011 - 3:56am

Comrade, the Bureau of Tractor Production needs you.

83
by bravehoptoad :: Wed, 01/12/2011 - 12:51pm

Tractors! With our tractors, we will bury you.

63
by tuluse :: Tue, 01/11/2011 - 4:30pm

This is weighted DVOA remember, not total. They're still 21st by total.

79
by Mr Shush :: Wed, 01/12/2011 - 3:55am

Doh. Yes, fair enough. Easy to see how a couple of decent performances against fading/imploding divisional rivals late in the season could have a disproportionate effect there.

15
by thankyoujimcaldwell (not verified) :: Mon, 01/10/2011 - 5:15pm

I've never really delved into how DVOA is built up to a single metric (I think it's been good as a predictive metric). With that admission made, I'm somewhat puzzled by the Jets-Colts DVOA. (Yes, I'm a Jets fan: so, homer alert, though I I'm trying to be unbiased here.)

The Jets outgained the Colts (by 40 yards), had more first downs, more plays, and a longer time of possession. They had one more penalty for 5 more yards (neither team was penalized much). They had a much stronger kick-off return game (other special team metrics look even, and that seems to fit the game as I remember it). Each team had one sack. The Colts did not turn the ball over, and the Jets had one turnover (in a safe part of the field at the end of the half). They also had one fumble that they recovered (yes, fumble-luck, though that particular type of "fumble" -- ball between hands the whole time, and essentially regained before ball or player goes to ground -- does not strike me as a 50/50 ball by any stretch of the imagination). And the Jets were playing away.

Turnovers matter, needless to say, but the damage from actual turnovers (as opposed to fumbles made but then regained) should be captured, I would think, but yardage, plays run, and the score itself. So it seems to me that the only "unrealized" advantage the Colts had was the Edwards fumble. Did that really swamp the other things that went the Jets' way? This was a close game, but the VOA and DVOA gaps seem quite large. Aside from being lucky to face Jim Caldwell, this didn't strike me as win won through luck rather than earned on the field. VOA seems to disagree.

18
by B :: Mon, 01/10/2011 - 5:43pm

I think you answered your own question. The Jets slightly out gained the Colts, but they had a turnover and another fumble which they recovered. Which is why the Colts got the slightly higher rating.

27
by Dave :: Mon, 01/10/2011 - 7:50pm

I'm not sure I'd call it slight.

That said, it surprised me too. Figured the third down conversion rate would make the offenses much closer, and with so much of the Colt offensive output coming on one play (albeit a scoring play) I definitely would've guessed the values incorrectly. By a lot.

32
by JSA (not verified) :: Mon, 01/10/2011 - 9:19pm

Part of it probably is that the Colts had success on 1st and 2nd downs to put them in a lot of 3rd and 1 situations. Those are positive plays in the eyes of DVOA, that are expected to lead to first downs.

52
by Purds :: Tue, 01/11/2011 - 10:46am

That sounds like a pretty good explanation. I was with jimcaldwell up there at first, but this explanation makes a lot of sense. FO guys, is that right?

57
by thankyoujimcaldwell (not verified) :: Tue, 01/11/2011 - 1:29pm

Yes, if their 3rd downs were much shorter (and I assume you're right that they were -- haven't looked) that could very well explain it. (Though to play devil's advocate, the Jets got 8 of their 23 first downs without needing a 3rd down at all; the Colts got only 3 of their 16 first downs without needing a 3rd down.)

51
by rk (not verified) :: Tue, 01/11/2011 - 10:43am

I think you're glossing over Sanchez's interception too much. Yes, it didn't lead to a good opportunity for the Colts, but it absolutely blew a big opportunity for the Jets. He essentially threw away 3 points in what ended up being a 1-point game.

17
by Waverly :: Mon, 01/10/2011 - 5:26pm

It's not often that one sees a yellow highlighted row in the DVOA table so close to the bottom. In fact, I didn't see it at all at first, until I scrolled down.

20
by N8 (not verified) :: Mon, 01/10/2011 - 5:54pm

Wow. Suddenly the Packers S.T. approaches mediocrity!

22
by scottybsun (not verified) :: Mon, 01/10/2011 - 6:12pm

I'll trade falling 3 spots in the rankings in return for a road playoff win. thank you very much.

23
by Andrew Potter :: Mon, 01/10/2011 - 7:06pm

Rankings aren't really the point; the actual DVOA numbers are far more useful. The difference between Baltimore and Green Bay (1 ranking spot; 2.5% DVOA) is not the same as the difference between Baltimore and Pittsburgh (1 ranking spot; 10.2% DVOA). This is one of the minor quibbles with reading DVOA as a ranking system instead of how DeltaWhiskey presents the information in grouped batches: a team can improve in actual rating, but drop in ranking because those around them improved - or decline in rating, but go up in the ranking because those around them declined more.

Which raises a point: assuming the Last Week column exists to indicate improvement or decline from the previous week, I really wish they would put in last week's rating instead of last week's ranking. If it's not meant to be a "power rankings" system, including last week's rank really isn't all that helpful.

30
by jebmak :: Mon, 01/10/2011 - 9:13pm

I really wish they would put in last week's rating instead of last week's ranking

Wow, what a great idea! I fully support this plan.

35
by Arkaein :: Mon, 01/10/2011 - 10:05pm

Even better, they could do it like the regular season updates to the Playoff Odds report, which puts the change since the previous week next to the new value, with improvements in green and declines in red, making it easy to identify trends.

41
by Athelas :: Tue, 01/11/2011 - 9:11am

I vote for this too.

24
by TomC :: Mon, 01/10/2011 - 7:19pm

46.1% DVOA advantage + home field + 92.1% playoff odds + 10-point Vegas favorites = Bears are DOOMED.

If it's not 35-0 Bears in the 1st quarter, I'm going to be a nervous wreck.

62
by ticttocs (not verified) :: Tue, 01/11/2011 - 4:25pm

Yeah, none of that nonsense can measure Carroll's "Win Forever" attitude. (heart of a champion, baby!)

34
by jonnyblazin :: Mon, 01/10/2011 - 9:43pm

Looks like the Ravens had by far their best game of the season yesterday. Their DVOA for every game this year has been between 10% and 60% except for the two games vs. Cincy.

75
by Bjorn Nittmo (not verified) :: Tue, 01/11/2011 - 7:41pm

Where on this site do they post game-by-game DVOA for each team?

76
by tuluse :: Tue, 01/11/2011 - 7:47pm

I think it's available in the premium database.

53
by DeltaWhiskey :: Tue, 01/11/2011 - 11:16am

Below are the means and SD for the WEI DVOA

WEI DVOA
AVG 2.24%
SD 20.77%

WEI OFFDVOA
AVG 3.99%
SD 15.79%

WEI DEF DVOA
AVG 3.05%
SD 9.37%

WEI ST DVOA
AVG 1.29%
SD 4.08%

Game by Game Breakdown:

BAL V PET

BAL
TOT GOOD
OFF AVG
DEF ELITE
ST ELITE

PIT
TOT ELITE
OFF GOOD
DEF ELITE
ST AVG

COMMENTS:
To compare BAL's OFF to PIT's DEF and vice versa, I've standardized the scores. This now puts both OFF and DEF in the same metric - number of SD's from the mean.
BAL OFF (0.32) v PIT DEF (-2.15)
BAL DEF (-1.76) v PIT OFF (1.26)
BAL ST (1.62) v PIT ST(-0.41)
This suggests a pretty significant disparity in the quality of BAL's OFF v. PIT's DEF and suggests that BAL's DEF is slightly better than PIT's offense. The ST disparity is huge.

GB V ATL

GB
TOT GOOD
OFF GOOD
DEF ELITE
ST AVG

ATL
TOT GOOD
OFF AVG
DEF AVG
ST ELITE
COMMENTS: Again, standardizing scores

GB OFF (0.60) v ATL DEF (0.03)
GB DEF (-1.91) v ATL OFF (0.29)
GB ST (-0.39) v ATL ST(2.40)

GB OFF appears to be a little better than ATL DEF, the same cannot be said for the other side of the ball, as GB's DEF is quite a bit better than ATL's. The ST disparity is huger in this game.

SEA V. CHI

SEA
TOT HORRID
OFF BAD
DEF HORRID
ST AVG

CHI
TOT GOOD
OFF AVG
DEF GOOD
ST ELITE

COMMENTS: Again, standardizing scores

SEA OFF (-0.83) v CHI DEF (-1.50)
SEA DEF (2.11) v CHI OFF (-0.44)
SEA ST (0.37) v CHI ST(1.91)

Simply put, CHI is way better than SEA; however, I'm sure if I'd done this last week, I'd have said the same thing about NO v. SEA. The only area that appears remotely competitive is ST.

NYJ V NE

NYJ
TOT GOOD
OFF AVG
DEF GOOD
ST GOOD

NE
TOT ELITE
OFF ELITE
DEF AVG
ST GOOD

COMMENTS: Again, standardizing scores

NYJ OFF (-0.27) v NE DEF (-0.43)
NYJ DEF (-1.43) v NE OFF (2.90)
NYJ ST (0.98) v NE ST(0.88)

NE's defense appears to be slightly better than NYJ's OFF and not surprisingly NE's OFF is considerably better than the NYJ DEF. S.T. are evenly matched.

54
by nat :: Tue, 01/11/2011 - 11:39am

I don't think it's valid to standardize the offense and defense scores using standard deviations. There is no a priori reason to expect defensive quality to have the same distribution as offensive quality.

64
by tuluse :: Tue, 01/11/2011 - 4:31pm

I was going to post something similar to this. DVOA is expressed as a percentage for a reason.

68
by DeltaWhiskey :: Tue, 01/11/2011 - 5:15pm

Yeah, but sometimes you ahve to do a lot of work to discover and come to a better understanding, and since stats is not a primary part of my day-to-day work, it helps maintain learning and understanding.

The more I play with DVOA the more I understand it strengths and weaknesses.

71
by nat :: Tue, 01/11/2011 - 5:25pm

I love the stuff you're doing. I sometimes disagree or suggest improvements. I get that you're defining Elite etc in terms of standard deviation, which means that an Elite offense is much tougher to face than an Elite defense in today's NFL. I even see why you started with standard deviation in comparing offenses to defenses, although I respectfully disagree with that way of doing things.

Keep at it.

77
by DeltaWhiskey :: Tue, 01/11/2011 - 10:25pm

Thanks, I've been enjoying it as well and as I've mentioned I've learned a lot about DVOA and its limitations.

As you can see in the post below, I generally discredited my whole methodology for comparing using SD's. I thought I'd come up w/ a better number, but instead discovered that DVOA was the right metric for comparing.

56
by DeltaWhiskey :: Tue, 01/11/2011 - 12:37pm

Perhaps you're correct.

The first question is what is the relationship of DEF mean to OFF mean. If two teams with both an OFF = 3.99% and a DEF = 3.05% each played each other an infinite number of times, they should have .500 record each and should avg around 21 points per game scored and allowed. The problem is if the means are not functionally equal (i.e. mean OFF is greater or less than mean DEF) then there is a problem; however, theyt appear to be pretty damn close to equivalent, as a DEF DVOA of 3.05% can be expected to give up 22.2 points per game, while an OFF DVAO is associated with scoring 22.1 points per game.

Next, what does a SD mean. On OFFENSE, a team that is one SD (15.79%) above the mean (3.99%+15.79%) can be expected to score 25.9 points per game for a difference from the mean of 3.8 pts/game. On DEF a team that is one SD (9.37%) above the mean (3.05%-9.37%) can be expected to give up 19.8 points per game for a difference of 2.5 pts/game.

Using Raw numbers: a team whose OFF is 10.00% above the mean on offense will be expected to score 24.6 points per game for a difference 0f 2.5 points expected points from the mean. A team whose DEF is 10.00% above the mean can be expected to surrender 19.7 points per game for a difference from the mean of 2.5 pts/game.

I think you're right, but there's got to be some use for all my hard work. Oh well...thanks for giving my brain a work out today.

60
by TomC :: Tue, 01/11/2011 - 2:51pm

Anyone who watches the Falcons regularly care to comment on the meteoric rise of Atlanta's special teams (#22 last year, now #1 in WDVOA)? The biggest contributor seems to be kickoff distance + coverage, in which they're actually ahead of Baltimore, despite Cundiff's off-the-charts touchback numbers. Are they forcing a fumble every third kickoff or something?

61
by Ezra Johnson :: Tue, 01/11/2011 - 3:34pm

Is there any adjustment for dome and/or turf kickers vs. outdoor/grass kickers? It seems there should be.

82
by DeltaWhiskey :: Wed, 01/12/2011 - 9:31am

Not sure what to make of "How to please a woman" add being replaced by Tony Dungy's "How to be a friend."

87
by TomC :: Wed, 01/12/2011 - 4:18pm

Me neither, but I do know that the perfect amalgam of those two creepy and/or borderline tasteless ads is Catholic Match Girl.