Writers of Pro Football Prospectus 2008

Most Recent FO Features

MasonTre14.jpg

» Word of Muth: Rams' Big Win

Ben Muth's tape analysis shows that St. Louis rookie Greg Robinson is still a work in progress -- as is Seattle's defense.

14 Nov 2006

Week 11 DVOA Ratings

by Aaron Schatz

Here's a look at this week's DVOA ratings. This week's commentary is now available at FOXSports.com.

One of the most common requests from FO readers is a Playoff Odds Report, similar to the one you'll find over at our sister site, BaseballProspectus.com. Well, a FO reader named Mike Harris has put together one himself, running the season 15,000 times and listing how often each team gets playoff spots 1-6. Right now you'll find that report on his website.

I've never done a playoff odds report like this because I've never had the time to really work at making it as accurate as possible. Mike's algorithm for figuring out the winner of a future game based on DVOA is completely untested, and doesn't include any variables except DVOA and home field. There's nothing that considers injuries, for example, and we're using WEIGHTED DVOA instead of the midseason projections. So don't take this as gospel, or as "FO says that the Jaguars have a 41 percent chance of making the playoffs.") Nonetheless, this gives us something to work with, and if I can find the time, I'll go through past years and help Mike improve the algorithm. We've already made one substantial change, adding in all of the head-to-head and division/conference record tiebreakers instead of just flipping coins to get playoff spots when there are ties. (Mike pointed out to me in e-mail that adding correct tiebreakers really helps the New York Giants.)

I know that a few readers are creating sortable graphs and so forth and linking them in the comments. I don't have the time to go through all the comments but if you are one of those people, e-mail me and tell me about what you are doing. Perhaps we can help improve those tables as well, or incorporate them into our display of stats here at FO.

Opponent adjustments are now at full strength. Offense, defense, special teams are updated; individual pages and adjusted line yards will be updated later tonight.

Remember that you can always use the keyword "DVOA" to access the latest DVOA commentary at FOXSports.com.

* * * * *

These are the Football Outsiders team efficiency ratings through Week 10 of 2006, measured by our proprietary Defense-adjusted Value Over Average (DVOA) system that breaks down every single play and compares a team's performance to the league average based on situation in order to determine value over average. (Explained further here.)

OFFENSE and DEFENSE DVOA are adjusted based on strength of opponent as well as to consider all fumbles, kept or lost, as equal value. SPECIAL TEAMS DVOA is adjusted for type of stadium (warm, cold, dome, Denver/Mexico City) and week of season.

To save people some time, please use the zlionsfan template for all complaints:

<team> is clearly ranked <too high/too low> because <reason unrelated to DVOA>. <subjective ranking system> is way better than this. <unrelated team-supporting or -denigrating comment, preferably with poor spelling and/or chat-acceptable spelling>


TEAM TOTAL
DVOA
LAST
WEEK
W-L WEIGHTED
DVOA
RANK OFFENSE
DVOA
OFF.
RANK
DEFENSE
DVOA
DEF.
RANK
SPECIAL
DVOA
S.T.
RANK
1 PHI 38.5% 2 5-4 35.3% 2 22.0% 3 -16.7% 4 -0.1% 17
2 CHI 36.4% 1 8-1 35.5% 1 -1.5% 18 -27.8% 1 10.2% 1
3 NYG 30.9% 4 6-3 31.0% 3 16.0% 4 -14.5% 5 0.4% 14
4 SD 29.9% 3 7-2 28.4% 4 25.7% 2 0.0% 17 4.2% 5
5 IND 22.5% 7 9-0 22.6% 5 34.4% 1 8.7% 24 -3.2% 29
6 DAL 20.3% 10 5-4 20.9% 6 9.8% 8 -10.9% 7 -0.4% 20
7 JAC 19.3% 6 5-4 16.6% 8 -7.0% 22 -25.4% 2 0.9% 13
8 BAL 19.1% 5 7-2 17.1% 7 -8.5% 23 -22.6% 3 5.0% 3
9 NE 12.4% 8 6-3 13.9% 9 8.5% 10 -2.6% 13 1.3% 11
10 DEN 10.4% 9 7-2 12.8% 10 4.4% 12 -5.7% 12 0.4% 15
11 KC 8.7% 11 5-4 10.0% 11 0.3% 17 -6.6% 11 1.8% 9
12 PIT 8.3% 12 3-6 9.1% 12 4.2% 13 -10.7% 8 -6.6% 31
13 NO 6.1% 13 6-3 5.4% 13 12.3% 6 9.4% 25 3.1% 7
14 CIN 4.5% 15 4-5 1.9% 14 15.2% 5 12.2% 26 1.4% 10
15 STL 0.9% 14 4-5 0.1% 16 10.7% 7 8.4% 23 -1.5% 24
16 CAR -0.9% 16 5-4 1.0% 15 3.1% 14 0.4% 18 -3.5% 30
TEAM TOTAL
DVOA
LAST
WEEK
W-L WEIGHTED
DVOA
RANK OFFENSE
DVOA
OFF.
RANK
DEFENSE
DVOA
DEF.
RANK
SPECIAL
DVOA
S.T.
RANK
17 GB -2.6% 20 4-5 -1.2% 17 2.5% 15 3.8% 20 -1.3% 23
18 MIN -5.7% 17 4-5 -5.9% 20 -18.7% 28 -12.9% 6 0.1% 16
19 WAS -5.8% 19 3-6 -5.1% 19 8.9% 9 15.9% 29 1.2% 12
20 MIA -6.8% 21 3-6 -5.0% 18 -15.6% 26 -10.5% 9 -1.7% 25
21 ATL -7.4% 18 5-4 -9.5% 22 -6.1% 21 -1.3% 15 -2.6% 28
22 SEA -8.6% 23 6-3 -8.8% 21 -5.5% 20 5.3% 21 2.2% 8
23 NYJ -13.4% 26 5-4 -13.8% 24 0.6% 16 17.6% 30 3.6% 6
24 CLE -13.6% 24 3-6 -12.2% 23 -22.1% 29 -2.5% 14 6.0% 2
25 BUF -15.5% 22 3-6 -18.1% 26 -14.8% 25 5.7% 22 4.9% 4
26 HOU -16.1% 28 3-6 -14.9% 25 4.8% 11 18.9% 32 -2.0% 27
27 DET -19.8% 25 2-7 -18.9% 27 -4.3% 19 15.2% 28 -0.3% 19
28 TB -25.3% 27 2-7 -24.2% 28 -23.8% 30 -0.4% 16 -1.9% 26
29 OAK -28.3% 30 2-7 -26.0% 29 -36.9% 32 -9.6% 10 -1.0% 22
30 TEN -29.6% 32 2-7 -26.6% 30 -16.4% 27 13.0% 27 -0.2% 18
31 SF -29.7% 29 4-5 -32.2% 31 -10.4% 24 18.6% 31 -0.7% 21
32 ARI -36.0% 31 1-8 -36.1% 32 -24.2% 31 3.2% 19 -8.6% 32

  • NON-ADJ VOA shows what the rating looks like without adjustments for strength of schedule, luck recovering fumbles, or weather and altitude on special teams.
  • ESTIMATED WINS uses a statistic known as "Forest Index" that emphasizes consistency as well as DVOA in the most important specific situations: red zone defense, first quarter offense, and performance in the second half when the score is close.  It then projects a number of wins adjusted to a league-average schedule and a league-average rate of recovering fumbles.  Teams that have had their bye week are projected as if they had played one game per week.
  • PAST SCHEDULE lists average DVOA of opponents played this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
  • FUTURE SCHEDULE lists average DVOA of opponents still left to play this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
  • VARIANCE measures the statistical variance of the team's weekly DVOA performance.  Teams are ranked from least consistent (#1, highest variance) to most consistent (#32, smallest variance).


TEAM TOTAL
DVOA
W-L NON-ADJ
TOT VOA
ESTIM.
WINS
RANK PAST
SCHED
RANK FUTURE
SCHED
RANK VAR. RANK
1 PHI 38.5% 5-4 30.4% 7.7 1 -0.3% 14 4.3% 8 5.1% 31
2 CHI 36.4% 8-1 48.2% 6.6 3 -10.4% 32 -7.6% 28 31.6% 2
3 NYG 30.9% 6-3 25.6% 6.6 4 6.1% 5 6.8% 4 8.9% 24
4 SD 29.9% 7-2 38.1% 6.7 2 -6.6% 31 -8.4% 29 5.6% 29
5 IND 22.5% 9-0 24.4% 6.6 5 -0.8% 16 4.3% 7 7.9% 26
6 DAL 20.3% 5-4 22.8% 5.5 8 -0.6% 15 6.5% 5 11.4% 17
7 JAC 19.3% 5-4 21.2% 5.3 9 4.3% 6 3.2% 12 36.7% 1
8 BAL 19.1% 7-2 31.6% 5.9 6 -5.2% 29 -1.0% 20 12.8% 14
9 NE 12.4% 6-3 19.1% 5.5 7 -3.7% 25 -2.7% 22 11.0% 19
10 DEN 10.4% 7-2 10.3% 5.3 10 0.2% 12 -0.2% 19 16.1% 10
11 KC 8.7% 5-4 9.0% 5.1 11 -3.0% 23 1.2% 15 30.6% 3
12 PIT 8.3% 3-6 2.6% 4.7 14 4.0% 7 -1.5% 21 15.4% 11
13 NO 6.1% 6-3 7.3% 4.9 12 -1.0% 17 1.7% 14 11.4% 16
14 CIN 4.5% 4-5 -0.2% 4.6 16 3.5% 9 3.5% 10 7.3% 27
15 STL 0.9% 4-5 11.6% 4.7 13 -6.2% 30 -10.0% 31 9.3% 23
16 CAR -0.9% 5-4 0.3% 4.3 18 -3.0% 24 10.2% 2 6.9% 28
TEAM TOTAL
DVOA
W-L NON-ADJ
TOT VOA
ESTIM.
WINS
RANK PAST
SCHED
RANK FUTURE
SCHED
RANK VAR. RANK
17 GB -2.6% 4-5 -4.7% 4.4 17 -0.2% 13 -4.0% 24 11.9% 15
18 MIN -5.7% 4-5 -2.5% 4.6 15 -3.8% 26 -5.9% 27 11.3% 18
19 WAS -5.8% 3-6 -14.2% 4.2 19 11.2% 2 6.1% 6 10.2% 21
20 MIA -6.8% 3-6 -2.1% 3.6 24 -1.3% 20 0.0% 18 9.7% 22
21 ATL -7.4% 5-4 3.3% 4.1 20 -5.1% 28 7.4% 3 25.6% 4
22 SEA -8.6% 6-3 -7.4% 3.9 21 -1.3% 21 -11.8% 32 10.3% 20
23 NYJ -13.4% 5-4 -9.1% 3.7 22 -2.1% 22 -5.5% 26 17.1% 9
24 CLE -13.6% 3-6 -16.8% 3.0 27 2.2% 11 1.1% 16 2.7% 32
25 BUF -15.5% 3-6 -15.8% 3.5 25 3.9% 8 0.3% 17 17.4% 8
26 HOU -16.1% 3-6 -26.1% 3.7 23 12.1% 1 -9.4% 30 13.0% 13
27 DET -19.8% 2-7 -13.6% 2.9 28 -5.1% 27 2.6% 13 5.4% 30
28 TB -25.3% 2-7 -35.7% 2.5 31 7.5% 4 4.2% 9 8.0% 25
29 OAK -28.3% 2-7 -34.6% 2.5 32 -1.1% 18 3.3% 11 13.3% 12
30 TEN -29.6% 2-7 -29.7% 2.6 30 7.7% 3 13.2% 1 19.0% 7
31 SF -29.7% 4-5 -31.6% 3.0 26 2.7% 10 -5.5% 25 24.3% 5
32 ARI -36.0% 1-8 -31.7% 2.8 29 -1.2% 19 -3.2% 23 23.1% 6

Posted by: Aaron Schatz on 14 Nov 2006

316 comments, Last at 18 Nov 2006, 11:50pm by B

Comments

301
by Andrew (not verified) :: Fri, 11/17/2006 - 1:42pm

That ball could have bounced in any direction, and it happened to bounce into Buckhalter’s hands.

Not true. Brown was turned towards the ball, and so was facing Buckhalter. Since he had essentially caught the ball, as opposed to it simply bouncing off his helmet or the ground, it was highly likely by basic physics that the ball would come out in the direction it went in, as opposed to suddenly developing significant sideways momentum or bouncing over Brown's head since the momentum of the ball going to Brown's hands and body would not fail to stop the forward movement of the ball.

It would surprise me if EVERY TEAM didn’t tell unblocked offensive players to try to get downfield and make blocks on a given play.

I'm sure they tell them to make blocks. What is not so clear is if they tell them to run to the ball, rather than running to any unblocked defender. We certainly don't see evidence of it in the games from many teams. I've noticed the Eagles, Bears, Jets, and Steelers do run to the ball.

Also that play in the 2005 playoffs involved a large number of players in one concentrated area (it was a play close to the end zone to start with, if i remember correctly). Not really the same type of thing at all.

Not really. It involved Mitchell, Smith, and perhaps 3 defenders.

The Buckhalter play involved Brown, Buckhalter, and 2 defenders.

The play area was similar as one was near the sideline and another near the back of the endzone.

302
by Andrew (not verified) :: Fri, 11/17/2006 - 1:44pm

Irishfan #300:

A year or two ago, it was noted that Mike Martz's St. Louis teams won significantly more games than would be expected. I think they were the most exceptional in this regard.

303
by Wanker79 (not verified) :: Fri, 11/17/2006 - 1:50pm

Andrew, you and me have both been fervently defending the Eagles by bringing up their ridiculously bad luck this year. So with all the times you've brought up their bad luck, defending something like that as being skill just seems disingenuous.

304
by Eddo (not verified) :: Fri, 11/17/2006 - 2:15pm

Thanks for the info, stan. Obviously there is good technique for recovering a fumble, and I guess I just needed proof that despite that, fumble recoveries can still be attributed mostly to luck.

Re: the Buckhalter play.
While yes, because it is a reception, it should count towards DVOA, I find an intersting argument can be made. The generally accepted argument (which I agree with) is that Bryant's 62 yarder tells nothing about the quality of the Eagles' play. However, it could also be argued that the crazy Buckhalter reception tells nothing about the quality of the Redskins' play. I find it hard to believe that it is anything but a "random" occurence, yet it will end up hurting the Redskins in the same way it helps the Eagles. Unfortunately, I cannot think of a way to compensate for plays like that without subjectively discounting it from the play-by-play stats. And as soon as subjectivity is brought into DVOA, the entire purpose has been defeated. To me, it seems like it's an unavoidable Catch-22.

305
by Ray (not verified) :: Fri, 11/17/2006 - 2:16pm

The used-to-be fumble that's now ruled a tipped pass to Buckhalter was COMPLETE luck. There's no way you can consider it anything else. Arguing that they're coached to run to the ball is irrelevant because it all depends on where the ball bounces. What if Buckhalter had 'run to the ball' a half second quicker and was engaged in a block already when the ball popped out? He would have been doing the same thing, but the ball wouldn't have gone to him for the TD.

Like Wanker says, you can't argue all the bad luck from earlier in the season without being able to admit when we finally do get good luck.

306
by Andrew (not verified) :: Fri, 11/17/2006 - 3:04pm

Andrew, you and me have both been fervently defending the Eagles by bringing up their ridiculously bad luck this year. So with all the times you’ve brought up their bad luck, defending something like that as being skill just seems disingenuous.

I never said Buckhalter catching it wasn't good luck. That was really good luck. Him being in the area to make a play was not luck though. That was good playing.

Back to the Giants game, it was lucky for the Giants that the ball bounced wildly into the endzone, and that Michael Lewis could not corral it. It was not lucky for Tim Carter to be in the area for the recovery, that was him hustling after a fumble.

Back to Buckhalter, he deserves as much credit for making a play as a defender catching that ball for an interception would.

Ray #305:

The used-to-be fumble that’s now ruled a tipped pass to Buckhalter was COMPLETE luck. There’s no way you can consider it anything else. Arguing that they’re coached to run to the ball is irrelevant because it all depends on where the ball bounces.

The laws of conservation of momentum dictate that an almost caught ball cannot bounce randomly. This particular pass was not tipped - it bounced backwards and up out of Brown's arms. And in that vein, it is worth noting that tipped passes either continue forward, or go up in the air - they don't go left or right much, so its wrong to say that their trajectory depends on the bounce. Similarly, balls bouncing off a receiver's body and arms tend to go back in the direction they came from.

307
by Wanker79 (not verified) :: Fri, 11/17/2006 - 3:07pm

Re: 304

Everyone was saying the exact same thing after that game where Kyle Boller threw like 3 tipped TDs. Everyone knew that it didn't say anything predictive about Baltimore, but you can't just ignore a few instances just because you feel like it. Either you go strickly off the play-by-play and live with the occational fluke tipped-reception, or you watch every play of every game and eliminate all of them. And since the second option just isn't reasonable, I guess we're just gonna have to live with it.

308
by Eddo (not verified) :: Fri, 11/17/2006 - 3:17pm

Re: 307
I couldn't agree more. And that's where you run into the limits of any statistical model. Too many DVOA-haters fail to realize that you can't judge the system as an absolute because of this simple fact. You have to take the good (how DVOA can show that a team plays really well or poorly despite the game's outcome) with the bad (the way certain factors--weather, tipped passes, funky bounces--can make DVOA look crazy).

309
by chris clark (not verified) :: Fri, 11/17/2006 - 4:22pm

re 287:

Yes, there is a difference between intangibles and luck (and unaccounted for tangibles and any other factors). However, all of the above should be captured in variance (or std. dev.). The variance of each team, plus their DVOAs should allow one to compute a probability distribution of the expected outcome of a game between two teams.

Such an outcome will not account for specific matchups, but it should predict the actual scores of games with associated error bars.

There are other measures (skew and Kurtosis?) which would allow one to know if a team's stats are off by one bad or good game or are just generally inconsistent. Of course, just looking at the data can often tell one that without measures. However, I have noticed that Aaron's notes have been pointing out that some measures have moved in counter-intuitive fashion (i.e. one shouldn't trust eyeballing the data too much). That is teams have gotten improved DVOA numbers when just barely winning against bad teams (and vice versa), because the play during the game hasn't necessarily followed the score nor the conventional wisdom about which teams are strong (and in which areas).

310
by Wanker79 (not verified) :: Fri, 11/17/2006 - 4:32pm

Back to the Giants game, it was lucky for the Giants that the ball bounced wildly into the endzone, and that Michael Lewis could not corral it. It was not lucky for Tim Carter to be in the area for the recovery, that was him hustling after a fumble.

Ok, I'll bye that logic. But that's an awfully fine line you're walking.

311
by chris clark (not verified) :: Fri, 11/17/2006 - 5:03pm

re: 308

Moreover, some factors (altitude and cold weather stadiums) are already factored into DVOA.

In those regards, I think Aaron's experience last year with the Colts is most instructive. For a while (1 or 2 weeks), Aaron adjusted the Colts position in the listed rankings to be more reflective of their W/L record; they were undefeated at the time. Note that he left the ranking the same, so he didn't destory the data, just altered its presentation. Doing so made the DVOA rankings more like other "power" rankings. However, as I recall, it didn't "work out", because it distorted what DVOA says and he stopped. (If I were Pat, here I would post a pointer to the page where Aaron explained that.)

One of the worries of adjusting the DVOA measures to account for teams that perform outside the norms, is that one begins to introduce "superstitious" factors into the rankings. For example, some teams may play better in domes, or on turf, or in the cold. I would love to see those statistics. They could inform my intuition. I would be less excited to see them part of the base DVOA calculations, they would seem to introduce some subjectivity into the measure. I definitely don't want to see meausres like "Brett Farve as QB in a dome against a team after a bye" adjustments in DVOA. (Ok, if each NFL team played 500 games/year and we had 5 years worth of data, maybe, but not with our small data set.)

So, we live with what DVOA can tell us, which is how teams do in playing the field position game (and that's under stating what DVOA says--i.e. it tells us more than that). And, I'm sure someone will say it tells us way more than that. All I know, is that is doesn't tell us everything, and that I still want to read other points of view and allow them to inform me also.

312
by Eddo (not verified) :: Fri, 11/17/2006 - 7:13pm

Re: 311
All I know, is that is doesn’t tell us everything, and that I still want to read other points of view and allow them to inform me also.
Bingo. I think a lot of the people who argue so heatedly on both sides need to think about your statement, combined with the fact that no one from FO has ever said that DVOA should be the ultimate judgement tool, just one judgement tool.
By the way, it's nice to see a sports-related message board with so few flamers and total idiots.

313
by Andrew (not verified) :: Fri, 11/17/2006 - 11:31pm

But that’s an awfully fine line you’re walking

Hustling to the ball is not luck, that is persistence and skill.

Getting the ball when you are hustling to the ball is luck. You can run all you want towards the play or towards a potential fumble, but that doesn't mean you will get it. Exhibit #1 - Michael Lewis hustled to the ball against the Giants, but Carter ended up with it.

314
by Bobman (not verified) :: Sat, 11/18/2006 - 3:39am

Hey, this has been fun reading back and forth on DVOA's merits.

What I want to know is why is Polaris pissed? He's in second place, right?

Kuato, I'm with you, but I am more worried about Philly. Dal has stupendous balance, but I really don't think they can keep up offensivesly. (Plus there's Vandy who is reliable, except when it counts the most.) Philly, however....

Actually, I wouldn't mind a nice safe loss to an NFC team to settle things down. no more 16-0 talk or inflated heads, etc., but not a loss that might hurt in tie-breaker terms.

315
by Oh, Mathematics (not verified) :: Sat, 11/18/2006 - 7:05pm

Andrew - Not only am I an Eagles fan, but I am also a physicist! While conservation of momentum certainly does dictate that the ball continue to fly in the direction of Buckhalter's hands once it has left Brown's, it doesn't really say anything about at what point in Brown's bobbling he might lose it, or what part of the ball he hits last. He didn't appear to have too much control over the ball, and in which direction it left his hands was completely luck. It's MUCH different than the NFC championship game, where the ball was popped up in the air, allowing Mitchell to adjust to the ball and move towards it. The fact that it popped straight up, and not in any other direction still makes the outcome of the play largely the result of luck, even if it was a heads-up play on Mitchell's part. I didn't see any evidence of Buckhalter adjusting to the ball or even breaking stride, which to me says that the play was mostly luck even after the ball left Brown's hands.

316
by B (not verified) :: Sat, 11/18/2006 - 11:50pm

314: I thought Polaris was a Seattle fan, which might explained why he's so upset.