Writers of Pro Football Prospectus 2008

Most Recent FO Features


» Futures: Nick Chubb & Sony Michel

The Georgia Bullddogs' dynamic duo should be on NFL rosters at some point in the next 72 hours. Which will be the better pro? That depends on what kind of running back you're looking for.

28 Nov 2006

Week 13 DVOA Ratings

by Aaron Schatz

Here are this week's DVOA ratings. The commentary is now posted at FOXSports.com.

The commentary for Indianapolis this week talks about Pythagorean wins. Some people have asked me why I don't include this in the tables during the year, even though I include it in the tables after the season is over. The reason is that I have to cut and paste the points scored and allowed from elsewhere, which takes extra time. Anyway, since I mention Pythagorean wins, I figured I would run the top 12 for the curious folks. These are all the teams with at least 6.0 projected wins.

1) Bears, 9.5
2) Patriots, 8.7
3) Ravens, 8.5
4) Chargers, 8.4
5) Cowboys, 8.2
6) Jaguars, 7.7
7) Colts, 7.3
8) Bengals, 6.6
9) Broncos, 6.6
10) Saints, 6.4
11) Eagles, 6.2
12) Chiefs, 6.2

Yes, the Colts are seventh, and still might end up as the "luckiest" team since the AFL-NFL merger. After the Colts, the top teams outperforming the projection are Seattle and San Francisco, by 1.7 wins apiece.

Mike Harris did his Playoff Odds Report again. You'll find it here, and once again there are two versions: one regular, one with Philadelphia at 0% because of the Donovan McNabb injury.

Speaking of the Eagles, I've received a number of e-mails about that strange Correl Buckhalter touchdown from a couple of weeks ago. Yes, the NFL changed this from a Reggie Brown fumble and Buckhalter recovery to a Buckhalter catch and a passing touchdown for McNabb. From my contact at the NFL:

"The play was originally scored as a catch by Brown, a fumble caused by Springs, with Buckhalter recovering it for a touchdown. After reviewing the play we determined that Brown never established possession, so it was a catch for Buckhalter and a receiving touchdown. Senior officials said that if Buckhalter had not caught the ball in the air, it probably would have been ruled an incomplete pass."

As for how I am treating this in DVOA, I haven't changed anything yet because frankly I don't feel like doing anything at this point that makes Philadelphia's rating go up. At some point I probably will go in and turn it into a Buckhalter catch, like the official stats say.

The other play people are asking about is the New England double fumble on Sunday. I'm counting that as a catch and fumble by Ben Watson and just ignoring the Reche Caldwell recovery and second fumble. Sometimes it makes sense to simplify things like this for the sake of my own sanity.

Don't forget to check out the Football Outsiders swag shoppe -- now featuring ROBO-PUNTER shirts! They have our logo on the front, ROBO-PUNTER on the back with uniform number 1.

All of the team and individual JUST THE STATS pages are now updated through Week 12.

* * * * *

These are the Football Outsiders team efficiency ratings through Week 12 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.

WEIGHTED DVOA is based on a formula which discounts games more than eight weeks ago in order to get a more accurate picture of how teams are playing now. This is the formula used for the rankings at FOXSports.com.

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

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>

1 CHI 32.1% 1 9-2 30.6% 2 -3.5% 18 -27.8% 1 7.8% 1
2 SD 31.7% 2 9-2 30.0% 3 26.3% 2 -0.5% 16 4.9% 5
3 BAL 31.0% 5 9-2 30.7% 1 -0.4% 15 -25.6% 2 5.8% 3
4 DAL 28.6% 6 7-4 28.5% 4 15.1% 4 -14.3% 4 -0.7% 19
5 IND 23.9% 9 10-1 24.5% 5 33.2% 1 7.7% 26 -1.6% 25
6 NE 22.2% 8 8-3 23.8% 6 10.0% 8 -9.8% 7 2.4% 9
7 JAC 21.1% 4 6-5 21.8% 7 -2.4% 17 -23.3% 3 0.2% 15
8 PHI 19.9% 3 5-6 16.5% 9 13.2% 5 -9.6% 8 -2.9% 29
9 NYG 17.8% 7 6-5 17.7% 8 12.3% 7 -5.6% 12 -0.1% 16
10 CIN 11.7% 12 6-5 11.5% 10 17.6% 3 7.7% 25 1.8% 11
11 KC 10.7% 11 7-4 9.0% 11 4.4% 10 -4.0% 13 2.3% 10
12 NO 6.9% 15 7-4 6.6% 12 12.8% 6 7.2% 24 1.4% 12
13 CAR 3.9% 13 6-5 5.5% 13 -3.8% 19 -10.3% 6 -2.6% 27
14 DEN 3.5% 10 7-4 3.9% 14 0.9% 13 -2.9% 14 -0.2% 17
15 PIT -1.5% 14 4-7 -2.1% 16 -0.9% 16 -6.2% 11 -6.7% 32
16 MIA -3.4% 20 5-6 -1.6% 15 -13.6% 28 -11.6% 5 -1.5% 24
17 MIN -4.8% 17 5-6 -4.5% 17 -10.8% 26 -7.1% 10 -1.1% 21
18 STL -5.9% 16 5-6 -7.7% 19 3.4% 12 8.8% 27 -0.5% 18
19 WAS -6.0% 18 4-7 -9.5% 21 5.2% 9 11.5% 28 0.2% 14
20 NYJ -8.9% 23 6-5 -10.1% 22 0.7% 14 13.6% 29 4.0% 6
21 GB -10.2% 21 4-7 -8.8% 20 -4.8% 20 2.0% 17 -3.5% 30
22 BUF -10.6% 24 5-6 -13.1% 24 -9.7% 24 7.0% 23 6.1% 2
23 TEN -13.1% 27 4-7 -7.4% 18 -10.4% 25 5.2% 19 2.4% 8
24 SEA -14.6% 25 7-4 -14.9% 25 -12.3% 27 5.6% 20 3.3% 7
25 HOU -14.7% 26 3-8 -11.5% 23 4.4% 11 17.9% 31 -1.2% 22
26 CLE -14.8% 19 3-8 -15.4% 26 -22.5% 31 -2.3% 15 5.4% 4
27 ATL -15.6% 22 5-6 -18.2% 27 -8.1% 22 4.6% 18 -2.9% 28
28 SF -22.4% 29 5-6 -20.7% 28 -6.4% 21 14.6% 30 -1.4% 23
29 TB -25.1% 28 3-8 -24.8% 30 -18.5% 30 5.7% 21 -0.8% 20
30 OAK -25.4% 31 2-9 -23.3% 29 -31.7% 32 -8.1% 9 -1.7% 26
31 DET -28.8% 30 2-9 -29.7% 32 -9.1% 23 20.3% 32 0.6% 13
32 ARI -30.5% 32 2-9 -27.8% 31 -18.1% 29 5.9% 22 -6.5% 31

  • 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).

1 CHI 32.1% 9-2 42.1% 8.2 2 -8.6% 31 -15.0% 32 25.8% 3
2 SD 31.7% 9-2 37.2% 8.5 1 -4.7% 25 -8.3% 28 6.6% 32
3 BAL 31.0% 9-2 37.4% 8.2 3 -3.4% 23 -0.9% 18 16.1% 12
4 DAL 28.6% 7-4 30.0% 7.4 6 -0.8% 17 0.0% 17 16.2% 11
5 IND 23.9% 10-1 23.1% 7.8 5 5.4% 5 0.3% 15 13.2% 18
6 NE 22.2% 8-3 24.1% 7.2 7 1.3% 15 -7.8% 27 15.4% 15
7 JAC 21.1% 6-5 22.7% 6.7 8 3.8% 10 8.0% 6 33.5% 1
8 PHI 19.9% 5-6 12.7% 7.8 4 0.6% 16 5.7% 9 15.9% 13
9 NYG 17.8% 6-5 15.6% 6.7 9 3.3% 11 10.7% 2 13.8% 17
10 CIN 11.7% 6-5 8.4% 6.2 11 3.1% 13 6.3% 8 10.3% 22
11 KC 10.7% 7-4 12.3% 6.5 10 -4.8% 26 8.7% 4 28.1% 2
12 NO 6.9% 7-4 9.1% 6.1 12 -3.8% 24 4.4% 12 9.5% 23
13 CAR 3.9% 6-5 4.8% 5.5 15 -1.7% 19 5.5% 10 9.5% 24
14 DEN 3.5% 7-4 3.2% 5.5 14 5.2% 7 -4.8% 24 16.2% 10
15 PIT -1.5% 4-7 -6.5% 5.2 17 5.2% 6 1.3% 14 19.2% 8
16 MIA -3.4% 5-6 3.0% 4.9 19 -2.5% 22 9.5% 3 9.4% 25
17 MIN -4.8% 5-6 0.0% 5.9 13 -6.2% 28 -4.4% 23 6.7% 31
18 STL -5.9% 5-6 4.8% 5.2 16 -8.5% 30 -6.9% 26 8.4% 28
19 WAS -6.0% 4-7 -12.5% 5.1 18 7.8% 4 4.6% 11 9.2% 26
20 NYJ -8.9% 6-5 -9.8% 4.9 21 3.3% 12 -10.9% 31 15.1% 16
21 GB -10.2% 4-7 -12.4% 4.5 23 -1.6% 18 -6.6% 25 16.3% 9
22 BUF -10.6% 5-6 -11.2% 4.9 20 4.6% 8 7.4% 7 15.7% 14
23 TEN -13.1% 4-7 -19.7% 4.7 22 12.8% 1 8.4% 5 21.9% 7
24 SEA -14.6% 7-4 -10.3% 4.2 26 -6.7% 29 -8.5% 29 10.3% 21
25 HOU -14.7% 3-8 -21.5% 4.4 24 8.2% 3 -1.4% 20 10.4% 19
26 CLE -14.8% 3-8 -21.9% 3.6 28 4.5% 9 0.1% 16 8.5% 27
27 ATL -15.6% 5-6 -4.8% 4.3 25 -2.0% 21 4.2% 13 24.9% 4
28 SF -22.4% 5-6 -21.2% 4.2 27 -2.0% 20 -9.0% 30 22.0% 6
29 TB -25.1% 3-8 -34.4% 3.4 29 8.6% 2 -2.9% 22 7.4% 29
30 OAK -25.4% 2-9 -32.4% 3.1 31 2.6% 14 -1.4% 19 10.3% 20
31 DET -28.8% 2-9 -20.4% 2.8 32 -8.6% 32 13.6% 1 7.3% 30
32 ARI -30.5% 2-9 -23.2% 3.3 30 -5.1% 27 -1.5% 21 22.0% 5

Posted by: Aaron Schatz on 28 Nov 2006

189 comments, Last at 02 Dec 2006, 1:59pm by jebmak


by Felden (not verified) :: Tue, 11/28/2006 - 7:58pm


by What (not verified) :: Tue, 11/28/2006 - 7:58pm

hmm I see

by Felden (not verified) :: Tue, 11/28/2006 - 7:59pm

Durn it, I added a smiley face, and it deleted my actuall CONTENT.

I said it was nice to see the Niners improvement on the field reflected in their increasing DVOA.

by Ali Nagib (not verified) :: Tue, 11/28/2006 - 8:01pm

Wow, not only did the Bears not drop, but their ranking AND raw offense, defense and ST numbers are almost identical to last week. I'm less surprised by the former than the latter, then again it may go to further the "they are who we thought they were" theory of the Bears. Whatever that is these days.

by Aaron Boden (not verified) :: Tue, 11/28/2006 - 8:02pm

I expected Chicago to fall, but there was not much movement. Is this because only Grossman played poorly on sunday?

by dbt (not verified) :: Tue, 11/28/2006 - 8:03pm

Chart updated. Clicky my name for the pretty pictures.

by Richard (not verified) :: Tue, 11/28/2006 - 8:05pm

Pythagenpat Table at my name.

by Josh (not verified) :: Tue, 11/28/2006 - 8:07pm

Dallas is only team in top 5 on both offense and defense, NE only top team in top 10 in all three categories.
Philadelphia only other team in top 10 in both offense and defense.

by Ali Nagib (not verified) :: Tue, 11/28/2006 - 8:07pm

To summarize the Bears:

Week 12 - Off -3.4%(19), Def -28.8%(1), ST 8.9%(1), Overall 34.3%(1)
Week 13 - Off -3.5%(18), Def -27.8%(1), ST 7.8%(1), Overall 32.1%(1)

by bsr (not verified) :: Tue, 11/28/2006 - 8:07pm

I assume that there was no adjustment to Philly and that opponent adjustments will just take care of themselves over time? To bad opponent adjustments couldn't be weighted for only recent activity but I guess that would just be much too arbitrary.

by The Broilermaster (not verified) :: Tue, 11/28/2006 - 8:07pm

Interesting to see that Denver's offense is actually ranked HIGHER than their defense. Poor Plummer. If only the Denver sportswriters wouldn't blame Jake for everything. New TE's, new RB's (kinda), new WR's, injuries on the O-line...but Jake takes all the criticism.

by mactbone (not verified) :: Tue, 11/28/2006 - 8:09pm

HAHAHAHAHAHA, I just saw an article that said the Raiders have replaced Tom Walsh with John Shoop. As any Bears fan can tell you, they will now have the most dink-dunk offense in the league. Shoopification is a horrible, horrible thing.

by Gus (not verified) :: Tue, 11/28/2006 - 8:10pm

Good to see the Pats up there near the top of the Pythagorean wins... but is anyone else getting kind of scared of the Ravens in the AFC? With the Ghost of Jamal Lewis playing well right now, they're actually almost into positive offensive DVOA. Not to mention Mark Clayton's improved play of late helping out McNair.

by James C (not verified) :: Tue, 11/28/2006 - 8:10pm

I expected the Bears to stay roughly where they were on account of playing reasonably well on defense and putting 150 yards up against the Pats top notch run defense.

by The Broilermaster (not verified) :: Tue, 11/28/2006 - 8:10pm

The Raiders have actually impressed me a little bit the last few weeks. They got totally screwed vs. San Diego.

by Jim (not verified) :: Tue, 11/28/2006 - 8:11pm

Hey, these power ratings are starting to look downright conventional. The top six teams are pretty much everyone's top six, just not in the same order. As Philly falls without McNabb and Seattle rises with Hasselbeck and Alexander back, it'll probably look even more "normal". Also: nice to see my Cowboys entering the conversation for best team.

by RecoveringPackerFan (not verified) :: Tue, 11/28/2006 - 8:12pm

13: I take full responsibility for Lewis's resurgence. It's doing a fine job of absolutely obliterating my Loser League team.

by Moridin (not verified) :: Tue, 11/28/2006 - 8:12pm

I guess Arizona's opening kickoff touchdown was finally enough to allow the Steelers to take the ST uh "crown".

by Richard (not verified) :: Tue, 11/28/2006 - 8:13pm

Pythagenpat is way better than Pythagorean...

by rk (not verified) :: Tue, 11/28/2006 - 8:13pm

Thanks to J.J. Arrington's kickoff return TD, the Steelers finally have the worst special teams in the league.

by Will Allen (not verified) :: Tue, 11/28/2006 - 8:14pm

I'm a little surprised that the Vikings have held their mid range position for the last few weeks, but I suppose that reflects that just a little better luck could have delivered them victories against the 49ers and Dolphins. I am more surprised that the yardage they yielded against Leinert dropped their defensive DVOA so much, but I didn't see the game, so I don't have a good feel for it.

Is it possible that Billick has been overdelegating to his offensive coordinator(s) for more than just this year, or it just a matter of McNair and the offense meshing better as the year progressed?

by Jim (not verified) :: Tue, 11/28/2006 - 8:16pm

Playoff odds report is for week 10? :(

by Moridin (not verified) :: Tue, 11/28/2006 - 8:17pm

I find it almost amazing that the Vikings are almost the most consistent team in the league, according to variance.

by Becephalus (not verified) :: Tue, 11/28/2006 - 8:18pm


by NewsToTom (not verified) :: Tue, 11/28/2006 - 8:19pm

Hey, congratulations to the Steelers! You finally passed the Cardinals for the coveted #32 slot in ST ranking.

Other thing that jumped out at me: TEN's schedule ranking thus far is currently the second-hardest in DVOA history, behind only the 2004 Browns, though I'd bet they'll also finish behind the 2000 Jets (at least) when the season is over. The gap between #1 TEN and #2 is greater than the gap between #2 and #9. Using that (silly) measure, TEN's schedule is more of an outlier than Oakland's offense!

Biggest weekly risers and fallers:
+4: MIA (20-16), TEN (27-23)
-7: CLE (19-26)

Biggest WDVOA-DVOA disparities:
+5: TEN (23-18W)
-2: WAS (19-21W), NYJ (20-22W), BUF (22-24W)

by kleph (not verified) :: Tue, 11/28/2006 - 8:26pm

done ordered my robo-punter swag. finally i have something even more awesome to wear watching the big screen at my brother-in-law's house than my EDSBS 'possession reciever' shirt. and that's saying something.

by Pacifist Viking (not verified) :: Tue, 11/28/2006 - 8:28pm

Early in the year, I predicted that regardless of whether they actually win or lose, the Vikings would bounce around between 16th and 17th. So far it's been holding up.

I was looking at points allowed today and noticed that the Patriots are right there with Baltimore. But Baltimore is a scary team, and I think the AFC team most likely to go to the Super Bowl. BORING!

by King_Biff (not verified) :: Tue, 11/28/2006 - 8:28pm

As a Colts homer, I hope the colts "luck" keeps up, because their defense is ridiculous. Runningbacks run through the defense like its made of pantyhose and their zone coverage is, like an old pair of pantyhose, full of holes. Hopefully the officials hand it to them this year like they did the Steelers last year.

by throughthelookingglass (not verified) :: Tue, 11/28/2006 - 8:28pm

What's really impressive to me is that the Colts offensive DVOA is further away from average than the Raiders offensive DVOA.
And the Bears are 3rd in variance. They aren't who we thought they were.

by Chris (not verified) :: Tue, 11/28/2006 - 8:29pm

Mike Harris' playoff predictions and mine are really starting to converge (Click my name for a comparison). If you compare last week to this week, in some cases the DVOA odds are moving toward mine and in others mine are moving toward DVOA.

by NF (not verified) :: Tue, 11/28/2006 - 8:29pm

I didn't think Philadelphia played that badly...

Jacksonville will find a way to have greater than 100% variance. I just have a feeling.

Looking at Detroit's DVOA, and looking at its future schedule, I'd say they have the same chance of winning another game as Chicago does of not getting HFA. Also, Seattle blows my mind. Is there any other division in the league where it would have a chance of winning the division?

by James (not verified) :: Tue, 11/28/2006 - 8:31pm

#23 - Yes, but the Chargers are the only good team in the top five of variance. The next best teams are Cincinnati and Indianapolis, which are 22, and 18 respectively.

Yay for my home team (Chargers)?

Doesn't the fact that the Chargers are consistently good make their DVOA ranking "better"? Or does DVOA already take that into account?

by Stillio (not verified) :: Tue, 11/28/2006 - 8:32pm

31: Jacksonville actually managed to drop 4 points in variance despite the implosion against Buffalo...don't ask me how that works, but the numbers don't lie, right?

by James C (not verified) :: Tue, 11/28/2006 - 8:35pm

I wonder if someone can help me with this it isn't phrased all that well but please bear with me.

If DVOA is based upon all the data that FO have acrued over several seasons, does the fact that only six teams in the NFC have a positive DVOA reflect how bad the conference is?

The total DVOA for the NFC is -54.7, that seems pretty bad.

by NF (not verified) :: Tue, 11/28/2006 - 8:36pm

Sometime in January, the Ghosts of Ravens/Titans Offense 2000 and the Ravens defense will play the San Diego Chargers, and it will be the greatest game of the year.

by James C (not verified) :: Tue, 11/28/2006 - 8:37pm

The total DVOA for the AFC is +63.4.

by Will Allen (not verified) :: Tue, 11/28/2006 - 8:37pm

#23, what is the source of your surprise? From the start, and to this point, they have been the same in most every game. Absolutely superb run defense. About average pass defense. About average running offense. Hideous beyond any description passing offense. Somewhat below average special teams. There haven't been too many games which have departed from this script.

by Adam (not verified) :: Tue, 11/28/2006 - 8:40pm

Does DVOA reflect the full difference in strength between the AFC and the NFC? Each team only plays 25% of its games against teams in the other conference. So if the AFC is a lot better than the NFC, the full difference might not show up.

by Richard (not verified) :: Tue, 11/28/2006 - 8:41pm

32: I believe Estimated Wins takes that into account.

by Catfish (not verified) :: Tue, 11/28/2006 - 8:45pm

re: 12

I remember one columnist in the Chicago Trib describing the Shoop offense as "using the run to set up the punt". How good is Randy Moss at running 2 yard curls on 3rd and 8?

by Will Allen (not verified) :: Tue, 11/28/2006 - 8:45pm

Aaron has enough to do, but it really would be interesting to me to know how Leinert's DPAR, and the Vikings' defensive DVOA in the last game, would have been affected if Chester Taylor had run in his goal line run to make it 38-12 with ten minutes left, instead of coughing up a 99 yard fumble return to make it 31-19. Does that fumble return make Leinert's passing yardage after that point much more statistically important?

by turbohappy (not verified) :: Tue, 11/28/2006 - 8:47pm

Re: 32

Variance doesn't affect DVOA, but having a low variance does help estimated wins.

Speaking of variance, if your team isn't ranked at the top isn't it more encouraging if they have a high variance? You know the Vikings aren't going to beat the Chargers, but on a good day the Steelers have a shot, right?

by MdM (not verified) :: Tue, 11/28/2006 - 8:48pm

Will someone rationally explain to this crazed Eagles fan how they could be so bad now? What happened!

by Richard (not verified) :: Tue, 11/28/2006 - 8:53pm

43: It would seem they all know that they have no chance in hell without McNabb.

by dbt (not verified) :: Tue, 11/28/2006 - 8:53pm

The Eagles were 31st or 32nd in variance until two weeks ago. Now they're 13th. 9th-16th are all bunched up -- they could easily be in the top 5-8 next week.

by Becephalus (not verified) :: Tue, 11/28/2006 - 8:58pm

32 It makes them a much ebtter bet in the playoffs where you need to play well 3 or 4 game sin a row :)

by Noble (not verified) :: Tue, 11/28/2006 - 9:02pm

16: My guess is that the rankings are starting to look "conventional" because sports writers' rankings become more accurate as the season goes along. The truly crappy teams remove themselves from contention and the truly good teams separate themselves from the pack.

Take a look at Peter King's ranking, the top six look like FO's (if you switch Indy and Chi), but after that the rankings become more muddled and he's prone to remembering only wins and not whys.

Actually, a weekly comparison of rankings would be interesting. *contemplates the project*

by Joseph (not verified) :: Tue, 11/28/2006 - 9:05pm


I don't remember--does Mike Harris' site account for tiebreakers? (My Saints are currently 4-1 in div, 6-1 in conf.)


by NewsToTom (not verified) :: Tue, 11/28/2006 - 9:05pm

Re #47
MDS has been doing something like that; see last week. But one could easily do more, and the results might be interesting.

by Jim (not verified) :: Tue, 11/28/2006 - 9:20pm

47: They probably look conventional now because the top 6 (the list, not the order) and the bottom 3 match up perfectly with W-L records (with Dallas in there as the top 7-4 team). The closer I look at it, the middle teams are probably still ranked quite differently from conventional power rankings.

by Chris (not verified) :: Tue, 11/28/2006 - 9:28pm

Re 48: He does consider tiebreakers, at least to some extent. From the week 11 DVOA report:
"...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."

It's not clear from the above statement if Mike Harris' simulations consider common opponent tiebreakers (which comes before conference record for breaking division ties), or if they accurately handle 3-way ties and some of the finer nuances of the tiebreakers. My software rigorously breaks all ties through strength of schedule, and you can run your own specific scenarios.

by PhillyCWC (not verified) :: Tue, 11/28/2006 - 9:35pm

McNabb had surgery today to repair his ACL. That's a positive, at least. More bad news for the Eagles, though...Mike Bartrum, the long-snapper, is out for the rest of the season with some kind of neck injury.

Seriously, haven't the Eagles paid the devil enough already for 2004's Super Bowl run?

by Marko (not verified) :: Tue, 11/28/2006 - 9:36pm

12, 40: My thoughts exactly about Shoop and the Raiders. And it's not like the Raiders' offense wasn't horrible and boring already. Putting Shoop in charge of it should add to the already high comedy value of watching the Raiders.

Shoop loves 2 or 3 yard passes toward the sideline on 3rd and long, which have no chance of getting a first down. The routes in his offense hardly ever generate yards after the catch. Maybe the Raiders are looking for every advantage they can get in trying to secure the #1 overall draft pick, considering the stiff competition they face from the Cardinals and the Lions.

by NF (not verified) :: Tue, 11/28/2006 - 9:36pm

12: A dink-and-dunk offense will work a lot better than a 7-step drop and get sacked offense. Does Al Davis know about this?

There is something absolutely remarkable in this weeks rankings. The Houston Texans no longer have the worse defense in the league. The last time the Texans had a defense that wasn't the worst in the league, TO was in Philly, the Randy and Santana Moss were in Minnesota and New Jersey, Carson Palmer was a struggling second-year QB, and Dave Wannestedt was chopping wood in Miami.

by Fnor (not verified) :: Tue, 11/28/2006 - 9:37pm

I am impressed by the ROBO-PUNTER shirt, but I would much rather have it in an actual jersey....

by Yaguar (not verified) :: Tue, 11/28/2006 - 9:46pm

Dallas and New England are the only top teams with a modicum of balance.

by Jason Scheib (not verified) :: Tue, 11/28/2006 - 9:54pm

I just have to mention how close the top of this list compares to Actual Turnover Ratio. The top three are the same:
Chicago +34
San Diego +27
Baltimore +22

and teams 4 through 7 are the same teams just in a slightly different order:
Indianapolis +20
New England +17
Dallas +15
Jacksonville +15

by Becephalus (not verified) :: Tue, 11/28/2006 - 10:08pm

how does ATR differ from TR?

by Pat (not verified) :: Tue, 11/28/2006 - 10:11pm

Mike Bartrum, the long-snapper, is out for the rest of the season with some kind of neck injury.

Mother of...

Paid the devil? At this point, we're way past that. It's black cats under ladders by broken mirrors time. So what voodoo master decided he was a Cowboys fan this year?

Seriously, for those who are wondering why losing a longsnapper is such a big deal, Bartrum was routinely selected to the Pro Bowl by the coaches as a "need" player. He's basically the best in the business.

This is just unreal.

by Lou (not verified) :: Tue, 11/28/2006 - 10:13pm

ATR relates to the Never Punting theory. Link below

by Becephalus (not verified) :: Tue, 11/28/2006 - 10:14pm

oh that. I thought it might be that :)

by admin :: Tue, 11/28/2006 - 10:21pm
by Rick (not verified) :: Tue, 11/28/2006 - 10:23pm

Quick question:

When Miami (negative offense, strong defense) plays Jacksonville (negative offensive, strong defense), is there a chance of getting a negative score?

Time to see what odds the online bookies are giving for a 0-0 result...

by Splat (not verified) :: Tue, 11/28/2006 - 10:43pm

What are the odds of getting say, a "Snapshot score" for a team, a DVOA score for just the past 3 weeks. Full season adjustments, etc, just averages of the last 3 weeks so we can see how teams are trending?

by Jason Scheib (not verified) :: Tue, 11/28/2006 - 10:46pm

re #60, 61. Actually ATR is a distinct concept that isn't tied to the never punting theory at all. The never punting theory, good or bad, just made use of that statistic. Just because the never punting idea may have issues doesn't mean ATR has to be guilty by association. It's a valid and meaningful stat in and of itself. The purpose of my post in #57 was to show that. Forget the never punting stuff, it's unrelated. Not related.

by ZS (not verified) :: Tue, 11/28/2006 - 10:52pm

Who was the other Rutgers starter besides L.J. Smith in the SEA-PHI game?

by Lou (not verified) :: Tue, 11/28/2006 - 10:53pm

I agree ATR is usefull. I just associate the two because they were introduced in the same article.

by Ben (not verified) :: Tue, 11/28/2006 - 10:57pm

re:66 It was Ind-Phi, and Gary Bracket, the Colts starting MLB went to Rutgers.

by Pat (not verified) :: Tue, 11/28/2006 - 10:58pm


To win it all, the Bears need to figure out how to get the Good Rex every week.

Didn't someone here note that Good Rex comes out very consistently? He comes out when the Bears play below-average pass defenses. Grossman's good weeks almost exclusively came against teams with worse than average (~0% DVOA) passing defenses.

Based on that, the only way the Bears are getting Good Rex out every week is if they manage to convince the league to let them play Detroit every week.

by Tim L (not verified) :: Tue, 11/28/2006 - 11:16pm

Aaron wrote about the Colts that "Over the past 25 years, only two Super Bowl champions did not finish among the top four teams in Pythagorean wins: the 2001 Patriots and the 1988 49ers."

There's actually six teams that meet the criteria, not two.

1982 Redskins finished fifth
1983 Raiders finished fifth
He got the 1988 49ers...
...and 2001 Pats
2003 Patriots finished sixth
2005 Steelers finished fifth

But the quote certainly makes a game attempt to try and put the Colts in their perceived place.

by Chris (not verified) :: Tue, 11/28/2006 - 11:25pm

Forgive my stupidity, but how do you calculate Pythagorean wins?

by Becephalus (not verified) :: Tue, 11/28/2006 - 11:28pm

pts squared divided by pts square + pts allowed sqaured

and 70 where did you get those numbers? I would love to see tables :)

by PhillyCWC (not verified) :: Tue, 11/28/2006 - 11:48pm

Here's the link to the story about the Bartrum injury...

by Jeremiah (not verified) :: Wed, 11/29/2006 - 12:08am

A slightly better formula for pythagorean wins is:

scored^2.37 /(scored^2.37 + allowed^2.37)

Then of course to get the number of games (instead of a percentage), simply multiply by the number of games played.

by Pat (not verified) :: Wed, 11/29/2006 - 12:09am

It's not exactly squared. It's 2.37.

The exponent essentially is related to the average points per game, going up as the points per game goes up. So it's lowest for baseball, medium for football, and ridiculously high (16.5) for basketball. For a game with an average of one point per game (i.e. a game with no scoring - say, chess), it'd be 1 (and the formula would be exact).

by NewsToTom (not verified) :: Wed, 11/29/2006 - 12:11am

Re #70/72
For baseball it's raised to the 2nd power, but for football it's slightly different (2.37, actually).

For 2003, though, I show the Pats as #1 in PYTH at 71.1%, narrowly ahead of KC's 71.0%. PIT last year was 5th, but only about .06% behind #4 CAR (exp. wins 11.6215 for CAR, 11.6120 for PIT). Raiders were 5th in 1983, about .15 expected wins behind #4 SF (10.61 v. 10.46, rounded off). In 1982, though, Redskins were 3rd with 71.8%, behind Dallas at 74.1% and Miami at 72.7% and narrowly ahead of the Jets at 71.6%. So, 4 teams outside the PYTH Top 5.

by NewsToTom (not verified) :: Wed, 11/29/2006 - 12:12am

Re #76
Ok, for baseball, as the linked article says, it's 1.82, not 2. And that's 4 teams outside the PYTH Top 4, not the Top 5.

by Tom Kelso (not verified) :: Wed, 11/29/2006 - 12:14am

Minor point, but the fellow's name is Ovie, not Ogie, Mughelli.

I wouldn't even mention it, but the same typo appears in the Audibles thread as well.

by Stillio (not verified) :: Wed, 11/29/2006 - 12:22am

"We have the Jaguars ranked seventh in the DVOA ratings, but can anybody imagine Jacksonville actually playing well in four straight games?" Sadly, no. Worse, though, is the fact that Jax will likely have to win in KC on the last week of the season to even make the playoffs. Can anybody imagine Jacksonville actually playing well in five straight games on the road? Didn't think so.

by dbt (not verified) :: Wed, 11/29/2006 - 12:25am

Team (pass off rank)

Good Rex:
Green Bay (23rd)
Detroit (31st)
Seattle (25th)
Buffalo (10th)*
San Fran (29th)
NY Jets (26th)

Spotty Rex:
Minnesota (18th) (at MIN)
NY Giants (8th) (bad until they stopped blitzing)

Bad Rex:
Arizona (19th)
Miami (14th)
New England (11th)

So here's the thing. They're not particularly great. It's about whether or not 1) they know to blitz the crap out of him 2) they can accomplish it against the Bears middlin' line.

by admin :: Wed, 11/29/2006 - 12:40am

I e-mailed FOX to fix the spelling on Ovie Mughelli, and add the 1983 Raiders to the list of teams outside the Pythag top four. My eyes must have glazed over or something. However, I would consider last year's Panthers and Steelers to be tied for fourth. If you see a big, significant difference between .7263 and .7258, you are probably reading the wrong website. Honestly, if they had let me put Baltimore and Chicago side by side this week, I probably would have.

by Len (not verified) :: Wed, 11/29/2006 - 12:48am

Anyone else here looking forward to a Parcells - Belichick Super Bowl?

That would be incredible.

by Ben (not verified) :: Wed, 11/29/2006 - 12:56am

re:82 I think you meant to type unbearable...

by paytonrules (not verified) :: Wed, 11/29/2006 - 1:07am

That mid-season projection with the Bears at 9th is looking pretty good right now, as the ST and defense regress to the mean consistently.

It's going to come down to one player for the Bears this year, more than any other team.

BTW - shouldn't the Bears be thinking real hard about Byron Leftwhich next season? Methinks he'll be available, since David Garrard just wins baby.

by Bobman (not verified) :: Wed, 11/29/2006 - 1:14am

If you look at the Steelers commentary about luck ("When a team suffers from terrible luck, but really isn't that bad, two things can happen. Some teams, like the 2004 Panthers and 2005 Redskins, will get their act together around midseason and surge in the second half. Other teams, like the 2003 Bucs, will collapse completely."), why can't this work in the reverse? Does it?

Something along the lines of: After the mid-point of the season, unreasonably lucky teams (like Indy) either revert to the mean or continue on their ridiculous run of luck.

The FO CW keeps dinging Indy for their opponents' ill luck (bad FG percentage, maybe a few fumble recoveries here and there). Yet there is evidence that statistically unlucky teams can either regroup or collapse. Isn't winning it all with good luck analogous to collapsing with bad luck?

I'm not saying it is any more likely that the team with the most projected wins and highest DVOA bombing out in their first playoff game. (cough cough gasp) But as a logical construct, it certainly seems to be one of the possible outcomes. Super super long-shot, or just an uncommon occurence?

by Becephalus (not verified) :: Wed, 11/29/2006 - 1:25am

85 what exactly are you trying to say, because its not at all clear.

by B (not verified) :: Wed, 11/29/2006 - 1:27am

If Good Rex only shows up against bad defenses, and Chicago needs good Rex to make it to the Superbowl, the only NFC team they have to worry about is Dallas. Well, that and root for the Colts to take the AFC crown.

by Rollo (not verified) :: Wed, 11/29/2006 - 1:30am

Doesn't good or bad Rex just depend on whether he throws off his back foot, which he does when pressured?

by kevinNYC (not verified) :: Wed, 11/29/2006 - 1:45am

For all the talk of Eli being the worst QB in football history, the defense needs players and to pick up their performance. Over the past 5 weeks, the defense DVOA ranking has gone 5-6-5-8-12. That's a pretty big drop over two weeks. That future schedule number never will get better for the Giants, huh?

As far as the commentary, the Giants actually have good depth on the o-line at every position... except left tackle. BTW, how many teams have good backups at left tackle?

by Fnor (not verified) :: Wed, 11/29/2006 - 1:54am

"Bad Rex" was out in force against Minnesota. Part of it is playing good defences. Another, more important part, is luck and whether a defence can make Grossman pay for his bad throws. Minnesota should have, but dropped an obscene number of interceptions.

I think "F--k it, I'm throwing deep." or whatever is the best way to characterize it. Except he's not particularly accurate on short throws, either. It's all just a confusing mess.

by centrifuge (not verified) :: Wed, 11/29/2006 - 2:10am

86: I think he's saying (and, Bobman, correct me if I'm wrong):

Good teams with bad luck have two potential outcomes for the second half of the season: luck regresses to the mean (leading to an improvement in results), or becoming an actual bad team (ladies and gents, your Pittsburgh Steelers!).

So do the same two results (inverted) also apply to bad teams with good luck? We've seen many such teams have their luck regress to the mean, leading to a horrific crash, but I can't recall many instances of such a team actually becoming good. I think Bobman wants to know if that ever happens, and if not, why not?

by centrifuge (not verified) :: Wed, 11/29/2006 - 2:18am

Oh, and 55: NFLShop customizable jerseys run about $100, and they take up to ten characters.

If only you could change the team name to the Outsiders...

by Independent George (not verified) :: Wed, 11/29/2006 - 2:25am

Huzzah! Pittsburgh special teams finally drops to 32nd! Congratulations to Santonio Holmes, Ricardo Colclough, and everybody else whose hard work made this possible.

Interestingly enough, it looks like the drop in ranking comes more from an improvement in Arizona's score than a drop in Pittsburgh's.

by DavidH (not verified) :: Wed, 11/29/2006 - 2:29am

I don't know if "lucky" teams ever become "actually good" teams, but if they don't, I think it might be because it's easier for Scenario 2 (below) to happen than it is for Scenario 4 to happen. What I mean is, it's easy for a team to get worse because of bad attitudes. It's harder for a team to get better because of good attitudes, simply because playing well is harder than playing badly.

Scenario 1:
Team is unlucky in the first half. Team plays the same in 2nd half, gets luckier, and wins a few.

Scenario 2:
Team is unlucky in the first half. Team kind of gives up on the season, and plays worse in 2nd half.

Scenario 3:
Team is lucky in the first half. Team plays the same in 2nd half, gets less lucky, and comes back to earth.

Scenario 4:
Team is lucky in the first half. Team starts to believe more in themselves, plays better, and wins more games.

by Becephalus (not verified) :: Wed, 11/29/2006 - 2:38am

91 My suspicion is that this dichotomy is completely false. And that their luck and performance continue on two bell shaped continuims. Luck centered around average, and play centered around what it was in frist half.

by putnamp (not verified) :: Wed, 11/29/2006 - 3:00am

100 to 1 odds? Really? ;)

by Bobman (not verified) :: Wed, 11/29/2006 - 3:25am

91 Centrifuge, you got my roundabout question right.

94/95, Both good points. Actually, this is kind of encouraging for Indy, which was DVOA ranked about 16 in DVOA (and 7th in DAVE) in Week 6. Luck was largely assumed to be the reason they were 6-0 at that point. But maybe they were just a better team putting their act together (and also lucky).

Now they are 5th in DVOA and slightly less lucky (if you look at their TOs the past few weeks), but as a record, 10-1 is pretty comparable to 6-0. So their play must be improving even as luck drops off a bit. So it's not quite DavidH's Scenario #4, since their play per DVOA has improved quite a bit from a DVOA of 4.6 to 23.9.

I guess what I am saying is if their luck remains pretty good, then they will win way beyond their Pyth. and estimated wins.

Luck aside, what we really need is a visual graph of which teams have upward and downward trajectories in their DVOA, and extrapolate that (if that's logical). I am crushed at work and do not have time, but it would basically take weekly DVOA and chart it from previous weeks, fit a line to it, and see where it projects to.

Just taking three teams and two points for a crude example, Indy went from 4.6 to 23.9 from Week 6 to Week 12, while NYG are holding steady at 17.8 for both then and now, and Chi went from 64.7 to 32.1. One nice climb, one flat-line, one steep drop (but still the top slot). Does this imply that Ind will continue to climb and Chi continue to drop? If I had to bet $1 I'd say yes, but if I had to bet $100, I'd say "beats me."

Looking to history for some meaning here, the final 2005 DVOA had Indy at 36.6, Sea at 28.2, and Pit at 26.1. Eight weeks earlier, those numbers were 36.3 for Ind, 25.4 for Sea and 23.9 for Pit. Not as dramatic as I hoped, but Indy was flat, while Sea and Pit were both on a mild upswing in the second half of the season. (based on only two data points)

So for postseason success, do we look for teams with a cumulative high DVOA which is somewhat bouyed by good performances two months ago, or teams that are peaking and on the upswing (who might be dragged down by a poor performance in week 10, but clearly improving)?

Okay, I have avoided actual paying work for too long. Sorry for the long post.

by Sid (not verified) :: Wed, 11/29/2006 - 3:26am

RE: 11

I think the key is that Gary Kubiak left. Look at how Plummer did with Kubiak and Lepsis there last season.

by Richard (not verified) :: Wed, 11/29/2006 - 3:56am

89: The Chargers.

by Lou (not verified) :: Wed, 11/29/2006 - 4:06am

The Bears will already be paying Grossman and Griese next season. They can't afford to spend any more money on that position. They have to find a way to retain Briggs.

I like the way you think.

by Moridin (not verified) :: Wed, 11/29/2006 - 4:21am

Re: 37
That's why I said 'almost' :D . My more analytical side realizes they play pretty close to the same all the time, but my still leftover before-FO fan side thinks they play more sporatically. Ah, perception.

I'm curious though what the biggest recorded differences in Passing D and running D has been in DVOA though. Right now the Vikings is about 50, and Washington is 40. Same question about the Offense side of things (where it looks like Indy, New Orleans and Atlanta have the lead in difference).

by IzzionSona (not verified) :: Wed, 11/29/2006 - 6:23am


I do know that DVOA tends to be wildly spread early in the year, when opponent adjustments are small and (because) there isn't much data in the system. So, there's a natural "mashing down" of numbers like Chicago's. So that'd be one bias there.

Of course, there is one "fitted line" type number provided to us by Aaron... from what I gather, WDVOA is supposed to show more of the "current trend" in DVOA, yes?

by Paulo Sanchotene, Brazil (not verified) :: Wed, 11/29/2006 - 7:56am

Why MDS just don't buy a Lions' shirt with Millen's name written on it?

by Gerry (not verified) :: Wed, 11/29/2006 - 8:22am

"I think Bobman wants to know if that ever happens, and if not, why not?"

I would guess that it may have ever happened, because even if the odds of it happening are slight, they are not zero. I would also guess that the number of times it has happened is small.

As for why not, look for explanations of why it could occur on the other side. Look at the Steelers. They were having bad luck, but why did it all collapse? One could easily speculate that as their chances of the playoffs diminished, their desire to sacrifice themselves (by playing hurt, by pushing hard in practice, etc) lessened, which lowered their quality in a measurable way.

But if a team was already working hard, and was unusually lucky, would they work harder then, or would they think that the good results they were getting showed they were working hard enough already? And even if they did work harder, is there a point of diminishing returns on the effort level? I think it is easy to believe that an unusually lucky team would not be able to raise their true quality level to the same extent or as easily as a truly unlucky team would be to lower theirs.

by james (not verified) :: Wed, 11/29/2006 - 9:00am

looking at qb dvoa over the past couple of years to determine sb winner

Most years the winner was in the top5.

so this year you have


I eliminated McNabb and Huard because they aren't playing.

Whats interesting is Huard being benched with a no. 4 dvoa rating. WTF?

by james (not verified) :: Wed, 11/29/2006 - 9:01am

Bears will probably get home field but have to beat 2 top 5 dvoa qb's to advance.

by Chris (not verified) :: Wed, 11/29/2006 - 10:19am

105. "Whats interesting is Huard being benched with a no. 4 dvoa rating. WTF?" Huard did a great job filling in, but it was a lot of smoke and mirrors. Passing only at very opportune times, single read patterns, a lot of passes to Gonzo. His performance was probably not sustainable as the Chiefs started facing better defenses down the stretch. This was evidenced by Huard having a pretty bad game against Miami. Even though Green is rusty, he's still making throws that you just wouldn't see from Huard, as well as making multiple reads, there by getting additional production out of Sammy Parker. Anyone who just watches the two, can see who the better QB is. I'm definitely much more comfortable having Green under center during our playoff race. Again Huard did a great job filling in; he saved our season.

by JasonK (not verified) :: Wed, 11/29/2006 - 11:04am


Bob Whitfield is also the backup at RT, so that's 2 spots on the line where the Giants have crappy depth. (I, for one, don't think that they would move Dave Diehl.) For a team that started the season with 10 OL on the roster, that's pretty shocking.

Personally, if the Giants lose to Dallas this weekend (effectively ending their playoff hopes), I'd like to see them throw Guy Whimper in at LT, just to see what he can do.

by zlionsfan (not verified) :: Wed, 11/29/2006 - 11:47am

Hey now, let's not forget hockey. I can't find an article more recent than my link; back then, they provided three factors, one for the early days, one for Original Six, and one for what is now the pre-rule-enforcement era. The most recent exponent ... was 2.

That's odd. I would have thought hockey would have a lower exponent than baseball because they'd be scoring fewer goals than baseball teams score runs.

Oh yes. I remember the days when the 49ers' DVOA was lower than network television standards and the Lions were pretending to be an NFL team ...

by paytonrules (not verified) :: Wed, 11/29/2006 - 11:59am

I was thinking a trade and releasing Greise to make Grossman the backup (or they could compete in camp).

There still could be payroll issues though. It's just so rare for a young QB of Byron's caliber to be available via trade - and for some reason the world has soured on him so....

by Devin McCullen (not verified) :: Wed, 11/29/2006 - 12:03pm

Actually, losing this week wouldn't end the Giants' playoff hopes, although it would put the kibosh on the division. They've got a good conference record and still have to play Carolina, so a wild card is well within reach. Of course, a sufficiently embarassing loss would probably send them completely into a spiral of collapse. But they still have to play the Eagles and Skins, and so they could wind up at 8-8 despite themselves.

by Ryan Mc (not verified) :: Wed, 11/29/2006 - 12:42pm

"As far as I know, there is no precedent for this, for a playoff contender voluntarily switching from a veteran quarterback to a rookie this late in the season."

What about the Raiders going to Todd Marinovich at the end of the 1991 season? I believe this was a coach's decision, or was Jay Schroeder actually injured? Raiders fans?

by jebmak (not verified) :: Wed, 11/29/2006 - 1:06pm

Re: 34+36 James C. (If you are still around)

DVOA (weighted or not) is for 2006 only. The NFC is the worse conference again just based on this year.

by CaffeineMan (not verified) :: Wed, 11/29/2006 - 1:12pm

re: 113

Wait a sec, I thought James C was correct, that the 0.0% League Average (aka Blutarski) DVOA was calculated using data from all seasons available. That way you can look at things like "best season in the history of DVOA", rather than confining conclusions to a single season.

by Wanker79 (not verified) :: Wed, 11/29/2006 - 1:19pm

I took the Total DVOA numbers for each week (back to Week 5 when the 'D' took effect) and fitted linear trendlines for each team. Here are the slopes of each trend:


Granted, it's alot easier for a bad team to trend up and a good team to trend down as opposed to vise versa, so I really don't know if this actually tells us anything.

by admin :: Wed, 11/29/2006 - 1:47pm

All stats pages are now updated.

To answer a question above: the offense and defense baselines are based on 2002-2005. The special teams baselines are a little different; I had to adjust them last year because kicking keeps getting better and returns keep getting shorter, so these are based much more on 2005 only.

by b-man (not verified) :: Wed, 11/29/2006 - 2:05pm

Click my name for some charts I put together from the historic DVOA data.

by b-man (not verified) :: Wed, 11/29/2006 - 2:07pm

Fixed link.

by james (not verified) :: Wed, 11/29/2006 - 2:13pm

re huard,

If you can smoke and mirrors your way to a no. 8 dpar and no 4 dvoa while not starting every game and the other guy has never gotten you deep in the playoffs...wtf?

by Lou (not verified) :: Wed, 11/29/2006 - 2:41pm

Green was 4th in DPAR last year and 7 in DVOA. Also I would think going to back to the regular starter would allow KC to open on the playbook and not rely on running LJ into the ground. But I havn't seen any KC games this year.

by Pacifist Viking (not verified) :: Wed, 11/29/2006 - 2:54pm

This may sound stupid, but here's my defense of Grossman's "F-ck it, I'm throwing deep" philosophy.

When you have a dominating defense but a questionable offense, you have to try run an offense that best supports your defense. Last year the Bears ran an ultra-conservative, runs and short passes offense that attempted to make few mistakes and let the defense help them win games 13-6 and such.

Another way to go is to say, "Screw it, we've got a dominating defense, we can afford to take risks." So they chuck it deep a lot. It might lead to turnovers, but the defense is so good the turnovers don't usually end up killing the team. And when it is working they have a chance to destroy teams.

Just a thought.

by james (not verified) :: Wed, 11/29/2006 - 2:59pm

re trent green
yeah well he sucks this year

The team went on a tear with Huard at the helm. The offense always relied heavily on the running backs. I don't remember Green winning any games all by himself with his arm. He played well within the gameplan.

by Rich Conley (not verified) :: Wed, 11/29/2006 - 3:03pm


Of all the games I've seen from Chicago this year, I dont think going more conservative on the offense would have lost them any more, and it certainly would have won them the games in Miami and NE.

by Rich Conley (not verified) :: Wed, 11/29/2006 - 3:05pm


From what I've seen, the never getting deep in the playoffs never had anything to do with Trent Green, and everything to do with the same problem Peyton has: Their defense couldnt stop a girl scout squad.

by Not saying (not verified) :: Wed, 11/29/2006 - 3:53pm

RE: 115

Are you using DVOA from each week's ratings? That would cause some problems in your trend lines, because the D adjustment now is different from the D adjustment in week 5, since we now have more data on how good teams are. Week 5 is much more dependent on early blowouts, for example.

by b-man (not verified) :: Wed, 11/29/2006 - 4:07pm

Sorry, the website field is not allowing slashes.

DVOA Charts 2006 Aggregate and Week-by-Week

by James C (not verified) :: Wed, 11/29/2006 - 4:19pm

If I am reading Aaron's post properly does it indeed mean that the mean DVOA of AFC teams is +4 and NFC is -3. That seems like a really big difference, am I understanding this correctly?

by Starshatterer (not verified) :: Wed, 11/29/2006 - 4:24pm

Centrifuge (#91 ) et al--

Possible candidates for the bad-but-lucky team upgrading to good in the second half of the season:

2001 Patriots
2003 Panthers

by kevinNYC (not verified) :: Wed, 11/29/2006 - 4:35pm

#108... The scary part is the Giants could lose the next two games and still have an excellent shot at making the playoffs. All the NFC teams that are 5-6 are almost guaranteed to lose at least 7 games. The Giants last three games (PHI, NO, @ WASH) give them a plausible shot to win 9 games. If they win 9 games, they will have no worst than 4 conference losses.

by Fnor (not verified) :: Wed, 11/29/2006 - 5:00pm

#124: I thought girl scouts were organized into troops?

#126: That chart is fraught with awesome.

by stan (not verified) :: Wed, 11/29/2006 - 5:12pm

Bobman and the "lucky team gets good" question --

Some thoughts:

1. Confidence is a huge factor in every sport. If a team gets "lucky early" and gets confidence, it very well could improve their development. It may also have a negative impact on the mental state of future opponents.

2. Some of what is measured as skill by DVOA is actually luck e.g. whether deflected pass comes down for INT or reception. See Brady's pass in AFC champ game after 2003 season where he threw it right at Freeney dropping on zone blitz. It hit Freeney in the arm or pads and bounced 15 feet in the air and came down right to the only Pat amidst the collection of Colts in the area. Perhaps such a pass should be treated like a fumble. It was a bad play by the QB which far more often becomes a turnover than a completion.

What about bad throws that defenders just drop? Maybe the defense shouldn't get credit because they lack the ability to make the play. But the same throw against a good DB gets picked. If DVOA is intended to be predictive, that throw shows a QB more likely to throw into coverage than his number shows.

3. Team DVOA is an average with offense, defense and special teams getting specific shares. Shouldn't the share allocated to each be adjusted depending on the team? Doesn't it seem likely that teams which are extraordinarily good on one side of the ball are more successful? Or should we explore the possibility that the DVOA percentage applied to each area of the game may need to change to reflect the fact that game outcomes are not determined in the same percentages?

Example -- take the Colts. Assume we have a offense that scores almost every time and a defense that can't stop anyone (like Colts vs. Chiefs in 2003 playoffs). In such a combination of offense and defense, special teams play becomes almost irrelevant. Any difference in field position from good or bad ST has no significance in a high scoring game where offenses are scoring on every possession regardless of field position.

If so, is the Colts' overall team DVOA too low because it overstates the actual impact of lousy ST play in their games? It may very well be critical to a team in the opposite condition (great D, no O). In that case, it might need to have a bigger share of overall DVOA in order to accurately predict the team's performance.

Curious to read responses.

by Rich Conley (not verified) :: Wed, 11/29/2006 - 5:22pm

1. Definitely. Take the Raiders for instance...they win a couple games, and Randy Moss may actually start running his routes. The lineman may actually start blocking, because its no longer a "it doesnt matter if I stop this guy, because the guy over there is going to get burned" etc.

2. Is completely correct, but at the same time, most of this stuff isnt in the play by play, so theres no way to use it in DVOA. The idea is that hopefully, it balances out, and for each one of those balls that brady throws off of freeny, and gets caught for a gain, he throws one that hits a reciever, and gets tipped to a defender. The problem is, theres no way to tell, without the game charting data, and thats too slow.

3. I dont know how you'd do it, but I agree. I remember screaming at the TV in that INDY-KC game when KC was down by 7 that they should just start onside kicking every kickoff, because the lost field position in a failure didnt mean anything.

by stan (not verified) :: Wed, 11/29/2006 - 5:30pm

One more thought on "luck" and DVOA --

The Colts are having extraordinary success on 3d down. Does DVOA adequately account for this? i.e. two bad plays plus one good play = first down.

Also -- the question of "clutch"
I know baseball number crunchers have concluded that clutch hitting is mostly a myth. However, I know that football coaches will often save certain really good plays for crunch time situations. I'm sure that the Colts have pulled a few of these from the special section of the playbook in critical situations.

And coaches often play vanilla in safe situations or safe games. A little like playing with a self-imposed handicap as far as DVOA is concerned. Thus, for a team that is in first place with a sizable lead, isn't it possible that the actual potential of the team is understated by the DVOA number?

by Rich Conley (not verified) :: Wed, 11/29/2006 - 5:39pm


The '3rd down success rate much higher than 1st and 2nd down rate' occurance is generally seen as almost a straw man. Generally teams that have that for the year tend to have a large regression the next year.

by Gold Star for Robot Boy (not verified) :: Wed, 11/29/2006 - 5:41pm

b-man, I think your VOA/DVOA charts are missing the Buzzsaw, aka the AZ Cardinals. Despite that, great stuff.

by stan (not verified) :: Wed, 11/29/2006 - 5:45pm


I agree that you can't expect DVOA to measure the lucky bounce in the air. Just pointing out that there are inherent limits to what stats can tell you because luck will always be embedded in the numbers.

In the case of the bad throw with good outcome, I was going to say that it can't ever balance out because they are all bad throws. But we also have the flip side of balls bouncing off receivers' hands and into DBs (or the WR just gets it taken away).

What is really curious is the case of the ball that is thrown into coverage, but so poorly it falls incomplete. If the QB improves his accuracy without improving his decision-making, his INTs will go up.

The other part of "luck" in the stats is due to playcalling. Sometimes Team A gets a TD because they hit the right play at the right time against a defense. Some playcalling advantage is due to superior analysis, but some is just dumb luck. The stats say it was skill, but you wouldn't want to bet on it happening again.

by Rich Conley (not verified) :: Wed, 11/29/2006 - 5:51pm


Thats the same argument I've been making for a while when people are like "BUT OMG, DVOA SAYS TEAM A IS BETTER THAN TEAM B"

Its a tool, and a good one, but its by no means even close to perfect.

by b-man (not verified) :: Wed, 11/29/2006 - 5:54pm

GSfRB: Thanks. ARI is on the charts but missing from the legend. The bottom three were cut off in the legend
VOA - bottom three are ARI, OAK, TB
DVOA - bottom three are OAK, DET, ARI
WDVOA - bottom three are TB, ARI, DET
I'll try to fix 'em later.

by stan (not verified) :: Wed, 11/29/2006 - 5:58pm


Couldn't you do regression analysis on offensive, defensive and ST DVOA with record to see. If a pattern developed over the last 6 years of data -- you might have certain exceptions such as -- when a team ranks #1 (or top 3 or whatever) in offense and has a D ranked lower than X, ST percentage is cut by Z% in figuring team DVOA.

by stan (not verified) :: Wed, 11/29/2006 - 6:05pm


What I mean is this -- I assume that the percentages assigned to O, D, and ST in computing team DVOA are a best fit of the data for ALL teams. But there is no reason to assume that special teams has the same impact for each team (my example of Colts-Chiefs is the obvious one, but we could come up with ones where ST are MORE important).

Is there any analysis which might show whether teams which are greatly unbalanced do better? Or worse? I.e. does a serious lack of balance require a different % for each of the three areas of the game to adequately explain a team's record?

by Andrew (A.B.) (not verified) :: Wed, 11/29/2006 - 6:27pm

Re: 134

I thought it was the other way around -- teams that were successful on 3rd downs were more likely to maintain that success than those who were more successful on 1st and 2nd downs.

by Wanker79 (not verified) :: Wed, 11/29/2006 - 6:28pm

Re: 140

I posed that exact question to Aaron through the contact form. Hopefully it'll end up in a mailbag or something.

by andrew apold (not verified) :: Wed, 11/29/2006 - 6:33pm

55 - customjerseybuilders has something more like what I think you want... I'm ordering my shirt, but would also like something in the jersey variety.

I used their customization tool to come up with http://agh2o.org/pix/robopunter.jpg

unfortunately I think there's like a minimum order of a dozen...

by jonnyblazin (not verified) :: Wed, 11/29/2006 - 6:34pm

Re: 121
That seems pretty counterintuitive to me, especially since the last two dominant Ds that won superbowls (Ravens, Bucs) employed the conservative offense/dominant defense (+ good special teams) mix.
I think with a dominant D the way to win is play the field position game. When you commit turnovers, you give the opposing team better field position where the D is less effective. A dominant D will be infinitely more effective when an opponent is pinned back rather than around the midfield area simply because at midfield offenses generally have a bigger play selection (with deeper drops and routes, flea-flickers, etc) because a turnover is less costly in that position.

by jj (not verified) :: Wed, 11/29/2006 - 6:42pm


Obviously unbalanced teams like the Colts stink it up in the playoffs.

A team that is unbalanced towards defense has a good chance in the playoffs.

However, Polian spends every waking hour relentlessly trying to change that formula to suit his own needs.

by NewsToTom (not verified) :: Wed, 11/29/2006 - 6:47pm

Re #141
Nah, #134 has it right. Teams that are good on first and second down and bad on third down are likely to improve the next year, all things being equal. Things can of course change; see the 1998 DVOA ratings and commentary section on the Rams--unsurprising defensive improvement judging by DefDVOA by down splits and moderately surprising offensive improvement based largely on upgraded offensive skill position players.

by Sean (not verified) :: Wed, 11/29/2006 - 6:51pm

Teams that are dominant in any area, offense or defense tend to do better than teams that are balanced. Yes, there are the Ravens and Bucs on defense, but there are just as legitimately teams like the 97-98 Broncos, the 99 Rams, the 02 Raiders and the 06 Seahawks. The important thing is to have a unit that simply overwhelms the opposition.

Sure, it's great if you're balanced and among the league leaders in both offense and defense, but the notion that a "balanced" team has some sort of inherent advantage over a team that is loaded on one side...well, that leads to people putting money on the 2000 Giants in the Super Bowl.

by Bobman (not verified) :: Wed, 11/29/2006 - 6:54pm

Stan and Rich,

What beautiful back and forth on the luck issue and DVOA. I agree also with Stan's #3 comment on the relative weighting of O, D, and ST, but concur with Rich that there's no practical (or objective?) way to apply this variable weighting. The 2003 Indy/KC game being a great example. How about last year's Indy/Cin game, or an Indy/GB game from the year before. Those "last team with the ball" games.

But it's hard to eliminate or reduce the importance of ST: taking the team I know best, Indy, which might seem like a likely candidate. This year they have greatly benefitted from other teams poor ST showing (kicking, numerous opponents) have nearly been done in by other teams return games (NYJ), and have greatly benefitted from the return of Wilkins in their own return game--they have at least one win to credit to him (JAX), as well as AV's consistent kicking.

Does this merit more or less weighting than DVOA currenly apportions? I don't know, but for one of the teams that comes to mind when somebody says "ST don't matter" (or matters less), I'd have to say when I think twice about it, ST matters quite a bit.

Rich, I love your idea of on-side kicking in those games--even if you get 1 out of 5, that means you might get the chance to win 42-35 because you deprive them of one drive and take the ball yourself.

How many Colt fans see an opponent on the move with 5 minutes left and think "I hope they score and leave enough time for Manning to win it in the end."? Last year I did not have this problem, but in 2000, 2002, 2003, 2004, 2006 the thought crossed my mind more than once. The onside kick is just the logical extension of that thinking. Talk about a coach throwing his D under the bus, though... not good for motivating them. What would some of the more outspoken D players say to the media on that issue: Warren Sapp: "Well,let me see, the coach gives them the ball at the 50 and a short field, and you're asking ME why we let up so many touchdowns...? You're asking the wrong guy."

by stan (not verified) :: Wed, 11/29/2006 - 7:01pm


I totally disagree. The Colts' losses in the playoffs (since 2003 when I started paying close attention to them) have all been basically due to their problems in the offensive line and a receiver corps that isn't at all physical. It isn't because their offense is far better than their defense. In fact, the defense hasn't been that terrible in the losses the last 3 years.

The Colts are a lot like Spurrier's fun and gun in Florida. If they get a mismatch or the slightest edge, they usually exploit it for big numbers. The Colts don't need great protection, just a bare minimum. They don't need wide open receivers, just a little window. Over the last few years, there have been a number of teams who actually played very good defense against them, yet gave up a lot of points (believe it or not, the 41 NE gave up in 2005).

But they are vulnerable to a strong, physical defensive front because they can't control a game with the run and they don't pass protect anywhere near as well as people think.

Joey Porter called out the Colts for being soft. They are. And the Pats beat them in 2003 and 2004 and the Steelers in 2005 using the same formula. The same one used by the Chargers last year.

by Bobman (not verified) :: Wed, 11/29/2006 - 7:03pm

Stan and Wanker79, I too posed that question to the FO gods a month or so ago, when Indy's O was freakishly high yet their non-DAVE DVOA was pretty ordinary (13th?) and they were one of two undefeated teams. Clearly something was amiss. IIRC, Aaron's view was that it was their record that was "wrong," not their O DVOA or overall DVOA, and taking a super-over-weighted aspect whether it's a +40% O, a -50% D or super ST play (calling ROBO-PUNTER) was not the answer.

by stan (not verified) :: Wed, 11/29/2006 - 7:07pm


FG kicking is not part of what I'm talking about with respect to ST mattering less. I only meant it with respect to the field position game.

And even if there isn't any practical way to quantify it in DVOA, it is worth thinking about when DVOA seems to be "off".

by Bobman (not verified) :: Wed, 11/29/2006 - 7:12pm

Stan #136
Regarding your final comment, that "luck" is what the Colts' whole O is predicated on: Manning putting them into the exact right call for the right D looks (and why the confusing, movement-filled 3-4s tend to diminish their "luck"/success). So I'd say no, there is no luck in that for them. Maybe other teams with "the first scripted 15 plays" or rookie QBs, getting the right play is largely luck, but for most teams with a vet QB, I think that is the goal rather than an "oh hey, lucky us" type of event.

For Indy I'd bet on it regularly. Probably NE and Cin too. Wouldn't you?

by Wanker79 (not verified) :: Wed, 11/29/2006 - 7:12pm

Re: 150

I was thinking more along the lines of weighting the offense and defense by the number of plays they are a part of. And for ST, you'd weight it by the number of drive that the the distinct units (FG, Punt, & KO) are a part of.

by Rich Conley (not verified) :: Wed, 11/29/2006 - 7:13pm

The issue I see with weighting ST/O/D differently depending on teams values for each, is that (in addition to adding a ton of complexity) if you start weighting differently, the currently coefficients for pretty much everything are going to need to be changed. If team DVOA correlates across the league to point differential, then simply changing the weighting of each part in a team may make DVOA initially look less accurate, but may in fact be more accurate, or it could be the other way around.

Theres also the situational questions : IE in that KC-INDY game, a lot of the reason that the special teams of Indy didnt matter was because KC's D was bad. If Indy was playing the Ravens, ST may matter more.

by Wanker79 (not verified) :: Wed, 11/29/2006 - 7:16pm

Re: 153 Follow-up

To clarify, I was talking about the percentage of plays/drive for the specific team, not the league as a whole. I re-read what I had written and figured that may have been a little ambiguous.

by Wanker79 (not verified) :: Wed, 11/29/2006 - 7:26pm

If Indy was playing the Ravens, ST may matter more.

I think that would work itself out. If a team only plays a couple games a year where their ST matter, that team's overall value for the year won't be effected by their ST all that much.

by Rich Conley (not verified) :: Wed, 11/29/2006 - 7:34pm


Yes, if you weight by total amount of plays, it very well may.

(I think we're talking just KO/KR P/PR here, right?)

by b-man (not verified) :: Wed, 11/29/2006 - 8:05pm

I fixed the legend so all teams show up now in the first three charts. If I get some time I will show the OFF/DEF/ST on the game charts.

DVOA Charts

by jgm (not verified) :: Wed, 11/29/2006 - 9:00pm

re:DVOA charts: I have been wondering for some time how to use the DVOA in a convenient way to predict how a team is really doing as the season progresses. Your DVOA charts do it for me. The teams with the steepest positive slope are probably going to play above their DVOA and vice versa. Great work!

by stan (not verified) :: Wed, 11/29/2006 - 10:21pm


Whoa, Bobman. I never said that the Colts are "lucky". I was trying to make the point that there is a degree of luck in every DVOA calculation and it isn't just the bounce of the ball. Team A happens (not due to scouting or analysis) to run a play that gashes Team B's defensive call and DVOA has no choice but to attribute that to ability. In reality, it was a fluke and DVOA won't be as accurate in its predictive capability because of it.

Anyone who has read my posts over the years knows that I think that Manning reads defenses and coverages better than anyone who has ever played the game. I think he dominates games more with his mind than his arm. For example, I thought that the Philly game was one of the most effective pieces of evidence of his dominance of the NFL than any other game this year.

It ain't luck.

by jonnyblazin (not verified) :: Wed, 11/29/2006 - 11:06pm

re: 158
These charts rock! Long live b-man!

by paytonrules (not verified) :: Thu, 11/30/2006 - 2:31am

You Oregon Trail fans need this shirt:


by Stiller Fan in Austin (not verified) :: Thu, 11/30/2006 - 2:39am

I just wanted to say that the Steelers' polynomial game DVOA fit is utterly amazing. They'll be around -200% by the end of the season...

Why couldn't Ben have led with his throwing arm instead of his head in his motorcycle misadventure? It would have been less painful for him and Steeler fans...

by DavidH (not verified) :: Thu, 11/30/2006 - 2:41am


where are you getting weekly dvoa game scores?

by DavidH (not verified) :: Thu, 11/30/2006 - 2:47am

also, b-man:

the increased spread of values at the beginning of the year makes it harder to figure out trends in the league charts. it might be interesting to see similar charts, but with z-scores plotted instead of actual dvoa values. (i hope i am using the right word. a z-score is the number of standard deviations from the mean that a datum is, right?)

and as long as i'm mentioning things that i would do if only i knew how... team fit lines using only the first 11 weeks would be cool. then we could see whether week 12 was predicted at all by the trends.

these graphs are wicked rad, man.

by Boston Dan (not verified) :: Thu, 11/30/2006 - 6:58am

Yes, Bman, thank you for posting those graphs. Chicago's linear shocked me a little bit.

by Yaguar (not verified) :: Thu, 11/30/2006 - 7:51am

Wow, check out the Lions' game DVOA graph.

Fire Millen.

by b-man (not verified) :: Thu, 11/30/2006 - 9:45am

I updated the top 12 teams game graphs with the OFF/DEF/ST. I'll do the rest later.
DavidH: Game DVOA was calculated as the value that the current week would be to get the posted DVOA.
DVOA_game=n*DVOA_cur - sum(DVOA_game_prev)
Its not going to be perfect since the first 4 weeks have no defense adjustment and then changes to adjustments for each week were not back applied.

by Bruce (not verified) :: Thu, 11/30/2006 - 11:26am

Aaron, have you considered renaming "weighted DVOA" to "Janet Jackson DVOA"? This would be a reference to her tune, "What have you done for me lately?" Of course with JJ, the double entendre abounds.

by Wanker79 (not verified) :: Thu, 11/30/2006 - 11:31am

Re: 157

Yes, if you weight by total amount of plays, it very well may.

(I think we’re talking just KO/KR P/PR here, right?)

Actually, I'm suggesting that the offense and defense portions of the TOTAL DVOA be weighted by the percentage of plays each unit accounts for. The way I figure it, if a particular team's defense is involved in many more plays than their offense, the quality of the defense should be more important. And vise versa. As for the STs, you'd weight each of the three distinct units (FG, Punt, KO) by the number of drives each unit is involved. If a particular team never scores, then the Punt portion of their ST would be much more important to the team than the FG or KO portions. Or if a team scores at will and hardly ever punts, the quality of the punt portion isn't as important.

I'm mot very confident about my suggestion to weight the offenses and defenses. The relationship between the quality of the offensive and defensive units and the time they spend on the field is pretty complicated. If either unit is very good they should be involved in less plays (quick scoring drives for the offense and 3-and-outs for the defense). But by weighting them by the percentage of plays, that would mean that the very good defenses and offenses would count less towards their overall DVOA, which doesn't make any sense. But I see no reason why you shouldn't weight the three ST units by the number of drives they are involved in.

by Andrew (not verified) :: Thu, 11/30/2006 - 11:59am

Wanker179 #170:

Actually, I’m suggesting that the offense and defense portions of the TOTAL DVOA be weighted by the percentage of plays each unit accounts for. The way I figure it, if a particular team’s defense is involved in many more plays than their offense, the quality of the defense should be more important.

To me, this appears to be a weakness with DVOA when I am looking at the Eagles.

The Eagles have a good defense that just cannot get off the field (tied for 6th in yards per play, but near the bottom in coughing up 1st downs). So although they are pretty good on a per play basis, they've faced more plays then any other team except only the Titans. Not a recipe for success.

by SGT Ben (not verified) :: Thu, 11/30/2006 - 12:01pm

I'm honestly surprised that my Cleveland Team is as high as it is! (28th, sadly enough.)

What's really odd is that it's our ST (4) that's keeping us from being the absolute bottom dweller, (thanx Dawson for being so freaking reliable!)

I can't even think of something remotely amusing/thoughtful/positive to say regarding all this after last weeks schellacking at home against our state-rivals!

by b-man (not verified) :: Thu, 11/30/2006 - 12:05pm

Updated all game graphs for OFF/DEF/ST:
DVOA Charts
WDVOA uses the VOA+proj then DAVE then WDVOA. Kind of cool to see how the projections matchup against the current.

by Pat (not verified) :: Thu, 11/30/2006 - 1:04pm

The way I figure it, if a particular team’s defense is involved in many more plays than their offense, the quality of the defense should be more important.

Except this isn't something that's controllable by the team itself. It's dependent upon the other team and the way the game goes.

Special teams really is the epitome of that problem: you don't know how many kickoffs, punts, or field goals a team will attempt. Indy, in 2004 and 2005, barely attempted any field goals, and yet they counted equally in DVOA.

I don't really think you can change the weighting of each component like that - I think you'd end up boosting agreement with, say, previous wins, but not helping a correlation with future wins.

by Sid (not verified) :: Thu, 11/30/2006 - 1:55pm

RE: 108

A loss to Dallas won't end the Giants' season. They'd still be very much in the mix for a wild card in the NFC. A loss to Dallas would mean they'd finish 2nd in the NFC East, that's all.

Something I noticed that's strange is that the Giants have a weighted DVOA that's just about the same (.1%) difference as their overall DVOA, despite bombing in recent weeks. Were the early games that now have less weight bad games for them DVOA-wise or something?

by DavidH (not verified) :: Thu, 11/30/2006 - 3:56pm

Wow, looking at those updated charts I'm struck by how much correlation there is between the offensive and defensive numbers. I mean, I've only looked at it, not calculated anything, but it seems like a lot of the time a team improves or declines in BOTH aspects from week to week. I'll have to take a closer look to see if this is true.

by b-man (not verified) :: Thu, 11/30/2006 - 4:34pm

DavidH: These look like they have OFF/DEF going in different directions:
SD - OFF rising, DEF declining, ST steady
NO - OFF rising, DEF/ST declining
CAR - OFF declining, DEF/ST rising
MIN - OFF rising, DEF declining, ST steady
WAS - OFF/ST declining, DEF rising

by DavidH (not verified) :: Thu, 11/30/2006 - 4:45pm

I mean on a weekly basis, not the overall picture. I'm curious as to whether merely having one of your units perform better lifts the DVOA of BOTH units due to incomplete separation of the two aspects.

I took a look at each team's change in offensive and defensive game results from week to week, and marked which of these 3 things happened:

A) Same - both offense and defense played better than the week before OR both played worse than the week before
B) Different - offense played better and defense played worse OR vice versa
C) Neither - one of the units played about the same

So there were then 32*11 = 352 observations of improvement/decline, and here's what I got:
Same - 166
Different - 136
Neither - 50

That's not nearly as much covariation as I thought I had observed. It must have just been me only noticing the graphs that fit my original idea.

by Sid (not verified) :: Thu, 11/30/2006 - 4:49pm

Interesting game Sunday. Atlanta at Washington. DVOA indicates that Washington has easily been the better team this season. Plus, it's a SYSTEM game.
I would tend to lean Falcons, however. Only thing that would tell me no is that they look like they're imploding.

by Wanker79 (not verified) :: Thu, 11/30/2006 - 4:52pm

Re: 174

I don't completely agree that the time a particular team's defense spends on the field is something out of their control. A very good defense should force the opponent's offense off the field. If they can't, then the defense isn't as good as the opposing offense and they should be punished in DVOA for that. Where is gets tricky is that the perfect defense would be one that forces a 3-and-out on every drive; therefore accounting for very few plays and consequentially (if weighting the offensive and defensive units as I suggested) they would for very little of the unit's weighted value. So by weighting a unit according to the number of plays they participate in you'd be undervalueing the very good offenses & defenses, and over-punishing the bad ones. The more I think about it, the more I convince myself that weighting the offensive and defensive units by the number of plays they participate in isn't a very good idea.

On the other hand, the more I think about weighting the individual ST units (FG/KO/Punt), the more I like the idea.

by MCS (not verified) :: Thu, 11/30/2006 - 6:12pm

The last two jerseys my son got for the Packers are:

Darren Sharper two years ago.
Samkon Gado last season.

For a small fee, I can probably be persuaded to get a Millen jersey for him.

BTW, we won't let him get another Packer jersey.

by FantasyStooge (not verified) :: Thu, 11/30/2006 - 7:14pm

Re: Yaguar and others:
If someone in a Detroit theatre yells "Fire", do the other patrons have the right to yell "Millen"?

by Pat (not verified) :: Thu, 11/30/2006 - 7:49pm

A very good defense should force the opponent’s offense off the field.

Yes. But said defense has no control over how long their own offense is on the field.

On the other hand, the more I think about weighting the individual ST units (FG/KO/Punt), the more I like the idea.

Yeah, but you can't do that either, because you can't expect that just because they didn't kick many field goals in, say, the first half of the season, they won't in the second. Or punt, or, etc. There may be certain classes of teams where this is true, but, now it's starting to get silly. It's not like you'd ever be able to come up with a single number that describes the strength of a team perfectly anyway.

by jgm (not verified) :: Fri, 12/01/2006 - 9:18am

b-man: your graphs made picking Cin last night much easier. Now, after comparing your graphs, how about Buf over SD, Was over Atl, Ind at Tenn too close to call, Minn over Chi, Miami beats Jac, Sea beats Den, Dal over NY, Car over Phi, SF beats NO?

by jgm (not verified) :: Fri, 12/01/2006 - 9:31am

I'd also add that Arizona is ascending as St Louis has been falling so it's Ari over StL and TB should have a great chance against Pit. The graphs would predict Den over Sea except that with the return of Hasselbeck and Alexander I just couldn't go there.

by chris clark (not verified) :: Fri, 12/01/2006 - 11:34am

re 183:

One could analyze it and see whether FG attempts is correlated or not. Fumble recoveries were shown to be uncorrelated, but FG attempts may or may not be correlated. Until someone looks at the actual data, it is all speculation.

However, to use the "Pats right" common knowledge--it is quite unlikely one number will ever capture everything about a team's strength. Still, as discussed many times before (and I'm sure Pat agrees), DVOA is one of the best metrics we have for approximating that knowledge. One just has to temper it with things one knows that DVOA can't capture (e.g. the importance of McNabb's injury, the RPS relationship between the Colts, Pats, and Broncos, and any other goat entrails that you may read).

I say this as a Den fan, who is saddened by their DVOA rating this year, but has seen on far too many occasions why that rating is justified. I am only mildly heartened by the predictions that Den has a better shot at making the playoffs than some of their competition (e.g KC, Jac, Cin, NYJ), especially since the last time I ran the playoff prediction software with some win-out assumptions but have Den lose to SD, they don't make the playoffs. The dramatic from 1st to 3rd in their division could have been predicted from the DVOA numbers, and the numbers don't suggest that they will climb back to the top spot. Those 2 losses also cause the downturn in the DVOA regression line/curve, which is further cause for gloom. Rooting for SD may have to be sufficient consolation. That seems more secure than expecting Jay Cutler to be another Tony Romo.

by Andrew (not verified) :: Fri, 12/01/2006 - 11:39am


Where is gets tricky is that the perfect defense would be one that forces a 3-and-out on every drive

ROBO-DEFENSE would intercept every pass attempt thrown and return it for a touchdown whenever possible based on field position and player location, and would force a fumble on every sack or rush, recovering at least the average of 50% of them. 3 and out would be rare, because against ROBO-DEFENSE the opposing offense would rarely run more than 2 plays, 3 tops.

The best strategy against ROBO-DEFENSE would be the first down punt, in hopes of pinning their team against the goal-line.

Seriously though, due to turnovers, the defense would have to average a bit under 3 plays per drive under the perfect 3 and out scenario. The arguably greatest defense ever - the 1991 Eagles squad - allowed just 905 total plays in the year - 383 rushes, 467 poasses, 55 sacks - that works out to 56.6 per game. Given the typical number of drives per year (roughly, 180, with a variance of +/- 10-15, and better defenses typically facing more drives), this works out to around 4.75-5 plays per drive by the 91 Eagles squad.

by ugarte (not verified) :: Fri, 12/01/2006 - 6:37pm

New Orleans is clearly ranked too low because Drew Brees is dreamy. People Magazine's Sexiest Man Alive is way better than this. The Saints Rule!

by jebmak (not verified) :: Sat, 12/02/2006 - 1:59pm

Wow b-man, nice job!