Writers of Pro Football Prospectus 2008

Most Recent FO Features

HackenbergChr15.jpg

» SDA: From Death Valley to Happy Valley

While Saturday won't give us an answer to which is the best one-loss team, it is still a big week for conference races across the country.

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

1
by Smeghead (not verified) :: Tue, 11/14/2006 - 6:03pm

Seattle: #22 in DVOA.

85.6% to make the playoffs.

Giggity-goo.

2
by Smeghead (not verified) :: Tue, 11/14/2006 - 6:05pm

P.S. -- Man, are you gonna get it for having Philly #1.

3
by jebmak (not verified) :: Tue, 11/14/2006 - 6:09pm

Finally!

Re:#2
Philly will still be #2 the rankings on foxsports are based on weighted.

4
by TonyFranklin (not verified) :: Tue, 11/14/2006 - 6:10pm

So will Bears fans blame the ranking drop on the field goal return not being considered in DVOA?

5
by Charles Jake (not verified) :: Tue, 11/14/2006 - 6:12pm

PHI is clearly ranked too high because the liberal media desperately wants a black QB to succeed. The NFC won-lost records are way better than this. All Rex Grossman does is win games! Bear Down!

6
by Wanker79 (not verified) :: Tue, 11/14/2006 - 6:22pm

Wow, the "tiering" is almost completely gone.

Linear-trendline R^2 = 0.9884

7
by INT (not verified) :: Tue, 11/14/2006 - 6:24pm

Jacksonville is 1st in variance.

I'm shocked.

8
by jonnyblazin (not verified) :: Tue, 11/14/2006 - 6:25pm

Ouch, the Ravens got significantly readjusted for their 'clutch' win vs the lowly Titans, especially their defensive rankings. Could Ray Lewis be more important to that defense than most people think?
And SD's defensive ranking is 0.0? What happened? They really need to be injected with some enthusiasm.

9
by Richard (not verified) :: Tue, 11/14/2006 - 6:29pm

Is Estimated Wins out of ten games or nine games at this point?

10
by Matt (not verified) :: Tue, 11/14/2006 - 6:29pm

Eagles have beat the Cowboys, Texans, Packers, 49ers, and Redskins. And they're #1?

When they loose to the Indy, Carolina, and either Atlanta, Dallas, or the Giants, they won't even make the playoffs.

11
by Matt (not verified) :: Tue, 11/14/2006 - 6:33pm

NYG managed to lose by 18 and pass SD who won?

Time for some more adjustments to DVOA

12
by Richard (not verified) :: Tue, 11/14/2006 - 6:33pm

8: Merriman is hugely important to the San Diego defense. So is Luis Castillo. They were both missing on Sunday. Carlos Polk is no Merriman and Phillips was coming off a leg injury. The secondary is better than it was, but the front seven was generating no rush.

13
by Ray (not verified) :: Tue, 11/14/2006 - 6:35pm

It's probably a good thing that the Eagles will be listed behind the Bears in the Fox rankings (a 5-4 team at #1?), but the actual difference for WDVOA is only 0.2%. They're statistically equal. Wow. If the 2nd half of the season is as lucky for the Eagles as the first half was unlucky... No, I have to stop. I can't get my hopes up again.

I can't wait for Phl at Ind. That should be a very evenly matched game. That's where the Eagles need more of that karmic payback.

14
by TGT (not verified) :: Tue, 11/14/2006 - 6:37pm

Matt is the troll of the day.
Richard, estimated wins should be out of 9 this week.

15
by Andrew (not verified) :: Tue, 11/14/2006 - 6:38pm

Matt #10:

After watching how the Panthers played at home against the Bucs, you really think the Eagles will lose at home to the Panthers? The Panthers could barely get a first down all game!

16
by Vince (not verified) :: Tue, 11/14/2006 - 6:40pm

Boy, that playoff odds report shows how boring the finish to the AFC is going to be, with 5 playoff spots pretty much already wrapped up, and Jacksonville and the Jets (wait -- THE JETS?!) battling for the final wild card berth.

17
by michael (not verified) :: Tue, 11/14/2006 - 6:41pm

#11: you simply cannot count 7 of those 18 points (from the 108-yard kick return) when talking about DVOA - as aaron mentioned in Audibles (or elswhere) it's so random, it's almost impossible. Otherwise, NYG played a tight game against the best team in the league (DVOA-wise), whereas SDG had their defense shellacked by a middling offense at best, hence the jump. This isn't the Harris Poll, here.

18
by theory (not verified) :: Tue, 11/14/2006 - 6:42pm

I've been noticing something in DVOA I don't quite grasp - offensive DVOAs seem to be "higher" than defensive DVOAs of similar rank, so having the #1 offense and #24 defense gives you a much higher overall DVOA than having the #24 offense and #1 defense. Is DVOA supposed to favor teams with strong offenses and weak defenses? Should the Chargers and Colts really be favored over the Ravens, Broncos, and Jags in the AFC? (I'm a Charger fan who'd like to believe the defense won't betray them in the playoffs).

19
by David (not verified) :: Tue, 11/14/2006 - 6:42pm

Chicago's offense is net negative now? Ouch.

Also, Philly is a full game above anyone else for estimated wins? Ye gods.

20
by turbohapy (not verified) :: Tue, 11/14/2006 - 6:42pm

SD had a little more fumble luck than NYG and gave up a ton on defense and didn't get that much credit for all the points scored since Cincy's defense is pretty poor. Plus the "freak" FG return. Still seems pretty odd though.

21
by Jim (not verified) :: Tue, 11/14/2006 - 6:44pm

The Indy-Dallas game this weekend looks better and better.

13: Eagles-Colts should be interesting, but the Eagles are one team that wouldn't seem to match up well against Indy. You can run all day through the middle of that defense, except that the Eagles don't commit to the run and don't go through the middle when they do run.

22
by Ray (not verified) :: Tue, 11/14/2006 - 6:45pm

RE# 11 Matt
NYG managed to lose by 18 and pass SD who won?

Time for some more adjustments to DVOA
When considering DVOA, you can't just look at the score. NYG lost by 18, but 7 of those points were the result of a non-predictive event (the FG return for a TD) that is not factored into DVOA.

Also, while SD won, they did it by giving up a TON of great plays to Cin, while they ran all over Cin's weak run D. So basically they get punished more for their poor defense than they get rewarded for their good offense because of opponent adjustments. On top of that DVOA factors in the first half of the game where the SD offense didn't produce very well.

Take a look at the Non-Adjusted VOA numbers (that don't factor in opponent strength). SD has a healthy lead over NYG there.

23
by Wanker79 (not verified) :: Tue, 11/14/2006 - 6:48pm

Re: 18

You may be forgetting to factor in Special Teams.

24
by doktarr (not verified) :: Tue, 11/14/2006 - 6:49pm

I enjoy trying to imagine the more absurd scenarios in Mike Harris's playoff projections, like the 22 times out of 15,000 when Indy misses the playoffs, or the 3 times when Arizona makes it. I imagine people using the word "swagger" when describing the Cardinals.

As Aaron notes, things are probably not quite as predictable as that table implies, because injuries can cause teams to decline more precipitously. But it is great to have that tool.

25
by Jerry (not verified) :: Tue, 11/14/2006 - 6:50pm

Why does the top 5 DVOA listing have Chicago at 1/Philly at 2 and the main table has it reversed? Shouldn't Chicago with 35.5 still be at 1 on both?

26
by jebmak (not verified) :: Tue, 11/14/2006 - 6:50pm

#18

Actually, if you ignore the Colts and their ridiculously good offense, the next 11 or so teams (O vs. D) match up almost percent for percent (I only checked the top 12).

26
by turbohapy (not verified) :: Tue, 11/14/2006 - 6:50pm

Re: 18

This is not always the case. In 2000, Baltimore's defensive rating was higher than any offense. I would say it's reflective of the fact that there aren't any defenses this year that are CONSISTENTLY dominant.

28
by Richard (not verified) :: Tue, 11/14/2006 - 6:51pm

For table with z-scores... click my name.

29
by Joe (not verified) :: Tue, 11/14/2006 - 6:53pm

Re: #21 - During the Eagles-Redskins game, Philly ran up the middle consistently when they needed to (2/3 of the 4th quarter). They also ran quite well up the middle against Tampa Bay. For the most part, the Eagles don't run up the middle because they believe the offense will be more effective by not doing so. FWIW, DVOA seems to agree with the results.

30
by Brandon (not verified) :: Tue, 11/14/2006 - 6:54pm

Unfortunately, I become more skeptical of DVOA every week. I think it's generally a great tool, and helps correct a lot of misperceptions. Too bad it seems deeply flawed and so counterintuitive in certain ways. As mentioned, it's goofy for the Giants to pass the Chargers by losing. I understand the principal, don't reflexively call me a moron - it just seems to me that there is too strong of an adjustment made on past performances of teams each week. Admitedly, I don't really know how to correct this.

31
by Ray (not verified) :: Tue, 11/14/2006 - 6:55pm

re #21 Jim
Eagles-Colts should be interesting, but the Eagles are one team that wouldn’t seem to match up well against Indy. You can run all day through the middle of that defense, except that the Eagles don’t commit to the run and don’t go through the middle when they do run.

The Colts aren't only weak against the run. They're also weak against passes to #2 WRs, RBs and TEs. The Eagles have a good #2 WR (whichever of Brown or Stallworth happens to be #2 that day), a good TE (Lewis, when he decides to hang onto the ball) and a GREAT pass catching RB in Westbrook. Even without pounding the ball very much (which, looking at the Was gameplay they just might do) they should have success in the passing game.

The question is whether they'll be able to slow down Peyton at all. If the pass rush can pick up some steam like it had in the beginning of the year and stop Manning from getting the ball downfield, the Eagles have a shot of keeping up.

32
by Pacifist Viking (not verified) :: Tue, 11/14/2006 - 6:57pm

It's sort of funny just how many non-predictive events happen--which is one reason I am skeptical about DVOA. When something doesn't fit into the system, it's called "non-predictive" or "luck"--but non-predictive events and hazard are affecting outcomes every week.

I'm not making a direct criticism of DVOA, but expressing a Dostoevsky-like skepticism of mathematical systems and their ability to predict reality.

33
by Bobman (not verified) :: Tue, 11/14/2006 - 6:58pm

Good to see Indy is about where I thought they should be all along, and essentially tied for third with est wins. Last few weeks they looked to low; I am surprised that a squeaker over a fairly ordinary Bills team helped them at all, even though I realize teams ahead of them did some falling and standing still to help out too.

But I have a question about future sked: Indy's is now ranked #7 when last week it was #15. That's a big leap and makes their sked look tougher, but really it's the same teams, so what gives? Phi moved up a bit, Jac is the same, Cin and Dal each moved up a hair... is that what it takes to bump the SOS from 15 to 7? Actually, I see Ten and Hou also moved up two each, but STILL they're in the cellar.

MY GOD, look at CHI and SD skeds. SD gets karmic payback for last year's killer sked, but CHI has feasted on the lowly for two straight years now--and this year they are a returning div champ! Ah well, as an Indy fan, I won't complain about anybody else's easy sked. I still say Indy's is pretty cushy from here on out. The CIN game looks to be a barn-burner--somebody had a great line yesterday in the Quick Reads thread about what does a runner in that game have to do to get positive PAR--run for 200 yards against those run D's?

34
by Andrew (not verified) :: Tue, 11/14/2006 - 6:58pm

Jim #21:

13: Eagles-Colts should be interesting, but the Eagles are one team that wouldn’t seem to match up well against Indy. You can run all day through the middle of that defense, except that the Eagles don’t commit to the run and don’t go through the middle when they do run.

Eagles - #1 rushing offense by DVOA. Eagles, #2 rushing offense by yards per attempt (trailing on Atlanta), #2 rushing offense up the gut (Middle/Guard rushes) - 4.89 yuards per attempt. #1 running back in yards per attempt: Brian Westbrook of the Eagles (#5 in DVOA, #8 in DPAR). Eagles run up the gut 48% of the time, and to the left end (Freeney) only 14% of the time.

Ps. the Colts also don't have much of a passing defense.

I don't know, do you even try checking up on what you write?

35
by Ray (not verified) :: Tue, 11/14/2006 - 7:01pm

How about looking at it this way. When you wonder how NYG passed SD by losing, remember that DVOA doesn't care about how many points a team scored or whether a team won or lost. DVOA only cares about how successful a given play was. Was X a successful play based on situation and opponent? If yes, then DVOA goes up. If not, then DVOA goes down.

DVOA doesn't tell you who won a game, point scored does that. Points scored doesn't tell you who played better in a game, DVOA does that.

36
by David (not verified) :: Tue, 11/14/2006 - 7:01pm

The point was not that the eagles have a bad rushing attack between the tackles. It's that they don't use it much. Hard to argue that part, I think.

37
by Wanker79 (not verified) :: Tue, 11/14/2006 - 7:04pm

Here are the top-heavy Division averages: (2*(1st+2nd)+3rd+4th)/6

NFCE....24.73%
AFCW....11.07%
NFCN....7.30%
AFCN....7.02%
AFCS....6.15%
AFCE...-2.35%
NFCS...-3.48%
NFCW...-14.28%

Wow. So the NFCEast is a mile better than everyone else. The AFCW is the best of the rest. The NFCNorth, AFCNorth, and AFCSouth are basically equal. The AFCEast and NFCSouth are equally crappy. And the NFCW is abismal. That feels pretty accurate to me.

The average Weighted DVOA for the two conferences are basically equal (NFC=-0.79% AFC=0.99%) and the top-heavy averages gives a very slight advantage to the AFC (6.10% compared to 4.80%).

38
by navin (not verified) :: Tue, 11/14/2006 - 7:04pm

All the bad teams seem to be improving. Both SF and Houston's defensive DVOA has jumped a lot in the past two weeks.

Also, San Fran two games behind Seattle with four games left to play in the division. I wish I could dream that SF could somehow play mediocre for the rest of the year and reach nine wins.
Wins: SEA, @STL, GB, ARI, @DEN--Denver resting week 17
Losses: @NO, @SEA

Imagine if this pipe dream came true! I think Arizona in 1997 was similarly horrible and made the playoffs and even beat Dallas in the first round. This is what reaching last year's win total in nine weeks does to a fan.

One concerning trend is the decline of the offense, especially the passing game. Alex Smith looks like he is regressing as the season goes on. I hope Vernon Davis' return to the lineup will open it up a bit. Opponents had been planning for his speed up the middle until he got injured, so hopefully his return opens things up a bit.

39
by B (not verified) :: Tue, 11/14/2006 - 7:04pm

17: The Cincy offense is ranked 5th, though. They have a lousy defense, but their passing game is usually considered to be one of the best.

40
by David (not verified) :: Tue, 11/14/2006 - 7:05pm

35: Specifically, it tells you who played better in ways that they are likely to repeat with some consistency. That' a facet of DVOA that some people overlook - it's not just about who was better in one game, but who's likely to be better over the entire season.

41
by doktarr (not verified) :: Tue, 11/14/2006 - 7:05pm

RE: 21, 31,

By contrast to PHI, who matches up only OK against Indy, Dallas seems very well suited to playing Indy. They have a good RB tandem and should be able to gash the Indy D for yards, particularly if Sanders remains inactive. Meanwhile, they run the 3-4 defense that tends to give Manning/the Indy O line trouble. Also, that game is in Dallas. On paper, it's a dead heat. Maybe it comes down to a Vanderjagt field goal?

42
by Andrew (not verified) :: Tue, 11/14/2006 - 7:06pm

Pacifist Viking:

I’m not making a direct criticism of DVOA, but expressing a Dostoevsky-like skepticism of mathematical systems and their ability to predict reality.

Over time (say a whole season), non-predictive events like fumble recoveries, critical penalties, touchdowns on returns, and injuries, tend to even themselves out between teams more than they would through half the season.

However, given the relatiely small number of games in a 16 game schedule, 1 or 2 flukey wins or losses can be sufficient to occasionally keep a team out of the post-season that should be in. See San Diego, 2005 edition.

43
by Brandon (not verified) :: Tue, 11/14/2006 - 7:07pm

#32 - agreed. There's an interesting article from Bill James, "Underestimating the Fog", that applies more specifically to baseball, but seems important here as well. Just because something may seems "random" or "non-random" over the course of our observations does not necessarily make it so. There are numerous instances where some may indeed be non-random, but appear to be caused by random factors because it's not sustainable over time. This does not mean that within the temporal occurence of the event that it was "random" and not dictated by the participants or predictable by viewers/coaches.

44
by Bobman (not verified) :: Tue, 11/14/2006 - 7:07pm

Ray #31, Indy is ranked 9th vs opposing teams RBs ion the passing game and 16th (exactly avg) vs TEs. If this is perfectly predictive, anybody would have to hope their #2/3/4 WRs pick up their game.

I think the game will be competitive til the end, but then again, aren't all Indy's games this year, with the exception of HOU #1? I bet even HOU #2 is less than 2 TDs in a few weeks.

Also, Indy with Sanders is clearly a different team vs Indy without. Current FO stats have 7 games without Sanders and 2 with, so if he's in the lineup, they are a different team what what the stats/rankings suggest. If I were Dungy, I might keep him on ice til the playoffs. If Reagor returns from his non-Romanowski-induced broken orbital bone, the DL becomes a little stouter as well. And Freeney had a breakout game vs McNabb a few years ago--multiple sacks and FF strips. Maybe it's the mobile guys (or just 'cuse grads) he's best at catching.

I doublt I'll get to watch the damn thing here in Seattle, however....

45
by Ray (not verified) :: Tue, 11/14/2006 - 7:10pm

RE #36 David
The point was not that the eagles have a bad rushing attack between the tackles. It’s that they don’t use it much. Hard to argue that part, I think.

Again, look back at this past week's game against Washington. 34 runs to 26 passes. Even if you assume all of McNabb's runs were designed passes (4), you still get 30-30, a 50% ratio. The Eagles CAN run, and they will even do it sometimes. However, until the game is played in a few weeks, we can't know if the will against Indy.

46
by NF (not verified) :: Tue, 11/14/2006 - 7:10pm

Am I the only person who thinks it's funny that the 49ers have a 6.9% chance of making the playoffs according to the projection, better than the Dolphins? Or that in 2 simulations the Cardinals won their division? Or that the two teams with the highest chance of being the #4 seed are the Patriots and the Seahawks?

47
by RecoveringPackerFan (not verified) :: Tue, 11/14/2006 - 7:12pm

32: Yes, non-predictive or lucky events have huge effects on individual games, but the bottom line is that most games are decided by the team that plays best in statistically quantifiable ways. Indeed, the fact that the rankings align loosely with W/L records confirms that, in general, the winners and losers on the scoreboard are doing the things the system expects out of winning and losing teams.

48
by MJK (not verified) :: Tue, 11/14/2006 - 7:12pm

Theory (18)

I bet it's because DVOA likes offenses that can sustain drives--e.g. get lots of 1st downs and defenses that keep the other side from sustaining drives (e.g. causing a lot of 3-and-outs). So an offense ranked #1 by DVOA will be on the field a lot, keeping their 24th ranked defense OFF the field. Hence, because the offense sees more snaps, the offensive DVOA will affect the total DVOA more than the defensive DVOA.

On the other hand, the #1 ranked defense will hardly ever be on the field, since they stop the other offense a lot and get off, whereas their #24 ranked offense will see lots of snaps and have lots of opporutnities to suck, despite sucking. So the good defense will contribute less to the total DVOA than it would if the offense were good.

This raises an interesting point about total DVOA, at least if I'm understanding how it works correctly. A good offense may indeed have a greater contribution towards a good total DVOA than a total defense, because the better an offense, the more it sees the field, whereas the better a defense, the less it sees the field.

49
by D (not verified) :: Tue, 11/14/2006 - 7:13pm

#25
The Top 5 listing is Weighted DVOA (where Chicago edges Philly), but the main table is sorted by Total DVOA where the Eagles reign supreme.

50
by Brandon (not verified) :: Tue, 11/14/2006 - 7:14pm

Oh, another thing I forgot to add is that I thought the playoff simulation was excellent. This is the type of situation where DVOA really excels - showing a big picture analysis. It seems a little too granular on a team by team basis, but gives interesting (and likely accurate) league wide results.

51
by Bobman (not verified) :: Tue, 11/14/2006 - 7:15pm

Hey, I just saw that Inty/Phi was moved to prime time in two weeks, so I WILL get to see it. Ya-freakin-hoo!

52
by Ali Nagib (not verified) :: Tue, 11/14/2006 - 7:16pm

Just for fun, I decided to see what the BP Playoff Odds were for the equivilent point in the MLB season for each of the eventual playoff teams(I'm going to say 90 games, as that's about 9/16 of 162):

Tigers - 94.41%
Yankees - 64.87%
A's - 15.93%
Twins - 6.47%

Mets - 99.08%
Cards - 77.91%
Padres - 52.39%
Dodgers - 38.10%

I don't think this tells us anything especially useful, I just thought it would be an interesting exercise. And it was!

53
by Brandon (not verified) :: Tue, 11/14/2006 - 7:16pm

#47 - while I understand in principle, one thing I'm arguing is that while certain results like fumble recoveries are certainly non-predictive and even "lucky" going forward, there are certain game contexts that subjectively may favor a certain team that are poorly characterized by DVOA. Mind you, I'm not flatly saying "DVOA doesn't work", just that it doesn't always do a great job of describing certain teams.

54
by Ilanin (not verified) :: Tue, 11/14/2006 - 7:17pm

29, et al. - Philly match up pretty well against Indianapolis in most areas except one, which is the slightly below league-average DVOA (1.3%) on defending passes to #2 wideouts.

This is not, to put it mildly, a good weakness to have when playing the Colts. I forsee a big day for Reggie Wayne.

55
by paytonrules (not verified) :: Tue, 11/14/2006 - 7:17pm

I really love DVOA BECAUSE it goes against conventional wisdom.

CV says the Bears proved they were the best team by winning. DVOA says "Well the Bears are darn good - but they were getting beaten like a drum for 28 minutes"

56
by MJK (not verified) :: Tue, 11/14/2006 - 7:17pm

Hmm, my own comment just got me thinking about how totally useless the "Time of Possession" stat that the networks always harp on is. What really matters...well, that depends on why you're citing the stat. If you want to know who had more scoring opportunities, than number of possessions would be better, but I would guess that number of possessions will frequently be close for the same team, since possession alternates. I know people like to talk about "tiring out the defense", but since defenses get to rest between plays, shouldn't number of plays be a better indicator of that? Time of possession really tells you nothing except that maybe a team is taking their time and trying to minimize the number of possessions total in the game.

57
by Ray (not verified) :: Tue, 11/14/2006 - 7:19pm

RE #44 Bobman
Indy is ranked 9th vs opposing teams RBs ion the passing game and 16th (exactly avg) vs TEs.
D'oh! You're right, I read the chart wrong.

I think the Eagles #2 WR is better than most other #2 WR, so Brown/Stallworth could still have a big game. But it will probably come down to how well the Eagles can run it and how well they can get pressure on Mannning.

I'm not too concerned about Freeney's game against the Eagles a few years ago. A lot has changed since then, including McNabb maturing even more as a QB. It all depends on whether the Eagles can keep up on offense. The Colt's O is #1, but it hasn't exactly been unstoppable this year either. The Eagles D can do it, but only if they can get pressure with their front four.

58
by hawksfan (not verified) :: Tue, 11/14/2006 - 7:23pm

Seattle is clearly ranked too low because Josh Brown should be worth 10% DVOA by himself (and even more against the rams).

Rating Seattle based on its win/loss record and all other teams in DVOA order is clearly better than this.

59
by Ray (not verified) :: Tue, 11/14/2006 - 7:25pm

RE #56 MJK
I know people like to talk about “tiring out the defense�, but since defenses get to rest between plays, shouldn’t number of plays be a better indicator of that?
Beyond that, what I always wondered about 'tiring out the defense' was, doesn't the offense get tired too? Does a defense really tire out quicker per play than an offense? Maybe the QB doesn't always do a ton of running around, but everyone else on O seems to be working pretty hard to me when I watch.

60
by kal (not verified) :: Tue, 11/14/2006 - 7:25pm

#56, number of plays is far better as an indicator. Both for telling the strength/weakness of a team, and for telling the relative fatigue levels. But even that can be something of a poor stat; truly dominant offenses take less plays to score than good offenses do. (at least usually). DVOA does seem (in my mind) to favor dink&dunk offenses over explosive ones, perhaps because explosive ones aren't sustainable, but it does favor those teams.

61
by Wanker79 (not verified) :: Tue, 11/14/2006 - 7:26pm

Re: 54

I would assume that a not-so-insignificant reason for Philly's poor defensive numbers against #2 WRs is due to 1)Lito Sheppard and Rod Hood being injured at the same time 2)Michael Lewis in pass coverage. Both of which have been resolved.

61
by TBW (not verified) :: Tue, 11/14/2006 - 7:26pm

Re 32:

When you apply a quantitative model that works on a large sample size to a single observation you are asking for trouble. Over time the non-predictive events even out and the model works, but for a single event there is no way to know how it will be impacted by non-predictive events.

If you had a coin rigged so that it would land heads 60% of the time, it doesn't mean to bet the farm on the NEXT coin flip. But if you have all day, to wager and wager smartly you will clean up. If it comes up tails 5 or 6 straight times it doesn't mean your coin is broken, it's just dumb luck.

Before anyone jumps on me for that last bit, I know it's not a prefect analogy. We don't necessarily know for sure how good DVOA is, so if it "comes up tails" a whole bunch it is a little more understandable to wonder if something has changes that is making DVOA less predictive.

Bottom line though is looking at a game here or there and saying you question DVOA because of the result is cherry picking and bogus. The real test is how does DVOA do over a whole season or several seasons. There will always be some absurd examples you can find to make DVOA look bad, but they are irrelevant because they are just isolated incidents.

Now if you can look at the results for a whole season or several seasons and find a pattern to the games DVOA gets wrong then you are on to something, and it's probably an opportunity to improve DVOA.

63
by Jeremy Billones (not verified) :: Tue, 11/14/2006 - 7:30pm

Re: 45

>>The point was not that the eagles have a bad rushing attack between the tackles. It’s that they don’t use it much. Hard to argue that part, I think.

>Again, look back at this past week’s game against Washington. 34 runs to 26 passes

Didn't they report that Reid gave up the playcalling to Morningweg last weekend?

64
by Zac (not verified) :: Tue, 11/14/2006 - 7:31pm

The estimated wins add up to 145, when there have only been 144 games played. It could be rounding error, I'm not sure. But I figured I'd mention it.

65
by Sully (not verified) :: Tue, 11/14/2006 - 7:31pm

Like the sport itself the attempt to quantify is just a game. Funny so many people take it seriously. And, for the third year I cannot help but make a comment. I make so very few.

The numbers may indicate Jacksonville is better than New England or Denver. How many people really believe that?

Who other than Philly fanatics believe the Eagles are really the best?

It is a game of mathematical masturbation. At least it has made the boys some money by signing on with FOX. So, how does that make it "outsiders"?

66
by Richard (not verified) :: Tue, 11/14/2006 - 7:31pm

63: Click my name for story.

67
by Richard (not verified) :: Tue, 11/14/2006 - 7:33pm

64: Rounding error would almost certainly explain that.

65: I can't think of a team I'd less like to play than Philly.

68
by Ray (not verified) :: Tue, 11/14/2006 - 7:34pm

RE# 63 Jeremy Billones
Didn’t they report that Reid gave up the playcalling to Morningweg last weekend?

Yes, although to what extent Morningweg was calling plays beforehand is not exactly clear. The real question will be who is calling plays during the Indy game. We'll just have to wait and see for that one.

69
by Bencoder (not verified) :: Tue, 11/14/2006 - 7:35pm

I don't think alot of people look at variance. This far into the season and from this point to the end it becomes more important IMO.

Speaking of variance, what is the coorelation between variance and strength of schedule (mean or median)?

Aaron, anyone?

70
by admin :: Tue, 11/14/2006 - 7:41pm

I'll try to answer some of these questions later tonight in a blog post. Some of them are addressed in the commentary. Don't assume that the use of the word "random" by readers in comments equals the use of the word "random" by me. Don't assume that I write off everything that isn't yet included in DVOA as a "non-predictive event." Don't assume that I've stopped working on improvements. And don't assume that, just because the Giants dropped less than the Chargers did, it means the Giants had a higher DVOA than the Chargers in Week 10.

71
by Wanker79 (not verified) :: Tue, 11/14/2006 - 7:42pm

Re: DVOA seems to like dink&dunk offenses

Offensive DVOA Rank - Team - YPA Passing
1.IND (3rd 7.90)
2.SD (8th 7.61)
3.PHI (2nd 8.36)
4.NYG (18th 6.73)
5.CIN (6th 7.73)
6.NO (4th 7.87)
7.STL (7th 7.62)
8.DAL (1st 8.78/Romo 15th 6.88/Brunell)
9.WAS (15th 6.88)
10.NE (17th 6.76)

Exactly how many of them would you consider dink&dunk? Out of the top 10 in DVOA, only the Giants, Redskins, Patriots, and Cowboys with Brunell are outside the top 10 in yards per passing attempt. Dal/Was/NE are the bottom. And I'm not really sure I've EVER heard the Giants, Cowboys, or Redskins refered to as dink&dunk.

72
by stan (not verified) :: Tue, 11/14/2006 - 7:43pm

57,

Sure, lots of different things could determine the outcome of PHI-Indy. But the last several years have almost always come down to this -- if your defensive front can resemble a stampede on Manning, you win. If not, you lose.

NE got 3 sacks and a whole host of almosts. And it wasn't enough.

Go back to SD and Pitt last year -- in both games, it looked like the defensive front was having an unimpeded footrace to see who could jump on Manning first.

That's basically how you beat the Colts. 'Cause that's what those who have beaten them have had to do to produce the few losses Indy has over the last 40+ times they have played.

73
by Josh (not verified) :: Tue, 11/14/2006 - 7:45pm

71: I think yards per completion, rather than yards per attempt, indicates if a team is a "dink and dunk" passing offense.

74
by C (not verified) :: Tue, 11/14/2006 - 7:49pm

Looking over the Playoff Odds, something to note: the Patriots ended up with the #4 seed the most of any team in the simulations. And who, pray tell, ended up with the #5 seed the most? Why, it would be the scissors to the Patriots paper, Denver. And then the winner of that game gets to play the Colts in the rock-paper-scissors half of the bracket.

Meanwhile, San Diego would host the winner of Baltimore-Jacksonville in the AFC.

You can't do that sort of analysis in the NFC because Seattle was the most common #3 and #4, so the simplistic approach doesn't work.

75
by Brandon (not verified) :: Tue, 11/14/2006 - 7:50pm

Aaron, thanks for the quick reply. I assume your reference to the Chargers/Giants situation just now is what I was talking about earlier with regard to teams SOS ratings changing and thus changing how earlier games are viewed? To a large extent, it seems that this could be a fools errand, given that the quality of the victory has not changed, only our certainty regarding the level of opposition. However, it's not REALLY the level of opposition for the week that's changed, but rather the overall season performance of a team, which is not really reflective of their quality at any given moment. It seems to me that this could lead to flaws on a team by team basis, but which would ultimately cancel out when considered league wide. This is one reason that I would say that DVOA is still somewhat granular in nature.

76
by Bencoder (not verified) :: Tue, 11/14/2006 - 8:01pm

In regards to variance, another interesting comparison would be in cases where a teams variance is falling (especially dramatically), is the team's win-loss record improving or declining....

I guess you could look at cases of increasing variance.

77
by doktarr (not verified) :: Tue, 11/14/2006 - 8:02pm

The most impressive drive from a DVOA perspective, I think, would be a steady drive with zero "failed downs". Every pass complete, every play making significant progress toward first down, with very few third downs faced and no fourth downs.

I suppose it's possible to characterize that as "dink and dunk", but it's more accurate to describe it as dominant. 8 yards at a time isn't much of a dink.

78
by Richard (not verified) :: Tue, 11/14/2006 - 8:10pm

77: Wouldn't that be the most impressive drive from everyone's perspective?

79
by BB (not verified) :: Tue, 11/14/2006 - 8:10pm

Having watched a lot of Detroit Lions football during the Mornhinweg era, I can't fathom that having him call the plays is going to be a positive for the Eagles over the rest of the season. Unless he's had some sort of epiphany since then that caused him to ignore everything he was doing while in Detroit. Yikes.

80
by Bencoder (not verified) :: Tue, 11/14/2006 - 8:10pm

#77 - Perhaps, but wouldn't that hurt a team like Philly which has feasted on big plays? Or perhaps it does indeed do that in some marginal way......

81
by Richard (not verified) :: Tue, 11/14/2006 - 8:17pm

Football Outsiders rankings correlations with:
ESPN: .844
FoxSports: .807
CBS Sportsline: .833
Jeff Sagarin: .879
Real Time Rankings from Sports Nation: .835

I find it amusing that the Fox Sports rankings are the most different.

82
by kal (not verified) :: Tue, 11/14/2006 - 8:20pm

#71 - I wasn't solely talking about passing. I'm just saying that a team that gets more big plays but less overall 'successful' plays will be lower in DVOA than one that is consistent but does not gain huge swaths. Another way to look at it is not yards per attempt, but yards per completion.

83
by Boots Day (not verified) :: Tue, 11/14/2006 - 8:21pm

DVOA says “Well the Bears are darn good - but they were getting beaten like a drum for 28 minutes�

The Bears fell behind 13-3 after giving up three scoring drives that totaled 13 yards. I wouldn't call that "beaten like a drum," and I doubt DVOA would either.

84
by jonnyblazin (not verified) :: Tue, 11/14/2006 - 8:22pm

Re: 12
Sure, I'll bet every single team in the NFL would have a better defense when their primary pass rusher is using steroids. My question to SD fans is whether they think getting Merriman back will mean that their defense will go back to being one of the better ones in the league.

85
by Bencoder (not verified) :: Tue, 11/14/2006 - 8:24pm

Richard

And that the actual coorelation figure is so consistent. Mostly because these are just a list of teams by win-loss record, with a few exceptions.

86
by kal (not verified) :: Tue, 11/14/2006 - 8:26pm

#78: it would be impressive, but it hurts teams that score on 80-yard passes. An 80-yard pass is worth something like 3 times less than what a drive that takes up 80 yards is worth provided success on that 80-yard drive. I don't know if this is as it should be or if it is not, nor do I know if it is more or less correlating. But it is a factor of DVOA as explained before.

87
by Bobman (not verified) :: Tue, 11/14/2006 - 8:31pm

I think both Stan and Ray are right on the Phil/Indy matchiup. Might come down to a battle of the #2 WRs--I'll take Manning to Wayne.

Wanker79, you meant Bledsoe, right?

MJK: As an Indy partisan, I think TOP is crap. Efficiency per play, yes, TOP, no. BUT, one underappreciated aspect of TOP as a measure of team quality is that it probably exposes the team depth--if you have good depth, you can handle large TOP on offense without getting tired, or can handle a few nicks and stingers without losing too much. If yo have poor depth, your guys might fatigue without being replaced, or one or two injuries might tank you. I assume the same works in reverse for the D--if you can rotate 8 linemen all game, it's no huge deal to have low TOP. But if all you do is rely on 4-5 guys, they'll be spent by the end and giving up Colts-like yardage to the RBs. Also, in either case, you get to expose backups to playing time, which always help when they do HAVE TO step in. It's a theory, anyway. So TOP can be a good thing, but dosn't have to. Even against Indy, one game this year they had about 20 min TOP vs 40 (Jax I think) and they still won.

I can't find the post, but somebody queried about TOP affecting O and D equally. I think CW, which we all know can be suspect, has it that the O is acting and the D reacting. The O knows the snap count and can relax for a couple seconds while the D is taut and flinching every second the QB is under center. The OL knows where it is blocking and how long it has to hold a block for, whereas the DL is fighting through those blocks furiously every second to get to the ball. In fact, a hard-charging D can have their speed and momentum used against them by a clever and calorie-saving OL--"Here, run past me, and out of the picture. And tire yourself out while you're at it. Thanks."

88
by Bencoder (not verified) :: Tue, 11/14/2006 - 8:37pm

#86 - And it brings up the possibility that a team could make some very impressive drives on every possession, get down to the opponents 20 each time and shank half their field goals if their kicker were abominable.

This team would look impressive in DVOA, but would likely almost never win a game that season.....with the exception of their Arizona matchup.

89
by Scott de B. (not verified) :: Tue, 11/14/2006 - 8:43pm

The numbers may indicate Jacksonville is better than New England or Denver. How many people really believe that?

Does conventional wisdom reflect team strength better than DVOA? This should be easy to test.

90
by BB (not verified) :: Tue, 11/14/2006 - 8:45pm

I just wanted to mention as well that one of my favorite things about the Fox blog is the 'censoring', which among other things causes me to see "BLEEP Enberg" in the post corresponding to Audibles. No way am I turning that censoring off, just seeing that alone makes it worth keeping on there for the laughs.

91
by Pacifist Viking (not verified) :: Tue, 11/14/2006 - 8:48pm

35: "DVOA doesn’t tell you who won a game, point scored does that. Points scored doesn’t tell you who played better in a game, DVOA does that."

I am a coach's son, so that might be the reason I disagree with that. Especially in a sport like football, where a single big play can determine the outcome, I do think the team with the most points played better/deserved to win.

42: Agreed. But that happens every year. Should a sport that has such a high percentage of fluky wins (as you say, 1-2 a year for a 16 game season) that determines so much depend very much on a statistical system that discounts those fluky wins?

43: That's a good point, and doesn't just apply to sports but to human history/knowledge: just because something seems random to us doesn't mean it is, and just because something seems non-random to us doesn't mean it is.

47: Agreed. One thing disturbs me though. Football is a sport in which turnovers have a huge impact on the outcome of the game, yet FO seems to discount fumbles as wholly luck. Now, I'm fine with recognizing luck as a factor in the outcome of games. But to create a statistical system, then discount an important factor as luck, seems flawed to me. Now, I don't know that there's a better way to do it within a statistical system, because I am convinced luck plays a large part in fumbles. But that's part of my skepticism of systems in general. I like most of what FO does; I think it tells us a lot. The total DVOA rankings are one area I don't like.

62: That's exactly my point, though: it may balance out on the model over time, but those single non-predictive events have a major impact on outcomes. That single non-predictive event can determine whether a team wins a game, which can determine whether it makes the playoffs.

"looking at a game here or there and saying you question DVOA because of the result is cherry picking and bogus."

Is that what I'm doing?

"they are irrelevant because they are just isolated incidents."

I would accept that, accept that those "isolated incidents" can determine whether a team wins a game or makes the playoffs. That means those isolated incidents, in football at least, are not irrelevant. Irrelevant to DVOA, not to football.

"Now if you can look at the results for a whole season or several seasons and find a pattern to the games DVOA gets wrong then you are on to something, and it’s probably an opportunity to improve DVOA."

I'm not looking to improve DVOA; I don't have an answer for how to do so. All I was originally doing is expressing skepticism about the use of mathematical systems to understand reality (or a sport) in general. I do actually believe in on-quantifiable, intangible factors having an impact on games. Now, I like FO (that's why I'm here reading this stuff every day), but a lot of the commentary seems to hold such intangibles in contempt. That could be my perception of commentors, not the FO commentary, I don't know.

92
by Pacifist Viking (not verified) :: Tue, 11/14/2006 - 8:56pm

Boy, my spelling and grammar need to improve when I whip off on-line comments. For the second "accept" I mean "except," and for "on-quantifiable" I mean "non-quantifiable."

93
by NewsToTom (not verified) :: Tue, 11/14/2006 - 9:01pm

Re #91
Fumbling isn't wholly luck. Certain situations are more prone to fumbling, and certain players (and perhaps teams) are more prone to fumbles. Aaron has stated before the propensity to fumble is taken into account in DVOA. What DVOA doesn't take into account is the likelihood of recovering fumbles, since that's essentially random 50% chance for both teams.

And I don't think anybody would disagree with you that a number of things in football will affect the outcome of a game that are not taken into account in DVOA, and effectively cannot be taken into account in DVOA; Hester's missed FG return for a TD is an excellent recent example of this. DVOA is just another number, like a team's won-loss record. In some ways, it's a better measure of the quality of a team, because it takes into account reproducible events and isn't prone to distortions the way a won-loss record is (1 and 41 point wins both count as 1 win). In a key way, though, DVOA is a worse measure. Wins and losses are what determines a team's fate out there in the world where they play the games. Think of DVOA as an added to coming up with the elusive quantity known as Team Quality.

DVOA =! Team Quality
Record =! Team Quality
Record + DVOA =! Team Quality
Record + Observation =! Team Quality
Record + Observation + DVOA = still not Team Quality, but getting closer

94
by Bencoder (not verified) :: Tue, 11/14/2006 - 9:03pm

Pacifist Viking

I don't think Aaron is saying that fumbles don't impact the game, but rather the team which recovers the fumble is random.

That's not the same thing as saying random events don't impact the score or result.

95
by Richard (not verified) :: Tue, 11/14/2006 - 9:04pm

91: The idea of DVOA is to give you an idea of which teams are better moving forward. Recovering 6 of 6 fumbles will almost certainly help you win that particular game, but recovering 100% of fumbles is not a skill, so it doesn't help us determine the quality of the team. The argument isn't that the team didn't deserve to win because they were lucky. The argument is that they're not likely to continue to win thanks to their fumble recovery skills.

96
by randomn00b (not verified) :: Tue, 11/14/2006 - 9:04pm

Record + Observation > Record + Obversation + Record when it comes to team quality.

97
by randomn00b (not verified) :: Tue, 11/14/2006 - 9:06pm

Sorry I'm tired.

I meant to say

DVOA (statistics) + Observation > Record + DVOA + Observation.

Record doesn't tell you anything that DVOA or Observation do tell you.

98
by Richard (not verified) :: Tue, 11/14/2006 - 9:06pm

I'm curious, are there really people who would think differently of the Colts if the Bills' kicker hadn't gone Norwood at the end of that game?

99
by dbt (not verified) :: Tue, 11/14/2006 - 9:11pm

by my understanding, DVOA does not assume fumbles are luck, it assumes recovering fumbles, over time, will average out.

pretty, clickable DVOA graph updated with week 11 data. Click my name.

100
by dbt (not verified) :: Tue, 11/14/2006 - 9:14pm

And as a Bears fan who thinks that that return was a huge play in the game, let me just remind everyone that the only reason the returner was back there was because of an incorrect assumption by the ST coach (he believed the Giants were calling a fake FG pooch punt).

101
by RecoveringPackerFan (not verified) :: Tue, 11/14/2006 - 9:15pm

53: I wholly agree that DVOA sometimes misses the point. There are also games (like OAK-DEN first time) where one team decides that the only way they lose is to open up and have flukey turnovers, and therefore they shell and protect a small or moderate lead. This hurts DVOA vs. the likely result of keeping the offense open, but it tells the thinking (read: non-statistical) observer that the winning team is still far superior.

102
by underthebus (not verified) :: Tue, 11/14/2006 - 9:20pm

#38 - I'm with you and I've been saying it for the last 3 weeks. SF has real chance to upset this weak division. With the schedule left, SEA does not have as good a hold as people think. I usually trust DVOA, but in this case I think it has something wrong. Look at who the 49ers lost to: SD, CHI, PHI, KC, AZ. (3 of 5 are the top 4 teams in the league). AZ game was close and considering it was the 1st game in a new stadium I think that could have easily went to the 49ers on a neutral field.

Yes I know about the whole Guts and Stomps thing blah, blah, blah... but now SF has avery winnable 2nd half schedule.

Also in regards to this easy shedule: (I posted this in ASG, but it seems relevant here too)
Has there been any articles in regards to division opponents? I'm thinking of the conventional wisdom of teams always playing divisional opponents well. Some recent examples are Hou-Jac, SF-STL, NYJ-NE.

SF has shown even with the horrible 2005 DVOA to be able to beat STL and also play a close game against SEA. HOU takes it to the Jags every year too. Does DVOA have a problem being predictive when it comes to divisional games?

103
by Fnor (not verified) :: Tue, 11/14/2006 - 9:20pm

dbt: That is indeed fraught with awesome.

104
by ZS (not verified) :: Tue, 11/14/2006 - 9:22pm

The Pats just signed Vinny Testaverde.

*prays for Extra Point*

105
by Gus (not verified) :: Tue, 11/14/2006 - 9:26pm

Ravens D is clearly ranked too low (meaning it should have a higher negative value) because the Frickin' Jaguars D is ranked better and that can't be right. Vic Carucci's power rankings suck, but Ray Lewis will eat your labtop if you don't change this.

There, I tried to stay with the formula.

But seriouly, how could the Ravens' D get killed from that Tennessee game? The Titans scored after a stupid safety by McNair, after an interception, and the Titans didn't do anything in the second half until their final drive. Does DVOA take into account how long the D is left out on the field? Because I'm pretty sure the Titans wore them down in the first half because the Nevermores couldn't hold onto the ball.

Does ANYBODY think think that Jamal Lewis has something left? Put in Mike Anderson!!!!1111 Billick, you're an idiot!

106
by David (not verified) :: Tue, 11/14/2006 - 9:35pm

You might want to look at the offensive DVOA of the Ravens' other opponents and how it's changed since last week. changing opponent adjustments can do more to a team's ranking than a new game added to the mix.

107
by Chris (not verified) :: Tue, 11/14/2006 - 9:37pm

#98: think differently? The missed FG was with six and half minutes left to go. The Colts ran the ball down the field when the Bills knew they were going to run it, and ran all the time out. If the field goal had been made, there is no doubt in my mind that the Colts would have drove the field and scored. I think they have done a few times already this year. So, I think of them the same regardless of whether that field goal was made or not.

108
by jonnyblazin (not verified) :: Tue, 11/14/2006 - 9:38pm

re: 105
I just think the Ravens defense simply played like crap in the first half, it didn't have anything to do with being on the field too long. What was odd that that were simply being beaten up front, whereas their weakness in previous games was giving up the big play in the passing game. I think in the future they'll still be a rough front 7 that gives up a big passing play in the future, getting pushed around in the first half vs Tenn I think was an aberration. But the Titans are so crappy that that bad first half wreaked havoc on their DVOA.

109
by Richard (not verified) :: Tue, 11/14/2006 - 9:40pm

107: Poor example on my part, perhaps.

110
by NF (not verified) :: Tue, 11/14/2006 - 9:46pm

91: Non-Adjusted VOA is intended to correlate with wins and losses. DVOA is primarily designed to be predictive. For example, Kansas City went 7-9 in 2004, but its DVOA was 11.4%, and the next year KC nearly made the playoffs in the best division of 2005. Similarly, Atlanta was 11-5 in 2004, but had a DVOA of -2.7%, and fell to 8-8 the next year while its DVOA remained almost exactly the same.

111
by RecoveringPackerFan (not verified) :: Tue, 11/14/2006 - 9:51pm

91: Speaking for myself, I think that "intangibles" do exist, that certain isolated incidents are a key part of why I watch football, and that points don't always show the full story. In the most obvious example from this year, 62 yard game-winning FGs are not good indicators of team quality; they indicate that the kicker managed to make a nearly record setting kick at just the right time. I think that two things give the best window into actual team strength: film study and statistics. I lack the commitment, expertise, and resources to analyze film of every game, so I look at things like DVOA and traditional stats to gain some idea of how good teams really are.

112
by Bobman (not verified) :: Tue, 11/14/2006 - 9:56pm

Oh the commentary on Fox kills me! "Does someone want to tell these guys that the season begins in September, not November?" Is this ironic for long-time fish watchers, or what!!!

Cowboys should be favored over Indy at home...? Aaron, do you mean "should" as in the world should be a just place, or the smart guys in Vegas, and all the pundits are expected to install DAL as favorites, because I don't see anyone in the mainstream sports media (MSSM) bucking the 9-0 trend. In all the FO stats, DAL is a VERY balanced team. Eerie. But so far this year, Indy's worst games have been at home--maybe not DVOA worst, but most lethargic on O.

With all their injuries, I wonder how long NYG will stay where they are.

113
by Rick (not verified) :: Tue, 11/14/2006 - 9:58pm

Re: fumbles

Those of you who think fumble recovery is random should probably discuss the matter with Bill Belichick. Doug Gabriel would appreciate your help.

114
by Rick (not verified) :: Tue, 11/14/2006 - 9:59pm

Re: Testaverde

Really looking forward to Bill Simmons on this issue. Somebody's going to need to talk him off the ledge.

115
by Jim (not verified) :: Tue, 11/14/2006 - 10:02pm

There is a lot of talk here about the word random. I think trying to figure out the predictive value of a fumble recovery will be of minimal value.

What amazes me isn't that people argue over the predictive value of things like fumble recoveries and interception TD returns, but how much value "experts" place on such things.

Take a look at the sites the list Philadelphia 14, 15 or worse. Now had Bryant missed the 62 yarder and the fumble in the Giants game gone through the end zone - all those sites would have the Eagles in the top 5.

116
by Peter (not verified) :: Tue, 11/14/2006 - 10:04pm

The best part of that playoff predicition page is that in 1 of the 15000 variations Houston overtakes Indy to become the #3 seed, which would happen if Houston win 7 and Indy lose 7.

I'm curious what would happen should this happen. Would the world end? Would Dungys head explode?

117
by admin :: Tue, 11/14/2006 - 10:06pm

Commentary now up: http://msn.foxsports.com/nfl/story/6171866.

I'm going to have to get the FOX people to stop writing such negative excerpts on the front page... seems like they always highlight some controversial negative thing rather than something positive like how strong the San Diego offense is.

Didn't even think about that with the Dolphins. I didn't feel well this morning, so don't expect every comment this week to be dramatically insightful.

118
by Jim (not verified) :: Tue, 11/14/2006 - 10:07pm

This will come off a little as preaching to the choir perhaps. But a few weeks back I sent an email to some friends who don't really believe in "stat rankings" like DVOA. I told them that I thought Mia, Oak, Det would have a better record for the remainder of the year than NE, Sea, and StL. At the time the Loser group was 3-17 and the winner group was 13-5. Since that time the losers are 4-3 and the winners are 3-6.

This stuff is the most valuable info out their for rating NFL teams. As a Viking fan I know this now - I hate you guys.

119
by IzzionSona (not verified) :: Tue, 11/14/2006 - 10:08pm

RE: Fumbles--to Pacific Viking and others

It's not that DVOA considers fumbles themselves to be random, but that the system considers each fumble as though it had been recovered at the average recovery rate for all fumbles of that type. The mailbag thread linked in my name contains the numbers used to represent each type of fumble (about two -thirds of the way through the column) Basically, you can figure out what the league average for the defense recovering each type of fumble by dividing the point value given by -8. If a team has forced 16 fumbles defensively on sacks, league average recovery rates would imply that they would recover 7.84 of these. If, instead, they've recovered 13, DVOA sees this as an abnormal variation that it does not expect to continue. Either way, the fumbles themselves are not random (DVOA recognizes causing fumbles as a positive defensive event and not fumbling at all as a positive offensive event), but the recoveries themselves are normalized to remove the "luck" of recovery rates that differ wildly from observed league averages.

120
by kevinNYC (not verified) :: Tue, 11/14/2006 - 10:10pm

#83... The Bears first 7 drives prior to their half-ending TD actually combined to give them less yards and first downs than that 1 TD drive. Those 7 drives ended with 4 punts, 2 turnovers, and a FG the Bears don't get without a personal foul on the Giants. However, the Bears defense did a good job after the Giants' opening drive.

Aaron, I'm still looking for a way to measure a coach's stupidity. Coughlin not punting at 24-20 was easily one of the dumbest things I've seen all season. I'm glad the Giants get a solid boost from their schedule strength. When you look at actual wins and losses/strength of schedule among teams with over .500 records, the Giants is WAY MORE DIFFICULT than the next closest NFC team. The Patriots are the only other team close.

121
by Subrata Sircar (not verified) :: Tue, 11/14/2006 - 10:25pm

Re: 96

As posted, you're saying that weighting a team's record equally with observation is better ("greater") than weighting their record twice as much as observation.

Which, while wholly unintended, still made me laugh ... because it's true.

122
by QB (not verified) :: Tue, 11/14/2006 - 10:27pm

First off, I do like and appreciate the rankings. They are no doubt a valuable tool. However...

If fumble recoveries are random, so be it, but I think the notion that defensive returns have no predictive value is laughable. A guy like Deion Sanders who made a career out of jumping routes and going the other way for 6 will always give his team some small boost by making such plays during the season. It's not something that happens every single game, but it does happen more for some teams (and players) than others over the course of a season, and it's certainly not as a result of luck alone.

There are a number of players like this in the league today. Ed Reed, Troy Polamalu, Ronde Barber, Dre Bly, most of the Bears D...these players will always look to score when they get the ball, and are well above average at doing so.

I'm not saying there should be a massive shift in the rankings, but certainly return TDs should be factored in on some level.

Or if you don't want to be too results oriented, maybe an interception of a short out route could be worth more than intercepting a pass downfield, as it's more likely to result in a defensive TD.

Just a thought.

123
by Stillio (not verified) :: Tue, 11/14/2006 - 10:29pm

120: I think you meant the Jaguars are the only other team close, seeing as how the Pats have the 25th easiest past schedule. Somewhat ironic that the two most tested winning teams still don't get a break and have to play each other. (Yes, I realize the Jags just got a break with Houston coming to town and blew it; don't bother me with facts)

124
by David (not verified) :: Tue, 11/14/2006 - 10:38pm

QB: if fumble recoveries are random, then so is whoever snags the ball, and his not being crushed under a pile of unfriendly jersies at the time. Running ability is not luck, but being in a position to move more than a few feet before being totally surrounded is, and separating the one from the other might be too much for a system based on play-by-play data.

125
by putnamp (not verified) :: Tue, 11/14/2006 - 10:45pm

Going way back for #1,

22nd in DVOA, but do you really want your team facing them when Hasselbeck and Alexander get back, especially now that the offensive line seems to have gelled and they've put together 2 straight 120+ yard games?

People talk about the Giants, for example, seeing a lot of injuries, but starting QB, starting LG, starting RB, starting RT, starting DT, 2nd/3rd WR (Engram), and Walter Jones, the best player on the team, has been playing injured since Week 1.

Gotta cut them a little slack for having a #22 DVOA. And underthebus, I anxiously await the next 49ers game, because I'll finally be able to watch the Seahawks in person instead of that stupid 49ers post-game show. I hate that thing so much, it's not even funny.

126
by putnamp (not verified) :: Tue, 11/14/2006 - 10:45pm

in person meaning on TV.. tickets for the game sold out long before I got to them :(

127
by BG (not verified) :: Tue, 11/14/2006 - 10:47pm

The bills kicker missed with 6 minutes to go in the game. It's not like it was a last second kick he missed.

128
by kevinNYC (not verified) :: Tue, 11/14/2006 - 10:48pm

#123... You're right. That's what I get when I don't look things over before I post them.

Re: Aaron's Eagles comments... How could any reasonable person expect the Eagles go 6-1 or 5-2 over their last 7 games? It's not outlandish to think they could lose 4 of their remaining games. I don't expect the Giants to collapse considering they had a good opportunity to win the Bears game prior to Tom Coughlin's total brainlock. They will get quite a few players back and luckily for them, they won't have to face the top-rated DVOA defense every week.

129
by Fnor (not verified) :: Tue, 11/14/2006 - 10:50pm

these players will always look to score when they get the ball.

And I'm sure Luke Pettigout always looks to wait for the snap to move, but it just doesn't happen all that often.

Joking aside, what does it matter what they want to do? There is almost nothing in their control when they actually do get the ball.

If you want a real analysis of this, look at Aaron's DVOA commentary from last week (or was it two weeks ago? Time sure flies....).

130
by Fnor (not verified) :: Tue, 11/14/2006 - 10:52pm

#128: Because their D is on life support, and they've lost an important O lineman?

131
by QB (not verified) :: Tue, 11/14/2006 - 10:55pm

But don't some kinds of turnovers result in defensive TDs more often than others? And aren't some teams more likely to create these kinds of turnovers?

Just as an example, I would assume that creating a fumble by blindsiding the quarterback and chopping his arm would result in a TD more often than creating a fumble by putting a big hit on a WR after a catch. So, if true, creating the first kind of fumble would be considered superior to creating the second type. This is very similar to what FO already does when they consider how likely different types of fumbles are to be recovered.

132
by navin (not verified) :: Tue, 11/14/2006 - 10:56pm

Re: 102
Hey underthebus, I agree with you but I also understand what DVOA is saying--that any Sunday the bad SF can show up and get blown out of the water--hence the high variance. Also the wins haven't been blowouts, except for the Oakland game. I wish SF had converted all those turnovers into touchdowns instead of field goals against Detroit. Oh well...

Also, DVOA rated SF around +20% in the week one loss to the Cardinals, pretty much because of fumble luck. I'm concerned about the road games because that's where the blowouts usually happen (@CHI, @KC), but this past week's game gives me hope, especially since it was an early game. Previously this year and in years past it seemed like SF would sleepwalk through the early games. Let's hope that that trend is over.

133
by Jason Scheib (not verified) :: Tue, 11/14/2006 - 10:57pm

This has got me thinking and so I just ran the numbers for Actual Turnover Ratio for the first time this season (including safeties).
The top five:
Chicago +32
San Diego +23
Indianapolis +16
Baltimore +14
4 teams tied for 5th place with +11
Philadelphia
Dallas
Jacksonville
New England

ATR may not be as good a measure as DVOA, but it's still a good measure. And it also has Philadelphia in the top 5.

Like I said, I'm just thinking.

134
by doktarr (not verified) :: Tue, 11/14/2006 - 10:59pm

122:

Interceptions on some rountes are worth a lot more than others. The amount is based on the average length of a return on an interception that far from the line of scrimmage, as well as where on the field it happens.

What FO says about "defensive returns having no predictive value" is that a player having a better INT return than average (for that field position) does not have significant correlation with later deviation from an average return.

I know this is hard to believe, but the perception that some guys tend to run back INTs more often than you would think, based on where they caught them, doesn't hold up. Deion Sanders returned 9/53 (17%) of his INTs for TDs in his career. Picking a couple active veteran CBs completely at random: Nate Clements is 4/20 (20%), and Deshea Townsend is 2/15 (13%). Of course these numbers don't correct for situation - some guys may make more of their picks close in where runbacks are easier.

135
by thok (not verified) :: Tue, 11/14/2006 - 11:00pm

33: Since I didn't see this answered by anybody, the change in Indianapolis's upcoming schedule strength is mostly about not having Buffalo as being on their upcoming schedule anymore. If you notice their past schedule dropped from 15 to 16.

Next week Indianapolis's future schedule will almost certainly drop back to around 15.

136
by doktarr (not verified) :: Tue, 11/14/2006 - 11:01pm

RE: 131,

Yes, again, all of this stuff is now in the system. Read the DVOA chapter in PFP 2006.

137
by QB (not verified) :: Tue, 11/14/2006 - 11:30pm

All right, thanks for the help dok. Maybe the section I read was misleading or maybe I just leapt to the wrong conclusion, but I thought their system did not account for those things.

Cheers.

138
by Pat (not verified) :: Tue, 11/14/2006 - 11:44pm

Re: Aaron’s Eagles comments… How could any reasonable person expect the Eagles go 6-1 or 5-2 over their last 7 games? It’s not outlandish to think they could lose 4 of their remaining games.

Because Philly obviously presents a poor matchup for the Giants. A team that, on defense, has an insane pass rush with just their front four, forcing tight throws into coverage, and a team that has receivers that basically only knows how to attack you deep. Not a good matchup for the Giants.

Ditto with the Colts, although the matchup on offense isn't as clearly good, although I think most people will be surprised. I don't think Indy's run defense problem is limited to power runs up the middle.

In Philly's last 7 games, they've got 3 games against teams above a 0% DVOA, although they're clearly competitive with two of them (Giants and Dallas) given previous experience. How is it unreasonable to say they'll win, say, half of those games (so 1 to 2 losses), and win the others (against teams with a DVOA of -0.9%, -5.8%, -7.4%, and -29.6%)? That's 6-1 or 5-2, and 11-5 or 10-6.

I'd say 11-5, 10-6, and 9-7 are all pretty equally likely. I doubt home field advantage is going to be a huge deciding factor in some of those games - especially Indy, where being in a dome actually helps to play to Philly's strength.

139
by Andrew (not verified) :: Wed, 11/15/2006 - 12:03am

Ilanin #57:

Philly match up pretty well against Indianapolis in most areas except one, which is the slightly below league-average DVOA (1.3%) on defending passes to #2 wideouts.

Context here is everything.

Normally, when the Eagles play at full strength, Lito Sheppard lines up on the offensive right side (strong side) vs. the normal position of the #1 wideout and Sheldon Brown lines up on the offensive left vs. the #2 wideout. When the formation goes to 3 wideouts, Sheldon Brown slides in to cover the slot receiver, and Rod Hood covers the offensive left.

For games 1 through 4, Lito Sheppard was out with an injury, with Sheldon Brown taking his place. However, from the 4th quarter of the Giants game (game 2) until the Jacksonville game (game 8), Rod Hood was also out injured, leaving Joselio Hanson and Dexter Wynn to fill in. These injuries forced changes in the normal style of play of the defensive backfield, so I would take the numbers with a grain of salt.

Normally, Brown and Hood are the better receivers at preventing completions, while Sheppard is more of a gambler who tends to come up with more interceptions and pass defenses, but also more completions. This plus the location they are played at tends to leave the Eagles weaker against #1 receivers, but strong everywhere else. I would expect Harrison to be the one catching a lot of balls against the Eagles, not Wayne.

However, that also comes with a caveat. Much of the weakness against #1's this year has been strong safety Michael Lewis in pass coverage. After giving up long touchdowns to Eric Moulds (1), Amani Toomer (1), Joe Horn (2), failing to recover a Plaxico Burress fumble he fell upon, and causing a huge pass interference call to Terry Glenn in terrible coverage, Lewis has been removed from passing downs and replaced by Sean Considine, who is apparently much better in pass coverage.

140
by smashmouth football (not verified) :: Wed, 11/15/2006 - 12:07am

Re: The Ravens' poor defensive play last Sunday
I watched the entire game and it didn't sink in at the time, because the announcing crew didn't realize it. Although everyone knew Ray Lewis sat out the game with an injury, practically nobody noticed that the BACKUP middle linebacker, Mike Smith, suffered a dislocated shoulder on something like the 2nd play of the game.
It took the Ravens until halftime to figure out how to adjust to playing with a practice squad rookie at a crucial position, especially given the scheme.
So although the injuries are obviously a concern, I think it's likely the poor first half D performance will prove to be somewhat of an outlier.

141
by Andrew (not verified) :: Wed, 11/15/2006 - 12:08am

Jeremy Billones #63:

Didn’t they report that Reid gave up the playcalling to Morningweg last weekend?

Yes, but Mornhinweg had already been doing about 50% of the play calling through the first 8 games. If Reid is not covering his mouth with the playsheet between downs, he is not calling plays.

142
by Kraig (not verified) :: Wed, 11/15/2006 - 12:08am

When will DVOA ease up on the 6-3 Seahawks? What are they, now, 21st after beating the 16th-ranked Rams twice? And the Rams are on a 4-game losing streak. Lopsided losses coupled with close wins sure makes the ranking look weird in their case.

143
by Moses (not verified) :: Wed, 11/15/2006 - 12:17am

There's something seriously wrong with this formula. Teams win and they go down the same as the team losing? The Eagles #1 after going 5-4 with their powder-puff schedule? The Steelers, who play like they don't care, are #12? There are massive disconnects between DVOA rankings results and records?

At one time I thought you guys might produce something useful. But this is trash. There's no predictive or explanitory value in this navel gazing fest. What a freakin' disappointment.

144
by Rodney Harrison (not verified) :: Wed, 11/15/2006 - 12:21am

Glad to see I'm finally getting some respect.

145
by Pat (not verified) :: Wed, 11/15/2006 - 12:33am

What are they, now, 21st after beating the 16th-ranked Rams twice?

Wait, beating an average team - both times on last-second field goals (and one of them a 54-yarder!) - is supposed to make you great or something?

Man. Considering the Seahawks beat the Giants, who are ranked #3, I think you're choosing the wrong results.

In any case, though, I'm guessing the Seahawks' DVOA will improve about the same time the actual team starts improving (as opposed to needing to rely on a kicker at the buzzer): when Shaun Alexander comes back.

146
by david mazzotta (not verified) :: Wed, 11/15/2006 - 12:34am

Current Vegas odds on Bears as Superbowl champs: 2/1

Current Vegas odds on Eagles as Superbowl champs:
20/1

20/1 is awfully appealing.

147
by NF (not verified) :: Wed, 11/15/2006 - 12:37am

I think Houston will win the 2005 Miami Dolphins Memorial Sleeper Pick Award. They are 3-6 right now, and have had the hardest schedule in the league. They actually have improved from being a league bottom-dweller to merely bad, and they have the third-easiest remaining schedule in the league. A 5-2 finish against Buffalo, NY Jets, Oakland, Tennessee, New England, Indianapolis, and Cleveland would put the Texans at 8-8, and probably undeservably cause a repeat of the general opinion that surrounded the Vikings and Dolphins this offseason.

148
by Andrew (not verified) :: Wed, 11/15/2006 - 12:38am

Pacifist Viking #91:

Especially in a sport like football, where a single big play can determine the outcome, I do think the team with the most points played better/deserved to win.

Lets say team A is leading team B, but as team B is marching down the field looking to score a final field goal to win, team A apparently stops them out of field goal range on 4th down as the clock winds down. Then team A's strong safety gets in a fight with someone on team B and spits in his face or knees him in the nuts right in front of the Ref after the team B player punches him in the face while no one is looking, the team A player is called for a personal foul and puts team B in field goal range with a fresh set of downs, team B kicks the field goal and wins. Was team B better, or did they just get lucky at the lack of discipline by one team A player?

How about if team A forces a fumble by team B to end the final drive, but the ball bounces wildly, is recovered by team B and run in for a touchdown?

Should a sport that has such a high percentage of fluky wins (as you say, 1-2 a year for a 16 game season) that determines so much depend very much on a statistical system that discounts those fluky wins?

Unless your goal in the sport is a perfect season, it seems reasonable to use a less than perfect statistical system to model your opponents abilities and tendencies, and to ignore the fluky plays, injuries, and bad officiating that you cannot control.

Football is a sport in which turnovers have a huge impact on the outcome of the game, yet FO seems to discount fumbles as wholly luck.

NO, NO, NO!!!! DVOA does not discount fumbles! It discounts fumble recoveries and their subsequent returns. Fumbles are definitely penalized on offense and rewarded on defense. DVOA attempts to normalize fumble recovery to ordinary recovery chances for the offense and defense for fumbles in different situations and in different areas. In this way, it becomes predictive because if team A can cause a sack on one of every 10 drop backs, and a fumble on one of every two sacks, and recover 3 of every 5 fumbles caused by sacks, over time, team A will normally recover an ordinary percentage of these fumbles, even if it has not to date during the season, or even if it has an 10 sack game with 5 fumbles of which it recovers only 1.

The Saints for example, while not recovering and inordinate number of their own fumbles, had through week 8 recovered all of their own fumbles when they were in scoring position, while only losing fumbles around midfield. That is good luck. It has nothing to do with skill.

149
by Joon (not verified) :: Wed, 11/15/2006 - 12:41am

aaron,

the foxsports page has the heading "2006 only" when it probably should say "full year" (for the non-weighted DVOA rank). maybe you could get them to fix it?

150
by SJM (not verified) :: Wed, 11/15/2006 - 12:50am

Re: #146

If I was a gambler, I'd place that bet in a heartbeat. Last week I picked the Eagles as my NFC champ, mostly because Grossman and Eli are too shaky and also the Giants are getting hammered with injuries. The Eagles meanwhile are solid and have a reliable QB in McNabb. They are better prepared to go deep in the playoffs, I think.

151
by Richard (not verified) :: Wed, 11/15/2006 - 1:11am

"Pork Chop Womack's return has stabilized the offensive line — raise your hand if you ever expected to read that sentence"

Well, in the sense that dead is stable...

152
by Polaris (not verified) :: Wed, 11/15/2006 - 1:15am

#145

A win is a win is a win, and Seattle [i]dominated[/i] St Loius for most of that game (in case you didn't bother to watch it).

I am not saying Seattle is a great team (yet), but they are certainly better than their DVOA (and the Eagles are certainly worse than their DVOA).

That's the point. If I were a niners fan, I'd be careful about hanging my hopes on the Seahawks low DVOA especially considering that the niners are even worse!

-Polaris

153
by Andrew (not verified) :: Wed, 11/15/2006 - 1:16am

kevinNYC #128:

How could any reasonable person expect the Eagles go 6-1 or 5-2 over their last 7 games? It’s not outlandish to think they could lose 4 of their remaining games.

Ummm ... because they've consistently done it year after year under Reid?

Eagles record in the final 8 games since 2000:

2000 - 6-2
2001 - 6-2
2002 - 6-2
2003 - 7-1
2004 - 6-2* (* last 2 games essentially forfeited by playing scrubs after locking in #1 playoff seed)
2005 - 2-6 (injury plagued disaster season)

The cause of this is pretty simple. McNabb has mostly played great in September, November, and December, and mediocre in October since the 2001 season.

Could they lose 4 of their final 7 games? Yes. In theory, they could lose them all. Will they? Probably not.

154
by Polaris (not verified) :: Wed, 11/15/2006 - 1:17am

#151 Seattle's O-Line has gotten much better the past two weeks. Look at the Line Adjusted yards. That line isn't to 2005 levels yet, but it's a lot better than you are giving them credit for.

-Polaris

155
by Pat (not verified) :: Wed, 11/15/2006 - 1:25am

NO, NO, NO!!!! DVOA does not discount fumbles! It discounts fumble recoveries and their subsequent returns. Fumbles are definitely penalized on offense and rewarded on defense.

You know what's funny? People say "fumble recoveries aren't random", blah, blah blah, but really, the main thing to ask is - does anyone really think it's okay to fumble the ball? At all? Regardless of who recovers it?

Steve McNair has fumbled 5 times. His team has recovered all of them. It's completely crazy to say McNair holds onto the ball extremely well. He doesn't. He's just gotten freaking lucky that when he's lost the ball after he's been sacked, he (or an offensive lineman) was able to fall on it.

Likewise, does anyone really think that causing fumbles is worthless? Does someone honestly think that Brian Dawkins didn't make one hell of a play where he caused a fumble that rolled into the end zone and was covered by a Giants player? Of course he did.

Don't most people believe that fumbles, even if they're kept, should be bad for an offense, and fumbles, even if they're retained by the other team, should be good for a defense?

And do people really think that when a fumble is caused, the fumbling player is usually trying to direct it to an intended target? Or the person who causes the fumble is trying to shove it to a teammate? I've watched enough football to know that's not true. Fumbling players are going "ooohhh crrraaaap" and reaching in desparation for the ball, and fumble-causers are only concerned with forcing the ball free.

Skill in picking up the football might exist, but it doesn't matter. It's such a small component of the actual play. The direction the ball goes is drastically more important. It won't matter if the team has a guy that picks up 100% of the fumbles that land right by him. The ball will only bounce to him a small fraction of the time, and the change in total recovery for the team will be negligible.

156
by Darth Goofy (not verified) :: Wed, 11/15/2006 - 1:27am

The Indianapolis Colts are clearly ranked too low because all they have done is win 30 straight regular season games where seeding has been set. Picking my nose and letting the boogers fly is way better than this. The Baltimore Colts ROCK and all others SUX!!! Colts will beat the 'Boys and EGals the nxt few wks!!!!!

157
by Polaris (not verified) :: Wed, 11/15/2006 - 1:30am

#145

So much then for the predictive value of DVOA then. Consider, if I were to tell you that one team in the NFC West was 6-3, two were 4-5, and one was 1-8 at this point in the season, and then I told you to look at DVOA (sans commentary), which team would you put where?

I'd almost certain say the Rams were 6-3, Seattle and San Fran were 4-5 (with San Fran outperforming their DVOA rating by a significant margin I might add), and the Cardinals correctly predicted at 1-8.

Funny, but that's not the way it's worked out....and that tells me the system while useful is flawed.

-Polaris

158
by Pat (not verified) :: Wed, 11/15/2006 - 1:38am

A win is a win is a win, and Seattle [i]dominated[/i] St Loius for most of that game (in case you didn’t bother to watch it).

SEA, + home field: +9.0%
STL: 0.9%
Difference: 8.1%

So what if Seattle dominated St. Louis? It's not like this is amazingly surprising. St. Louis isn't an amazing team. Seattle was supposed to beat them at home.

And I don't think you and I have similar ideas of "dominate". "Dominations" do not involve putting up fewer yards, being on the wrong end of the turnover ratio, being sacked 6 times, and really only being strongly ahead in special teams performance on the day.

Do I think Seattle should've likely won by more than 2 points? Probably, yeah. Would I call that performance a "domination"? Um. No. I usually use the word "dominate" for a game like, oh, I don't know, Seattle's victory over the Giants.

159
by Pat (not verified) :: Wed, 11/15/2006 - 1:45am

and then I told you to look at DVOA (sans commentary), which team would you put where?

Seattle's DVOA is -8.6%. Their variance is 10.3%. St Louis's DVOA is 0.9%. Their variance is 9.3%. These two teams are not tremendously different in ability. It's not surprising that one's 6-3 and one's 4-5. It's only two games difference.

160
by MdM (not verified) :: Wed, 11/15/2006 - 2:08am

By the way, the Zwinky is hotter than that weird Catholic Match girl, imo.

161
by jonnyblazin (not verified) :: Wed, 11/15/2006 - 2:23am

Pat, I agree with your point, but your example with McNair is flawed. You are calling a fumble a play where McNair is in the shotgun, the snap is a bit off, it softly bounces off his hands to the ground in front of him and he recovers it. He has 3 of these own recoveries this year. Its not the same as being blindsided by a DL and the ball flies loose; this hardly ever happens to McNair because of his pocket presense.

162
by kevinNYC (not verified) :: Wed, 11/15/2006 - 2:46am

#138... So you're saying PHI playing 3 of the top 6 DVOA teams ON THE ROAD isn't something to be concerned about? Regardless of the overused "luck didn't go our way" excuse, the Eagles are 1-3 against teams with positive DVOA. If we're going to just assume victories against teams with negative DVOAs, I might as well chalk the Giants up for three wins now. I love how you only look at PHI's offensive matchup against NYG & INDY, while ignoring what can happen the other way. PHI allowed Eli Manning to have EASILY his best game of the season and INDY is a solid first in offensive DVOA.

#153... see above.

163
by Nolan (not verified) :: Wed, 11/15/2006 - 2:54am

Once again with Pitt not in the very bottom DVOA you invalidate you so called rankins

164
by Chris (not verified) :: Wed, 11/15/2006 - 2:58am

Very cool, Mike Harris. For a comparison between the DVOA-based playoff forecast and my nfl-forecast.com predictions, click my name. The comparison table is linked in the article. In many cases the two forecasts are within statistical error. In other cases (AFC west and much of the NFC) there are some interesting differences.

165
by Polaris (not verified) :: Wed, 11/15/2006 - 3:16am

Pat,

Did you actually watch the 'Hawks vs Rams game? Your comments indicate that you did not. My point is that while statistical analysis is useful, it has to match with the 'eyeball' test.

STL had ONE...count that ONE sustained drive all game long. They got seven points out of it. The rest of their points came from being on Seattle's side of the field pretty much all the second half due to a boneheaded play by FS Hamlin. Seattle by Contrast had four sustained offensive drives.

That IS domination.

DVOA does not count fumbles returned for TDs....so remove 7 points for STL right there. Also Seattle was within the 10 yard line....so add (at least) three to Seattle right there. Also Burleson's TD return DOES count in DVOA (special teams). Thus it's really Seattle 27 and STL 15....not a blowout, but not terribly close either which is why Seattle gets more DVOA credit for this game than you want to admit Pat.

The fact is that the system is flawed. It favors teams like Philly and Pitt too much and unduly punishes teams that win narrowly (like Indy and Seattle).

If you can't admit that, then you don't understand football. Period.

-Polaris

166
by Bobman (not verified) :: Wed, 11/15/2006 - 3:22am

Darth Goofy, you sith lord you, was that an intentional TV announcer-like flub, or are you old enough to have cheered for them in Balt? It's been a looooong time since I made that verbal mistake. I think all my old Balto Colts stuff is gone, except for a mug with a broken handle. I may still have a hat somewhere....

167
by Tally (not verified) :: Wed, 11/15/2006 - 3:44am

Again, DVOA does not try to explain past results; it tries to predict future performance based on repeatable past results.

168
by zip (not verified) :: Wed, 11/15/2006 - 3:52am

If you can’t admit that, then you don’t understand football. Period.

False dichotomy for the win!

169
by Mnatr (not verified) :: Wed, 11/15/2006 - 3:53am

#48

Is it true that overall DVOA is affected by the side of the ball that sees more plays? I'd think that DVOA is a synthesis of O, D, and ST, not weighed by how much time each unit has seen.

170
by Richard (not verified) :: Wed, 11/15/2006 - 3:55am

168: Zing!

171
by centrifuge (not verified) :: Wed, 11/15/2006 - 3:56am

Every week, I see this, and every week, I laugh.

DVOA exists, fundamentally, because the Outsiders wanted to find a way to judge teams more accurately than common wisdom. Still, every week, people are in here complaining about how it doesn't match up with common wisdom.

If DVOA always matched up with common wisdom, it would have no reason to exist.

172
by MarkV (not verified) :: Wed, 11/15/2006 - 4:15am

is it just me or does the Fox rankings list (inaccurately) Oaklands defensive rank at 32.

173
by Polaris (not verified) :: Wed, 11/15/2006 - 4:31am

#171

No. DVOA (as I understand) was created as a better way to evaluate which teams were playing better football, i.e. had the better team.

You want to be able to use it as a predictive measure, but to *do* *that*, you also need to sanity check it. That is does DVOA accurately predict W-L records currently on the books?

Last year, it did a pretty good job.

This year, a coin is a better predictor than DVOA.

That tells me at least that there is something wrong with DVOA this season.

-Polaris

174
by Sid (not verified) :: Wed, 11/15/2006 - 5:22am

this is beyond awesome. The 5-4 Eagles ranked 1st, ahead of a 9-0 team (ranked 5th) and both 7-2 teams (one of which is ranked 10th).

Truth is, Eagles have outplayed the Colts this season. It's still funny to see, though.

175
by putnamp (not verified) :: Wed, 11/15/2006 - 5:34am

To respond to whoever attributed Seattle's "turnaround" on Shaun Alexander's return:

Several factors will herald their turn-around moreso than Alexander, I believe. One is if Floyd Womack and the rest of the OL can continue at the level they've played the last two games, and two is if the defense can remain consistent in pass defense. Hasselbeck returning would be #3, possibly swapped with the pass defense at #2. I really think Alexander is #4 on the list.

176
by Matt (not verified) :: Wed, 11/15/2006 - 6:43am

this is beyond awesome. The 5-4 Eagles ranked 1st, ahead of a 9-0 team (ranked 5th) and both 7-2 teams (one of which is ranked 10th).

Truth is, Eagles have outplayed the Colts this season. It’s still funny to see, though.

I really cannot understand that logic. How have you "outplayed" an undefeated team when you have 4 losses? The Colts beat Denver in Denver. I think that's more impressive than anything Philly has managed to accomplish this year. Having a 5-4 team #1 is just crazy. But all is well, Philly plays the Colts on Sunday, Nov. 26 on NBC. We'll see who is #1.

177
by brasilbear (not verified) :: Wed, 11/15/2006 - 7:25am

Regarding the Bears return of a field goal for a TD:

(Warning Bears Fan present)

While the Chicago media is reporting that the return was a called play (return right according to the papers) I agree that it has to be judged as a "random" result. However, some are missing that point that the result of that play will end up in the DVOA-ST somewhere down the line. At some point on a long field goal, some ST coach will make an adjustment telling his holder to watch out for Hester on a return and the snap will be dropped, or a TE on the end of the line will be tasked with holding the side line if there is a run back, and he will miss a block allowing Tillman or Vasher (just making up names) to shoot past him and get a hand on the kick.

That one play will have a ripple effct everytime some team lines up for a 50+ field goal and Hester lines up on the goal line.

Also:

Bears is clearly ranked too high because Evil Rex still has two cover your eyes bad games left in him. Having my beginner orchersta class pick winners using free throws is way better than this. This isn't a real sport anyway. those guys wear armor so it doesn't hurt anyway. Futebal is a much better sport (Go Flamengo go!!!)

178
by JasonK (not verified) :: Wed, 11/15/2006 - 8:36am

Hey, Aaron cited me in the Fox commentary! I'm famous!!! ;)

179
by centrifuge (not verified) :: Wed, 11/15/2006 - 10:50am

DVOA (as I understand) was created as a better way to evaluate which teams were playing better football, i.e. had the better team.

And does it not do that? How do you know that it does not? Because it violates what you see as common wisdom. But, as Tanier put in the blog a bit back, "DVOA remembers" even when the common wisdom forgets.

That DVOA does not predict single games well this season doesn't worry me. Asking it to do this mistakes its purpose -- it's the entire non-predictive event debate again. DVOA has been the most accurate predictor of end-of-season records for multiple years.

180
by BadgerT1000 (not verified) :: Wed, 11/15/2006 - 11:33am

Reading some of these comments reminds me of the stories my brother tells about his consulting work.

My bro' has his Phd in QA (stats nerd) and does all kinds of analysis/modelling work. One of the areas where he gets involved is sales compensation models. Say Company A just acquired Company B and they have two different comp models for their sales force and the execs want to show everyone why one is "better" or maybe develop a hybrid. So they call my brother.

Anyway, at some point with the "final" model there is a presentation to the salesforce at large. And repeatedly my brother tells the company "don't show them the model at a transaction level, only show them based on annual maybe monthly summary level". When they ask why he explains that the salesperson will immediately see differences and ASSUME those differences ripple across at the same magnitude to his other transactions. Which is not correct.

So the execs ignore him, present, somebody asks for an example, and they give a granular example instead of based on annual activity and all h*ll breaks loose if the person sees on that singular event he would have made less money then using the old method.

I am sure some reading this will ignore the larger message and attack my brother's work/methodology as surely HE must have made a colossal mistake if this type of difference occurs. Because clearly if something doesn't "work" on a single event how can it work on a collection of events?

Me, I appreciate the efforts here. I think this site and its authors are excellent. I understand they are trying to provide an alternative perspective as opposed to definitive answers. And if they wanted to be popular they would input half-*ssed correction factors to keep the comments traffic positive.

Keep fighting the good fight dudes.

181
by TracingError (not verified) :: Wed, 11/15/2006 - 11:42am

Washington is mediocre, and I'm glad they benched Brunell. But I just want to point out that they have lost to #1, 3, and 5, all on the road coming off their bye, they lost to #6 on the road and beat #6 and #7 at home. They did lose close ones to Tennessee and Minnesota, but they are not nearly as bad as the media seems to think.

Which brings up the following question: how do teams do after their bye, and if they do better, does DVOA care?

182
by jebmak (not verified) :: Wed, 11/15/2006 - 11:49am

On Nov 26th, I bet that nearly everyone here will be rooting Eagles, because the fate of the message board is riding on it.

Even though the game will be 52%-48%(or whatever), only the winner matters.

If the Colts win, extra filters may need to be applied.

183
by Sophandros (not verified) :: Wed, 11/15/2006 - 11:52am

Question regarding Cincy and New Orleans, since they play this weekend and seem to be virtually identical according to DVOA:

You discuss Cincy's Back Zone and red zone defensive woes, so I'm curious what NOLA's offense looks like in those areas. Might be something interesting to see during Sunday's shootout.

184
by J.D. (not verified) :: Wed, 11/15/2006 - 11:53am

Looking at the Harris playoff odds, I find it interesting that 5 out of the 6 AFC playoff seeds are all but set (according to this). NE, IND, BAL, SD, and DEN all have odds of 85% or above to make the playoffs, with JAX, KC, and NYJ all relatively even fighting for the last spot.

Maybe this goes back to PacifistViking's Dostoyevsky criticism, or maybe it's because of the lack of accountability for injuries/infighting/whatever else, but I can't imagine that only halfway through the season, things are nearly that sure.

185
by Darth Goofy (not verified) :: Wed, 11/15/2006 - 12:06pm

RE:#166

It was an homage to BLEEP Enberg. I am just old enough to have known that the Balt Colts moved to Indy (in fact, the Colts moved to Indianapolis the year I moved from Indianapolis at the age of 8... concidence, I think not)

186
by SJM (not verified) :: Wed, 11/15/2006 - 12:06pm

TracingError:

The Skins might be better than their media perception, but I would attribute that to players returning from injury more than strength of schedule (although the NFC East is by far the best division).

There was an article in PFP 2005 that basically said (if I recall correctly and with a mild slant of my own) that winning teams coming off a bye have no advantage over other winning teams, but do have an advantage over losing teams. (Actually it says that winning teams have a disadvantage against other winning teams, but that is due to an apples and oranges problem. Winning team having roughly a .500 record vs other winning team is what you would expect. It is lower than the pre-bye win % because that includes games against losing teams.) Losing teams coming off the bye are .500 against other losing teams (there is another apples and oranges mistake in the book in that case. The writer incorrectly interprets the data to say that a losing team has a better chance to beat another losing team after the bye, when in fact the bye is irrelevant and the team has a shot because the other team sucks. Ryan Wilson, you should be ashamed!)

Eliminating these errors, the bye week actually doesn't significantly help or hurt anyone. Teams go .500 against teams with similar records, good teams beat bad teams and bad teams lose to good teams, just as you would expect in any given week. So the bye is a non-significant (seriously in this case, not non-predictive or non-repeatable or random) event when taking about expected wins the next week.

187
by bearhater (not verified) :: Wed, 11/15/2006 - 12:07pm

just like last year, chicago will be shown
to be the pretenders they are come playoff time

188
by C (not verified) :: Wed, 11/15/2006 - 12:08pm

A point about DVOA favoring "dink and dunk" offenses: Phila. leads the league in offensive plays over 20 yards.

189
by MRH (not verified) :: Wed, 11/15/2006 - 12:11pm

"In retrospect, the shutdown of Chris Chambers and Marty Booker made a lot of sense. Kansas City has the best DVOA in the league against no. 1 receivers, and ranks ninth against no. 2 receivers. However, they're last against other receivers and 20th against tight ends."

My subjective view: Lenny Walls (KC's 3rd cb most of the year - Sapp has been mostly hurt) sucks, with one exception. Most of the big plays the Chiefs' have given up seem to have been with Walls in coverage or when Law falls down (which I think is a non-predictive event). When an offense can isolate a #1 or #2 on Walls, he mostly gets burned. I wonder if the game charters can tell if my subjective opinion matches with the record?

The one exception where I like Walls in coverage is vs. fade routes to tall receivers in the end zone. At 6'4", he seems to defend those well. A pretty specialized role but unfortunately he is being asked to do that.

I wonder how he'd do if asked to matchup with TEs?

190
by ToxikFetus (not verified) :: Wed, 11/15/2006 - 12:19pm

Scramble for the Ball or some other column should give out a "Message Board Troll of the Week" award. Polaris is the front runner for this week.

Indy fans be forewarned. The FO Message Board Curse doesn't take kindly to irrational rants against DVOA.

191
by Chris (not verified) :: Wed, 11/15/2006 - 12:50pm

Re: 184

Based on my experiences making playoff forecasts for almost 10 years now, I definitely believe them. Last year, on week 8 of the regular season my predicted odds of making the playoffs in the AFC were:

Indy 98.74
Denver 90.06
Pittsburgh 89.18
Cincinnati 81.18
Jacksonville 71.64
New England 77.3

The odds for all of these favorites got progressively better as the season went on, except for Pittsburgh, who went through a dicey period. But ever if they had failed, 5 our of 6 would have been about the expectation of 6 events with the odds listed above. The NFC was not nearly so resolved.

In 2003, I had 4 of the playoff qualifies from each division at 70% or better by week 8. I missed on Minnesota, who I had in the high 90's, who crashed badly.

192
by Andrew (not verified) :: Wed, 11/15/2006 - 12:50pm

kevinNYC #162:

So you’re saying PHI playing 3 of the top 6 DVOA teams ON THE ROAD isn’t something to be concerned about?

If you are the top team on a per play basis, why should you be "concerned" about playing anyone? Aren't you theoretically better than all the rest? You speak as though a visting team has no shot here in these games merely from the fact of playing a strong team on the road.

Anyway, those games are a number of weeks away. Lets see what these teams look like then before we talk about how good or bad they are.

All I am saying is that Reid has gotten the Eagles to consistently go 6-2 or 7-1 when healthy. Its not unreasonable to expect him to do it again.

And anyway, look like what a steaming pile of crap the Giants have looked like at home against the only good teams they've faced there - the Colts and Bears - heck, even against the Texans!

Are you telling me that because the Giants crushed (in a relative sort of way considering they didn't break 20 points) Washington and Tampa Bay in the Meadowlands, the Eagles don't stand a chance to repeat what the Bears and Colts did to them at home?

Regardless of the overused “luck didn’t go our way� excuse

I know you hate to hear it, but for the Giants to win they had to:

1) Have Lito Sheppard out and Rod Hood injured and out in the 2nd quarter.
2) Recover all three of their offensive fumbles, one for a touchdown.
3) Recover the only Eagles offensive fumble and have the fumble be in scoring position.
4) Have a Brian Dawkins interception in overtime ruled a catch for Shiancoe with no chance for a challenge.
5) Have Trent Cole kick McKenzie in the nuts at the end of game to put Feeley in his field goal range
6) Have Kearse injured in overtime.

That sort of a perfect storm isn't going to happen again. If any one of those things didn't happen, the Eagles probably win.

I love how you only look at PHI’s offensive matchup against NYG & INDY, while ignoring what can happen the other way.

????

Philly is strong against TE's, RB's, and other wideouts, and weaker against #1's and 2's for a variety of reasons which may or may not hold up now that the team has a healthy defensive backfield for the 2nd half. Giants have Plaxico and Moss injured and Toomer on IR. I don't see the favorable match-up. If the Giants try the pound the ball method, they probably will find themselves in another 24-7 hole because points come from the passing game.

PHI allowed Eli Manning to have EASILY his best game of the season and INDY is a solid first in offensive DVOA.

Eli's best game?

50 drop backs resulting in:
8 sacks and 1 intentional grounding for -65 yards and 2 fumbles
10 incompletions
1 interception (and 1 more interception the refs couldn't bring themselves to award Dawkins)
30 completions for 372 yards and 3 touchdowns, most of the completions coming against third year #4 and #5 cornerbacks playing out of position for the first time in their careers because of injuries?

Gosh, that isn't something to really hang your hat on.

193
by QB (not verified) :: Wed, 11/15/2006 - 12:51pm

For reference, teams coming off of a bye went 17-11 this year and outscored opponents 629-489. Perhaps too small of a sample size to prove anything, but I'm certainly intrigued.

(It only adds up to 28 W/L's because I did not count games where two teams off a bye played against one another)

194
by Andrew (not verified) :: Wed, 11/15/2006 - 12:55pm

Nolan #163:

Once again with Pitt not in the very bottom DVOA you invalidate you so called rankins

I love these comments! Is this guy Borat from Glorious Nation Kazakhstan?

195
by BadgerT1000 (not verified) :: Wed, 11/15/2006 - 1:04pm

Very nice!!

196
by Andrew (not verified) :: Wed, 11/15/2006 - 1:17pm

JD #181:

I can’t imagine that only halfway through the season, things are nearly that sure.

Most playoff spots are determined by the head-to-head competition of divisional teams on similar schedules. The Giants, for example, have a leg up in the NFC East because they've won all their divisional games so far. The Eagles and Cowboys on the other hand, aren't out of it, because the Giants lost both their strength of schedule games, while the Eagles and Cowboys won both of theirs, and the Giants lost to the Colts, while the Eagles and Cowboys haven't yet played them and could possibly beat them.

Those sort of contests set 4 of the 6 spots. The other 2 spots are generally an open competition among at most 3-4 other teams. Typically, 8 of the teams in a conference just are not competitive for a spot, so they can be discounted.

Half way through a season, enough key games have been played to give a pretty good level of confidence in who will make it and who won't for at least 5 of the spots (the 4 likely divisional winners, and the next strongest team in the conference). The rest is a dog fight for spot #6, or an occasional battle royale between two or three closely matched divisional opponents (this year Charges/Broncos or Giants/Eagles/Cowboys) for possibly the #2 or 3 seed as division winner vs. the #5 or 6 seed for the runner-up.

The Eagles and Cowboys, for example, can put themselves right back in the NFC East lead by simply beating the Giants, if all other games go equally. If one of them also beats the Colts, they get an additional boost, because it is a win for them vs. a loss for the Giants at the end of year accounting.

197
by Andrew (not verified) :: Wed, 11/15/2006 - 1:46pm

SJM #186:

So the bye is a non-significant event when talking about expected wins the next week.

OTOH, Andy Reid's Eagles are 8-0 in the regular season, and 4-1 in the post season when given 2 weeks between games (only loss was the 2004 Super Bowl).

Sometimes, non-significant events are very significant for certain teams. For example, since 1990 there is no abnormal homefield advantage in the championship game of the playoffs. But most of this non-advantage stems from an inordinate number of homefield losses by the Steelers (1-4), Eagles (1-2) and 49ers (1-3). Absent those three teams, home teams have won 15 of 20 contests, a winning percentage similar to the home team in the divisional round. Also, since 1990, AFC #2 seeds have faired poorly in the playoffs, losing 50% of the time in their first game. But much of this has been the kryptonite-like reaction of the Chiefs (0-3), Colts (0-2), and Titans/Oilers (1-2) to that seeding. Other AFC #2 seeds are 7-1.

198
by admin :: Wed, 11/15/2006 - 2:13pm

Seattle (comment 142 above) now addressed in blog post (click my name).

Sorry about individual stats still not being updated. Caught up in charting TB-CAR last night, so they may not go up until tonight.

199
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 2:17pm

Let me clarify.

If you give me offensive or defensive DVOA to tell me that Team A's defense is better than Team B's defense, or Team C's offense is better than Team D's offense, I will say, "Yep, you're probably right." If you give me individual DPAR and tell me that, in context at least, wide receiver A is better than wide receiver B, I'll say, "Hmm, you've really made me think of those players differently."

But if you try tell me that 5-4 Philadelphia is better than 9-0 Indianapolis because the DVOA says so, I'll say, "Whaaa?" That's the overreach. I trust DVOA on total offenses and total defenses; I'm interested in the DPAR of individual players; I think using DVOA to come to one number to show that team A is a better quality team than team B regardless of records is going beyond what I'm willing to accept from a computer statistical system.

Statistical and scientific conclusions tend to lose accuracy when they overreach, by either trying to get too individual/specific or by trying to be too broad. In my opinion (which counts for little), most of the FO stats are great, and total team DVOA is in the overreach zone.

200
by SJM (not verified) :: Wed, 11/15/2006 - 2:20pm

Andrew #197,

I agree that in the Eagles case it appears that the bye is an advantage. I could argue that we have to look at their DVOA in those games relative to their season DVOA, but in the interests of laziness, I will grant you that certain teams might gain an advantage.

All I'm saying is that generalizing league-wide, the bye does not provide any advantage.

201
by Bulgaroktonos (not verified) :: Wed, 11/15/2006 - 2:56pm

I'm not sure I buy the idea that the missed field goal return is random or non-predictive. How often does this happen? Maybe I'm missing them, but it doesn't seem like they happen all that often. For the same team to do it twice in two years indicates, something, even if it's just that there's a Chicago special teams coach who's telling them to try it.

202
by Insancipitory (not verified) :: Wed, 11/15/2006 - 2:59pm

Polaris, from one Hawks fan to another.

DVOA isn't saying the Seahawks can't win, or won't win. It's not even saying the Eagles will be the Seahawks in Qwest. It's saying that on the average play, it remembers a lot of problems on both sides of the ball.

I ask you:
Have not Spencer at guard, Tobeck, and Gray occasionally be DOMINATED in the interior of the line, FORCING the Seahawks to introduce the shotgun? Did not Seneca Wallace in the shotgun catch more than a couple snaps that were less than on target just last sunday, and weren't there two fumbles that occured farther away from other offensive players, which put the game in jepordy, as a result of this? (Not that Seneca couldn't carry the ball like a running back when he's done being trying for the pass.) There's been a lot of instability on the offensive line. It has shown up on the field. Mack Strong isn't crushing people with his blocks like he has been either. To some extent that's why Seneca is playing at all.

Then there's the defense. They'd been playing poorly, out of position, giving up big plays. For most of the season. One could even say they lost a game in KC that Seneca (even with his interceptions) did enough to win. They have.

In both cases, these are often problems that don't solve themselves easily. The Seahawks have that, and seemingly a little bit of a return of the drops. In the case of the O-line, I do expect it to improve. It has, Oakland's defense is for real, even if the Rams aren't fans of playing the run. But DVOA *will* be predictive of further instability in the O-line. Gray almost didn't come back this year. The way this years gone for him, it's hard to imagine him choosing another year away from his family. So the line will almost certainly be shuffled again, and as great a local personality Pork Chop is, his injuries are a concern, and it's hard to see him as the man over the long term.

With defense it's a little murkier. On paper I have no reservations saying this is the most talented defense the Seahawks have ever had. For most of the season they set a franchise record for futility. What is a person to believe? Perhaps out of willful delusion I'll choose the former. But even if true this doesn't mean DVOA isn't doing it's job. The Seahawks defense DOES have a mountain to climb. Something a lot of bad defenses don't do. And for the Seahawks the 2006 defense has been historically bad.

DVOA only measures what's put out on the field. Linehan giving away the game was really nice. It gave me a nice warm Coughlin feeling. Let's remember Linehan coaches smart the Rams lead 29-21 and Seneca has to drive half the length of the field and convert for 2 points to face the Rams offense in overtime.

The short of it is the Seahawks have done just enough to be where they're at. Everyone knows if they're to go forward, they need to be prepared to do a lot more. DVOA makes it easier for us to estimate how much more.

203
by Pat (not verified) :: Wed, 11/15/2006 - 3:27pm

I think using DVOA to come to one number to show that team A is a better quality team than team B regardless of records

With only 9 games, you clearly can't make a linear ranking of teams based on wins/losses alone. No one does this.

If you try to reorder them based on who each team has played, and won, that rating is simply going to be inaccurate. It's too little data.

When you add in margin of victory, or the game results, Indianapolis drops quite a bit. I don't understand why people think it's so unreasonable that a more complicated evaluation of the game would drop them even more. Win-loss records are just far too quantized to use them as an evaluator of team quality.

Just to give an example, using only points, Sagarin's predictor ratings put Philly 6th, and the Colts 4th.

I don't get it. Football's a game. It's not intended to measure which team is better. Why would anyone think that wins/losses would be a good measure of a team's quality?

204
by Andrew (not verified) :: Wed, 11/15/2006 - 3:35pm

Pacifist Viking:

But if you try tell me that 5-4 Philadelphia is better than 9-0 Indianapolis because the DVOA says so, I’ll say, “Whaaa?�

DVOA is saying that given a large sample of 2006 Eagles and Colts playing on a neutral field, the Eagles will win more games and appear a better team.

The Colts this year have had a large number of close games (some against truly abysmal teams like the Jets, Titans and Bills) all break their way at the final buzzer. That is why they are 9-0. The Eagles have seen all 4 of their close games go against them, so they are 5-4. A couple of uncontrollable situations such as fumble bounces, referee calls, opponent coaching decisions, or opponent penalties going differently, and both of these teams could have records anywhere between 9-0 and 5-4.

Any game won by less than one score - while certainly won - is generally a game that could have gone either way with only a single play or a single bit of luck going differently, and is essentially a draw.

The Pythagorean Scoring Theorem holds the Colts as a 5.71 win team, and the Eagles as a 6.04 win team through 9 games. If the Jets had kicked a field goal on 4th down, the Bills not missed one, and Matt Bryant hooked his 62 yarder, would you be saying the same things about a 7-2 Colts team and a 6-3 Eagles team? Yet what could either the Eagles or Colts have done to actually cause those thing to not occur? Weren't they completely out of their control?

You are way too enamored of the Colts being a 9-0 team in terms of predicting future games by virtue of their escaping from the grasp of some bad teams through blind luck.

205
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 3:44pm

Well, Pat, I'm not suggesting abandon DVOA, but use some other system of ranking teams. I don't care about ranking teams overall; I check the standings for such things. I do care about ranking defenses or offenses in order, or even players in order; I think that can tell us something. But I won't depend on any 1-32 system.

"I don’t get it. Football’s a game. It’s not intended to measure which team is better. Why would anyone think that wins/losses would be a good measure of a team’s quality?"

I understand what you are saying, but I simply couldn't disagree more. The goal of a team is to win the game. It's not unreasonable to suggest that the wins and losses tell us things about the quality of a team. If a team is successful at achieving its goal--winning--I'll say that team is a better quality team than a team that fails to achieve its goal. It doesn't matter how well you play if you lose. If you want to use the DVOA to tell you that a 5-4 team is a better quality team than a 9-0 team, well, fine, do it. But I'm not going to. Maybe I'm using the word "quality" differently than you. I'm simply saying that a team that is more successful in achieving its primary goal is a better quality team than a team that is not successful in achieving its goals. There are a lot of min-successes in a game, of course, and isn't that what DVOA measures? The success or failure of particular plays? That's good, and it can tell us something, but I'm much more interested in the major success--a win--than in the quality of successful plays throughout a game.

206
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 3:51pm

"Any game won by less than one score - while certainly won - is generally a game that could have gone either way with only a single play or a single bit of luck going differently, and is essentially a draw."

But it DIDN'T go the other way. It WASN'T a draw. That matters to me, even if it doesn't to DVOA. Maybe chance was the main reason it went one way or another. Maybe intangibles do matter (leadership, confidence, nervousness, conditioning). Maybe a better team made necessary late plays to achieve the goal of victory. Whatever the reason, that matters to me.

For example, the Eagles lost a close game to the Saints. Maybe it could have gone either way. There was one penalty late that had a big impact, if I remember. But the Saints still managed to hold the ball for the last 9 minutes or so of the game. For whatever reason, they were more successful in that situation. It wasn't a draw--they took the win.

Perhaps I'm looking at it the wrong way. Whether it's predictive or not of future success, I don't know. If that's what it's intended to show, OK. Perhaps I see a 1-32 ranking and think it means that, right now, the #1 team is a better team than #7 despite the record. That I would disagree with. If it's predicting that the #1 team will have more future success than the #7, OK, I'm not qualified to disagree with it. If it's saying that thus far in the season, the #1 team (5-4) has been better than the #7 team (9-0), I would disagree.

207
by Wanker79 (not verified) :: Wed, 11/15/2006 - 3:51pm

Re: 199

That's a reasonable stance. But the reason you look at the total DVOA for Indy and Philly and say "Whaaa?", isn't necessarily because the system is overreaching. What that should be telling you is that you're not looking at the whole picture. You probably think of Indy and think about their offense and what the team did against Denver and New England in consecutive weeks. And say to yourself that there's no way a 5-4 team could measure up to them. Since you trust the unit-by-unit numbers, lets look at them.

Offense:
Indy has the move valuable offense in the league (34.30% WDVOA). They have the move valuable passing offense (59.80%), and the 8th most valuable rushing offense (4.10%).

Philly's offense may not be as valuable as Indy's, but they're still ranked 3rd in the league behind only Indy and SD (19.70%). Their passing offense may not be as valuable as Indy's, but they're still ranked 4th in the league (24.80%). But their rushing offense is more valuable than Indy's (1st at 18.10%).

Defense:
Indy's defense just hasn't been very good, ranking just 23rd (7.90% WDVOA). They've had a below average pass defense (18th at 7.10%) and a miserable rush defense (29th at 9.90%)

Philly has had one of the better defenses in the league, ranking 4th (-15.80%). They've had terrific pass defense (3rd at -21.10%) and pretty good rush defense (12th at -12.10%).

So Indy has the best offense in the league and one of the worst defenses in the league. Philly has one of the best offenses in the league and one of the best defenses in the league.

And don't give me injuries. Bob Sanders may have a big impact on Indy's rush defense, but don't forget that Philly spent 3 or 4 games with Joselio Hansen and Dexter Wynn seeing significant time on the field. And with Jimmy Johnson's blitz schemes, those injuries are very arguable a bigger impact on Philly than Sanders' injury has had on Indy.

208
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 3:58pm

207: And when the two teams play each other in a few weeks, I honestly don't know who will win. Philly seems to have a better balance of offense-defense than Indy. However, the fact that one team was able to finish out it's games, and another wasn't, means something to me, even if it doesn't mean much to DVOA. I haven't put all my faith in statistical systems--I still believe in intangibles that can affect the outcome. For whatever reason, through 9 games, Indy has done what it has needed to to achieve the primary goal: winning. Philadelphia has been less successful at that primary goal, even if they have been more successful at the mini-goals of successful plays. Thus far, to me, that means Indy has been the better team. Whether it means Indy will be better than Philly in the future, or whether Indy will even beat Philly, I don't know.

209
by admin :: Wed, 11/15/2006 - 4:03pm

I think I may need to add some "what is DVOA supposed to do" questions and answers to the FAQ. In the meantime, let me repeat a comment that I also made in the discussion thread five weeks ago:

With all of this discussion of whether DVOA is or is not more accurate than other ratings systems, I wanted to point out something important, something I mention every so often.

DVOA is not valuable because it is the most accurate total power rating. In fact, I would not be surprised if there was another stat-based rating out there that was slightly more accurate when it came to determining that Team X was slightly better than Team Y. (That's not to say I don't do a ton of work to make it as accurate as possible, of course.)

DVOA is valuable because it breaks down every single play. And because it breaks down every single play, it can be broken down into every single play — and into groups of plays, and situations. Thus, you can use it to analyze matchups, to analyze specific players, to look at how teams do on different downs, or in the red zone, or in the second half. You can use it to figure out why certain teams are winning instead of just saying "Team X is the best team in the league." You can use it to figure out trends that help you to predict how teams will develop in the future (i.e. the third down off-season thing).

No game in the NFL is decided by the fact that Team X is, in total, better than Team Y. It's ALL about matchups and specifics. DVOA can analyze that. Sagarin's rating, Beatpaths (no disrespect to Tunesmith) and the other stat ratings can not.

Sometimes, the fact that we run a list of teams from 1-32 each week will obscure this important fact.

Just to add: If it was as simple as "Team with higher DVOA should always be favored over team with lower DVOA," we wouldn't do those game previews on Fridays that feature specific matchup analysis, or the really complicated game previews come playoff time.

210
by Wanker79 (not verified) :: Wed, 11/15/2006 - 4:08pm

Pacifist, if you really believe that no matter what occurs in a particular game...no matter how the teams play in a particular game...that the only thing that matters is the final score, I really don't understand why you even come to FO. That is the EXACT reason for the existance of FO. And if you completely disagree with the very basic fundamentals DVOA is based on, why in the world would you waste your time coming here? I'm not trying to be confrontational. I'm absolutely serious and am interested in your answer. If that's all that matters, you can just go to any number of sources to find a ranking of teams by thier records.

In fact, here you go:

1.Colts 9-0
2.Bears 8-1
3.Ravens 7-2
..Broncos 7-2
..Chargers 7-2
6.Patriots 6-3
..Giants 6-3
..Saints 6-3
..Seahawks 6-3
10.Jets 5-4
...Jaguars 5-4
...Chiefs 5-4
...Eagles 5-4
...Cowboys 5-4
...Falcons 5-4
...Panthers 5-4
17.Bengals 4-5
...Packers 4-5
...Vikings 4-5
...49ers 4-5
...Rams 4-5
22.Bills 3-6
..Dolphins 3-6
..Browns 3-6
..Steelers 3-6
..Texans 3-6
..Redskins 3-6
28.Titans 2-7
..Raiders 2-7
..Lions 2-7
..Buccaneers 2-7
32.Cardinals 1-8

211
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 4:09pm

Aaron, that's actually what I wanted to hear and is sort of my thoughts on DVOA. I like DVOA because it does break down every play and give us valuable info on matchups and specifics. But I don't trust the total team DVOA because it only takes into account the mini-successes of particular plays and not the major-successes of wins and losses. It measures the success of the small goals but not the total goals. And that's fine, and it makes it a valuable tool.

212
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 4:12pm

210: As it happens, I do believe the final score is the only thing that matters. For the most part, so does every team in the NFL.

However, I also want to make sense of things beyond a W or L. FO does a better job of doing that than any other free source I've found. The data and commentary is valuable, helpful, and entertaining. All I've expressed in this threat is a skepticism about a statistical system that boils everything about teams down to one number and ranks them as such while ignoring the outcome; I haven't expressed doubt about the ability of statistics to help us make sense of things.

213
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 4:14pm

thread, not threat. I'm not setting up barricades or anything!

214
by Pat (not verified) :: Wed, 11/15/2006 - 4:32pm

But it DIDN’T go the other way. It WASN’T a draw. That matters to me, even if it doesn’t to DVOA. Maybe chance was the main reason it went one way or another. Maybe intangibles do matter (leadership, confidence, nervousness, conditioning). Maybe a better team made necessary late plays to achieve the goal of victory. Whatever the reason, that matters to me.

Keep this in mind: there is absolutely no predictive value to a win by less than 7 points. You can prove this: look at the divisional rematches. If a team wins by less than 7 once, there is no statistical improvement in their chances to win the next game. Against the exact same opponent.

215
by Wanker79 (not verified) :: Wed, 11/15/2006 - 4:34pm

All I’ve expressed in this thread is a skepticism about a statistical system that boils everything about teams down to one number and ranks them as such while ignoring the outcome.

See, you have me right up until that very last part. I really don't have a problem with anyone having skepticism about a statistical system like DVOA. And I really don't have a problem with anyone being skeptical that everything can be boiled down to a single number (Aaron as much said the same thing more or less). But the idea that the Jets only being able to get to their 32 in the final minute of their game against Indy while TB got the ball a whole 26 yards further downfield in the final minute of their game against Philly to kick the 3rd longest FG in the history of the league, says something good about Indy and something bad about Philly just makes me ill.

216
by Mnatr (not verified) :: Wed, 11/15/2006 - 4:41pm

#214

Then what is all your clamoring about how strength of victory should not be part of the BCS computer ratings about?

217
by Pat (not verified) :: Wed, 11/15/2006 - 4:43pm

Actually, the Eagles v Bucs, Bears v Cardinals is more apt. The idea that Rackers missing a short field goal (and even if it was tipped, it was tipped because it was a bad kick) means the Bears can win close games, and the Bucs kicker making a ridiculously long field goal meaning the Eagles can't win close games just doesn't hold water.

The smarter thing to do is say "Philly clearly isn't that great because they play too many close games". That's treating the close games exactly as they should be. But if that's true, then the Colts aren't that great either.

218
by Pat (not verified) :: Wed, 11/15/2006 - 4:45pm

Then what is all your clamoring about how strength of victory should not be part of the BCS computer ratings about?

Because statistical rankings don't decide who goes to the Super Bowl. Means a lot more when the statistical ranking is attempting to replace the game.

219
by jebmak (not verified) :: Wed, 11/15/2006 - 4:50pm

#216

The same reason that everyone does everything. Because they are idiots, thats why!

220
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 5:20pm

214: I believe you. That's why if DVOA team rankings are meant to be predictive, I'm not qualified to argue.

215: I understand: you're saying the idea that luck tells us something qualitative about a team is rotten. But look at human affairs: take out luck and tell me something qualitative about just about anything.

217: Yes, luck makes a difference. Of course. But it's incredibly hard, maybe impossible, to eliminate luck from anything and come to a statistical formula. Some luck still goes into DVOA (does it discount TDs on tipped balls?), some doesn't.

221
by Pat (not verified) :: Wed, 11/15/2006 - 5:31pm

215: I understand: you’re saying the idea that luck tells us something qualitative about a team is rotten. But look at human affairs: take out luck and tell me something qualitative about just about anything.

I think the basic problem is that you're not keeping qualitative and quantitative statements separate. "I think the Colts are a great team because they win close games" is a qualitative statement - so long as you don't expect anything from great teams, because expectations are quantifiable statements. That's a lot like saying that someone is a great Bingo player. Which is fine, although expecting a great Bingo player to win Bingo a lot is mixing a qualitative and quantitative statement.

222
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 5:38pm

Though I didn't say "I think the Colts are a great team BECAUSE they win close games." It just matters to me, in a perhaps intangible sense in a conversation about a statistical system, who does win those close games.

And you can't totally eliminate luck from the statistical system. Fumbles are penalized the same regardless of who recovers it. OK. But there are all sorts of other forms of luck that cannot be eliminated (nor am I suggesting they should be). Was a ball tipped? Did a defender fall down? Did the ref miss a hold? Etc., Etc. DVOA does it's best, but all I've been trying to do is suggest why I have a different view and different level of trust in statistical systems.

223
by Pat (not verified) :: Wed, 11/15/2006 - 5:48pm

It just matters to me, in a perhaps intangible sense in a conversation about a statistical system, who does win those close games.

But why does it matter? What does it change in your opinion about that team?

Besides, what's the point of a rating other than to measure possible future performance? You don't need a rating to measure past performance. That's what wins and losses are for.

And you can’t totally eliminate luck from the statistical system.

What's the point of this statement? Are you trying to suggest that just because all luck can't be eliminated, none should?

I don't think anyone's suggesting that all luck could be eliminated in any way.

224
by Richard (not verified) :: Wed, 11/15/2006 - 5:48pm

222: The refs miss a hold on every play where a hold isn't called.

225
by David Brude (not verified) :: Wed, 11/15/2006 - 5:53pm

I don't even think individual DPAR does a good job in evaluating players especially when its boiled down to a single number. EDGE didn't go from being one of the best RB's int he league to one of the worst overnight. I'm not sure any current RB wouldn't have a negative DPAR behind that terrible ARI offensive line.

It's only useful when you think about the context. Brandon Stokely wasn't one of the best WR's int he league a few seasons ago. He may have performed as such according to DVOA or DPAR but that was more due to the little attention he was receiving as the Colts 3rd WR.

I like to think of DVOA as a consistency rating. Sure on average the Eagles may be performing at a higher level than the Colts but due to a combination of bad luck or poor lapses at crucial moments they have not won as many games as the Colts.

Pacifist makes some really good points in this thread I think. DVOA doesn't honestly tell me much even when he get into more detail such as the team offense pages. The problem with football is there are many complex interactions and dependancies. I can't think of any other sport where the success of one player is so dependant on the play of his other teamates.

226
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 5:58pm

"But why does it matter? What does it change in your opinion about that team?"

Why does it matter? Like I said, the reasons it matters to me may be intangible in a discussion about a statistical system. But why shouldn't it?

"Besides, what’s the point of a rating other than to measure possible future performance? You don’t need a rating to measure past performance. That’s what wins and losses are for."

Since I'm not a gambler or a prognosticator, I prefer to use statistics to make sense of what is rather than use them to predict the future. As I've said repeatedly, if DVOA is meant to predict the future, I'm not qualified to comment on it.

"What’s the point of this statement? Are you trying to suggest that just because all luck can’t be eliminated, none should? I don’t think anyone’s suggesting that all luck could be eliminated in any way."

The reason for the statement should be pretty clear. I'm just expressing skepticism about a statistical system. Wouldn't the fact that a fair amount of luck does go into the system be a reason for me to be "skeptical"? Notice I'm not saying I abandon it totally. I'm saying I'm SKEPTICAL. Why is that a problem? Some luck is eliminated from DVOA, but a lot isn't and can't be.

You seem to suggest I should give less weight to a win-loss record because luck is a factor; at the same time, you question why I would be skeptical about a statistical system in which some luck is a factor.

227
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 6:03pm

Of course, even as I say that, I'd be even more skeptical of a system that attempted to eliminate all luck, as luck is such an important factor in life and in sports.

228
by Andrew (not verified) :: Wed, 11/15/2006 - 6:07pm

Pacifist Viking:

But it DIDN’T go the other way. It WASN’T a draw. That matters to me, even if it doesn’t to DVOA.

But you keep asking DVOA to do something it is not supposed to do. DVOA tells you how a team performs on a per play basis in the three phases of the game and by each skill player. It does not tell you whether or not a team will spread out having one of its occasional really stinky plays or mental breakdowns at exactly the wrong moment in 4 different games (like the Eagles), or concentrate them all into 1 or 2 games (like the Bears).

For example, the Eagles lost a close game to the Saints. There was one penalty late that had a big impact, if I remember. But the Saints still managed to hold the ball for the last 9 minutes or so of the game. It wasn’t a draw–they took the win.

Yes, they did, and DVOA reflects that by showing a poor Eagles defensive outing, because they coughed up 27 points. But, the Saints managed to hold the ball and win because of the penalty you mention (a 12 man on the field penalty on a 3rd down play that resulted in a sack out of field goal range). The penalty says nothig about the playing capability of either team for the future. It says something about the mental state at that moment of an Eagles rookie.

Perhaps I see a 1-32 ranking and think it means that, right now, the #1 team is a better team than #7 despite the record. That I would disagree with. If it’s saying that thus far in the season, the #1 team (5-4) has been better than the #7 team (9-0), I would disagree.

Why? The Eagles are 5-4 on a handful of flukey plays. The Colts are 9-0 on a handful of similar plays. Why would you think a team that could barely beat three bad teams like the Jets, Titans, and Bills is better than a team that was barely beaten by three good teams like Giants, Saints, and Jaguars? If the Colts were truly dominating, they would have crushed those three like they did to the Texans and Redskins.

I think it says something about how good the Colts really are that the Colts - if they continue to perform as they have and continue to win and go 16-0, would end up almost 6 games above their Pythagorean projection, when no other team in NFL history has ended up more than about 3 games over. Its says their winning ways are unsustainable unless they pick up their play, so you are putting too much into the 9-0 record.

229
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 6:16pm

geez, Andrew, the main thing I've said is that I have SKEPTICISM about the statistical system. Part of that skepticism is because fluky plays happen fairly frequently and do determine a lot, that luck plays a big role, that I do believe in intangibles that aren't quantifiable, and other things.

I recognize DVOA measures particular plays; that is part of its strenght, but it's part of my skepticism. It measures the overall success of mini-goals (particular plays) without taking into account the major-goal (a win). That's what I've said. It's a good tool, I like FO stats, but I don't trust the DVOA team rankings because it focuses on the mini-goals and not the major-goals. It's a different worldview--I'm not quibbling with what the system can do, but pointing out why I don't trust it, or, if you prefer, why I have a different worldview.

A lot of commenters here love to critcize "conventional wisdom" and specific media commentators. That's fine. I enjoy that too. But even expressing doubts about the statistical systems here is seen as ridiculousness. I believe there are a lot of non-quantifiable, intangible factors that can determine outcomes: luck, confidence, leadership, nervousness, conditioning, etc. Because of my belief in these non-statistical factors, I have doubts about DVOA team rankings. THAT'S ALL I'M SAYING.

230
by Andrew (not verified) :: Wed, 11/15/2006 - 6:16pm

Pacifist Viking:

Of course, even as I say that, I’d be even more skeptical of a system that attempted to eliminate all luck, as luck is such an important factor in life and in sports.

But luck is not predictive. The Eagles are a really good team, but in part due to poor luck recovering fumbles, they are 5-4. If their luck in the future recovering fumbles is league average, they will win a lot more games than they have. That is what DVOA tells us.

Why is that so hard to understand? DVOA doesn't say the Eagles are 5-4, so they suck. You can read Pete Prisco for those insights. DVOA says that despite being 5-4, the Eagles are a really good team.

231
by Crackertl82 (not verified) :: Wed, 11/15/2006 - 6:18pm

I just watched something about your ranking system on ESPN and had to reply to it. I could care less about 2 through 32, what I want to know about is Chicago being number 1, instead of Indy. You claim that you base your rankings based on luck, and strength of schedule. Strength of Schedule, The Colts have beaten Jacksonville, The Giants, The Patriots, The Broncos, who else do they have to beat to have a hard schedule in your eyes. The Bears have beaten the NFL's cellar dwellers before this week, in the weakest division in football outside of the NFC West, and beat the Giants mainly due to the momentum swing on a 108 yard play that has happened like that twice in NFL history. The Bears are by far the most overrated team in the NFL since the Chiefs a few years ago, who played I believe 12 losing teams, and lost to 3 of the 4 winning teams they played, before being exposed by the Colts in the playoffs. This system is whack, sorry.

232
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 6:20pm

And why is it so hard to understand what I've said repeatedly? IF DVOA IS MEANT TO BE PREDICTIVE, THEN I AM NOT QUALIFIED TO ARGUE WITH IT.

Aaron's comments in #209 answered a lot of my questions. I look at a 1-32 ranking and disagree with the ranking. But if DVOA isn't meant to rank what has happened but to be predictive, then I'm ootu of my league.

233
by Crackertl82 (not verified) :: Wed, 11/15/2006 - 6:32pm

This is why my mother generally gets a better record than all of the supposed experts on football. She has won her football pool 5 times this year, and has a better record than all of the HBO columnists, and ESPN columnists, because they're too wishy washy, and jump off and on bandwagons like they're full of hay and on fire. I picked the Giants this week because they were at home, and they are a better team, the better team doesn't always win. The Cardinals in baseball weren't the best team in the major leagues, but they still won the world series. Who cares who's supposedly better, when the better team in most peoples eyes hasn't won it all in alot of the major sports throughout sports history.

234
by Crackertl82 (not verified) :: Wed, 11/15/2006 - 6:33pm

I also love how the 49ers are ranked lower than all 4 teams that they've beaten. Is there a mentally retarded kid running this show or what, I mean come the fuck on.

235
by Wanker79 (not verified) :: Wed, 11/15/2006 - 6:35pm

You seem to suggest I should give less weight to a win-loss record because luck is a factor; at the same time, you question why I would be skeptical about a statistical system in which some luck is a factor.

You realize you're agruing that a system that is based on 16 data points and takes absolutely zero effort to remove the effects of luck is better at determining the quality of a team than a system based on a couple thousand data points that tried to eliminate as much luck as it can?

236
by Wanker79 (not verified) :: Wed, 11/15/2006 - 6:41pm

Re: 190

Scramble for the Ball or some other column should give out a “Message Board Troll of the Week� award. Polaris is the front runner for this week.

I think you spoke too soon.

237
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 6:42pm

235: Yes. Yes I am. But you seem to not pay attention to what I've been expressing, which is a distrust of statistical systems in general. Is that a problem? I mean, am I OK? Am I a complete idiot because I don't believe math can explain the world?

238
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 6:48pm

Wanker, I'm not sure the numbers of the statistics matter that much. The English won virtually every battle in the Hundred Year War; the French won the war. The war was devestating for France, but they won the war. Why did that happen? Lots of reasons. I suppose you could argue that because the English won most of the battles they were a better quality military nation than France. And yet, France won the war.

Look, if the fact that I distrust math to explain all the world for us makes you think I'm an idiot, fine. Think I'm an idiot. I don't think you're an idiot for trusting the math, if that makes a difference. But if my different view makes me inferior and foolish, then let it be so.

239
by Wanker79 (not verified) :: Wed, 11/15/2006 - 6:52pm

factors that can determine outcomes: luck, confidence, leadership, nervousness, conditioning, etc.

How could you have forgotten about swaggar?!? Or a feeling of disrespect?!?

I mean, am I OK? Am I a complete idiot because I don’t believe math can explain the world?

Absolutely not! I assume that every single person that come to FO has some level so skepticism (including Aaron otherwise we'd still be using DVOAv1.0). I think the problem most of us are having is that you are trying to tell us that random events (while certainly having a huge impact on the outcome of the games) tells us something useful about a particular team.

240
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 7:03pm

See, in your response to my claim about intangibles, it seems pretty clear you hold intangibles in contempt. That's fine. That's where we disagree. I have a different view.

241
by Insancipitory (not verified) :: Wed, 11/15/2006 - 7:06pm

It seems to me there's a wide world of difference between a statistical model revealing all truths of the physical world and a statistical model proving to be a useful tool to catch intuition's gaffes. It's not going to be right all the time; a merciful concession I gladly accept as a Seahawks fan. But it is instructive and proven (at least somewhat)predictive.

242
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 7:06pm

Re: random

Some of the things assumed to be random may not be random; some of the things assumed to be non-random may be random. Does that make more sense? Is that reason enough to be skeptical?

If the Colts are a lesser team because they've gotten lucky, or the Eagles are a better team because they have had bad luck...well, are there some areas where maybe the Colts have had bad luck but that is still in DVOA? Or the Eagles have had good luck and that's not in DVOA? I don't have specifics, I'm just saying.

243
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 7:09pm

241: I agree. If we can use statistics to prove some conventional wisdom wrong, or to get a more accurate understanding of what is actually going on, I'm all for it.

In this thread I've probably taken on an "anti-FO, anti-stats" mantle. I don't mean to. I find statistics useful to help us make sense of reality--but I use them to help understand reality, not encapsulate it all.

244
by TomC (not verified) :: Wed, 11/15/2006 - 7:17pm

just like last year, chicago will be shown
to be the pretenders they are come playoff time

Denny Green is so innovative, his haikus have only two lines.

245
by Andrew (not verified) :: Wed, 11/15/2006 - 7:17pm

factors that can determine outcomes: luck, confidence, leadership, nervousness, conditioning, etc.

How could you have forgotten about swaggar?!? Or a feeling of disrespect?!?

Don't forget "emotion", "heart", "experience", "concentration", "focus", "determination", "nobody believes in us", "chip-on-the-shoulder", "come out flat", "playing with attitude", etc., etc.

Marques Colston is the NFL's leading receiver because he has more heart and swagger than Holt, Walker, TO, Moss, and Johnson. Right?

246
by stan (not verified) :: Wed, 11/15/2006 - 7:18pm

225,

There are other problems which neither DVOA nor any other stats formula can address -- coaching strategy and changes thereto. Let's face it, some gameplans are better than others. Some coaches are better playcallers than others. And some days they are better than others.

And teams are not the same from week to week (even without considering injuries). Some players improve over the season, some decline.

Add matchup specific issues from game to game, and you can only do so much with stats.

One thing that drives me crazy about some stat geeks is the way they look at pass numbers or rush numbers and make conclusions about the players on team X as a result. Often their conclusions are totally ass-backward.

If you have a weak defensive front against the run, you may choose to bring one or both of your safeties into the box. This puts your corners on islands and leads to more passing yardage and less rushing yardage given up. Or perhaps you leave a better run defender in at LB rather than the pass defender because the line needs the help.

Bottom line -- unless you closely analyze the tape, you have to be very careful relying on stats to tell you what part of a team is good or bad.

247
by admin :: Wed, 11/15/2006 - 7:21pm

Here's the problem with intangibles: They are not tangible. Therefore they cannot be measured in numbers. Therefore they are not used in a statistical system.

We return to this line that I publish in every week's DVOA commentary, and I publish it for a reason: "Remember, of course, that any statistical formula is not a replacement for your own judgment, just a tool to use in analyzing performance."

248
by Zac (not verified) :: Wed, 11/15/2006 - 7:23pm

Re: 112 (Dallas/Indy line, in case you don't want to scroll up that far).

According to USA Today ( http://covers.usatoday.com/data/odds.aspx ), Dallas was in fact a 1-pt favorite when the lines opened. Since then, they've moved to Indy being favored by 1 or 2 points. Which means the general public (which anyone can tell you are idiots) moved the line.

248
by Andrew (not verified) :: Wed, 11/15/2006 - 7:23pm

Pacifist Viking:

See, in your response to my claim about intangibles, it seems pretty clear you hold intangibles in contempt. That’s fine. That’s where we disagree. I have a different view.

Its not a contempt. Its a healthy suspicion of the same types of buzzwords you hear thrown about on Financial TV shows to explain the random fluctuations of the stock market. Your know:

"The market went up today on confidence that the Fed would cut interest rates".
"The market went down today because of profit-taking."
"A earnings warning in the tech sector sent the entire market into a -300 point nosedive."

Teams don't win because of something like heart (do you really believe one team comes out saying to themselves, "you know what, we don't care if we get our ass really physically kicked today, this game doesn't matter, so lets let oursevles have the everliving shit kicked out of us for 60 minutes"?), just like profit-taking doesn't really make stocks go down (gosh, don't people take profits in the market every day?)

250
by Pacifist Viking (not verified) :: Wed, 11/15/2006 - 7:36pm

245: just like in Wanker's response, your response suggests a contempt for intangible factors. They do exist; the fact that the conventional wisdom uses absurd intangibles and catchphrases doesn't mean there aren't intangible factors at work in sports.

247: Aaron, that's all I've been saying. The intangibles can't be measured, but because I believe they exist, I have skepticism about the statistical systems. I have nothing to disagree with you with in that commment. I think my semi-rants have been directed at how commenters respond to DVOA and to doubts about DVOA. I accept totally your statements that DVOA is a "tool to use in analyzing performance,� and that only tangibles can be measured in numbers.

249: Maybe teams don't win with heart, but maybe they don't win with numbers, either. It's fine to be suspicious of conventional wisdom's use of intangibles. I hope it's also fine to be suspicious of statistical systems because they (inherently) must ignore intangibles. I believe they exist and can affect outcomes.

251
by Pat (not verified) :: Wed, 11/15/2006 - 8:07pm

Some of the things assumed to be random may not be random

When people say "random" here, I don't really think they mean what you think "random" means. They mean "uncontrollable." Things can be random for one team and not for the other (Bryant's success kicking the FG against the Eagles is random for Philly, not for Tampa Bay).

doesn’t mean there aren’t intangible factors at work in sports.

This is the problem: you're suggesting that intangible factors have an effect on winning games, and yet one of the ones that you quote (an ability to win close games) does not.

If something's measurable - and teams' future performance after a close win is measurable - then it's not intangible.

252
by Mnatr (not verified) :: Wed, 11/15/2006 - 8:29pm

#251

Exactly. The intangibles people often bring up should actually show up in the stat line.

253
by bravehoptoad (not verified) :: Wed, 11/15/2006 - 8:29pm

The fact is that the system is flawed. It favors teams like Philly and Pitt too much....

I wonder what kind of team that is? Pennsylvania teams?

254
by MdM (not verified) :: Wed, 11/15/2006 - 8:35pm

I think people either believe in stats or they don't. If they dont...might as well go read Peter King, they're not changing their minds.

255
by Richard (not verified) :: Wed, 11/15/2006 - 8:36pm

253: I think he means teams that play well, but lose.

256
by Stillio (not verified) :: Wed, 11/15/2006 - 10:14pm

How dare DVOA tell us teams that play well but lose might be better than teams that play poorly but win? I'm outraged! Outraged I tells ya!

257
by Icky (not verified) :: Wed, 11/15/2006 - 10:17pm

HaHaHa,

The Hawks are ranked 22nd, and they'll prolly still find themselves in the NFC Championship game, if not the SuperBowl.

Silly NFC. -Hawk Out!

258
by navin (not verified) :: Thu, 11/16/2006 - 1:48am

Re: 248,
Vegas lines don't work the way you described them. Vegas wants minimal movement on their lines. That way they draw equal money on both sides, and are guaranteed a profit because of vigorish.

If Indy ends up covering, Vegas is going to lose money on the game, and the oddsmakers are going to feel a lot of wrath because of it.

FWIW, a lot of LSU and Florida lines have moved in college this year, and the team that it moved towards ended up covering a couple times. I always get a laugh when that happens.

So Vegas thought that Dallas -1 Indy would draw equal amounts of money on both sides, and because it didn't, the lines have moved.

259
by QB (not verified) :: Thu, 11/16/2006 - 2:57am

I doubt that Vegas linesmakers were naive enough to expect equal money on Dallas -1. Personally, I don't think they mind gambling a little bit when they're fairly sure the money is on the wrong side. They get the vigorish either way. One regular season game will not break the bank for any decent sportsbook.

260
by Polaris (not verified) :: Thu, 11/16/2006 - 5:11am

#190 et all,

Anyone that views DVOA with a healthy degree of skepticism (and I have said repeatedly it's a very useful metric) because of it's failure as a team predictive measure is automatically labeled a "troll" here.

Nice. You are a real piece of work.

-Polaris

261
by Pacifist Viking (not verified) :: Thu, 11/16/2006 - 10:40am

Look, I really didn't intend to start an argument in which I'm questioning everything. Really, I didn't. All I've tried to do is explain why it is reasonable for a person to be skeptical of a statistical system. If I've gotten more defensive and taken the arguent further than intended, it's because some commenters seem to deny the plausibility of this pretty modest idea. All I'm saying is that I'm not unreasonable because I have doubts about statistical systems.

Football players aren't pixels, and they aren't out there with calculators. They're human beings dependent on their mental and physical skills for success. If in the fourth quarter, a particular offensive tackle is tired, or hurt, or angry, or confident, or frustrated, or nervous, or motivated, that can make a difference in his performance, which can effect in outcome. Is this quantifiable? No. Should this be fit into a statistical system? No. Does this mean that all statistics should be abandoned? No. But it does mean that there's more to a football game than numbers are capable of showing. And if that's the case, is it such a preposterous idea that it's OK to be skeptical of statistical systems?

I am perfectly willing to accept FO's statement that DVOA is to be used as a tool in analyzing performance.

251: I am completely willing to accept that my use of "random" is not the use of random of most FO writers and readers.

254: "I think people either believe in stats or they don’t."

Is it really an either/or proposition? That one must either have full faith in the stats, or none at all? That's like a Communist regime taking over and saying, "Your previous ideas were based on the brainwashing of bourgeois ideology. You must now accept our ideology."

260: That's how I've been feeling. Some of the commenters, not FO writers, are like, "We have successfully debunked bourgeois ideology. We now expect you to fully follow communist ideology, or we will assume you still hold remnants of bourgeois ideology."

262
by Andrew (not verified) :: Thu, 11/16/2006 - 11:12am

Pacifist Viking #261:

Football players aren’t pixels, and they aren’t out there with calculators. They’re human beings dependent on their mental and physical skills for success. If in the fourth quarter, a particular offensive tackle is tired, or hurt, or angry, or confident, or frustrated, or nervous, or motivated, that can make a difference in his performance, which can effect in outcome. Is this quantifiable? No.

But of course it is measurable!

If a O-Tackle is playing poorly for some intangible reason, it is going to show up as QB sacks, QB hits, QB hurries, and running plays blown up on his side of the line. His intangible reasons for poor play will come out measurable in tangible accomplishments of himself, his teammates, and his opponents.

By definition, if intangibles are not measurable, they are not affecting the game, and are therefore irrelevant, because the game of football is a result of tangible, measurable, actions - yards, first downs, tackles, touchdowns, safeties, defensive stops, penalties. Nobody wins football games by "showing more emotion" or "having more heart", etc. without producing on the field.

If an allegedly inferior O-Tackle has "more heart", if you will, or is "hungrier for the win", or similar blather, he will shut down his supposedly superior D-End opposing him, and we will be able to measure this in a lack of the normal pressure this D-End is able to generate off the edge, and a greater production in running yards than this D-End normally allows.

No one is denying the reality of human emotion and emotional factors. But if those intangibles play a real role in the game, they will cause measureable results, and DVOA and associated statistics will pick it up.

We see it every week when a supposedly top-notch cornerback "loses his focus", for example, and is torched a few times by a middling receiver. Like the Great Thanksgiving Sunday Burning of Champ Bailey in the snow in Denver in 2004 by Jerry Porter.

263
by Pacifist Viking (not verified) :: Thu, 11/16/2006 - 11:18am

But the mood of a player, while it will show up in the stats somehow, is not consistent or predictable.

"By definition, if intangibles are not measurable, they are not affecting the game,"

I question the logic of that statement, but I understand the point you are making about intangible factors leading to tangible results.

By the way, I hate cliches. So I hope you're not pinning cliches on me just because I believe that some of the factors that affect human existence--including sports--are not tangible.

264
by Pacifist Viking (not verified) :: Thu, 11/16/2006 - 11:30am

All I've suggested is that stats can't tell us everything and that it's reasonable to have doubts about a statistical system, but the way some readers are responding, it's like I'm trying to tell you that Santa Claus exists and that he possesses Eli Manning and that's why his passes are inaccurate.

265
by B (not verified) :: Thu, 11/16/2006 - 11:49am

But if Santa Claus exists, wouldn't he dislike the Eagles? And therefore he'd want to make Eli's passes more accurate, not less.

266
by Andrew (A.B.) (not verified) :: Thu, 11/16/2006 - 12:01pm

Some surprises (to me, anyway) in this week's DVOA:

Run offense
#11: Jets
#24: Pittsburgh
#28: Kansas City (!)

Run defense
#17: San Diego

Pass defense
#6: Cleveland

Any thoughts on these? I think DVOA tells a much different story than the conventional wisdom on all of these.

267
by Wanker79 (not verified) :: Thu, 11/16/2006 - 12:12pm

But the mood of a player, while it will show up in the stats somehow, is not consistent or predictable.

It really doesn't have to be. If a particular player's performance is greatly effected by insonsistent intangable (like nervousness or concentration) the tangable result (which should show up in DVOA) will be a player with a high degree of variability. DVOA won't tell you WHY a player is good one week and bad the next, and it doesn't have to. All it has to do is tell you that he IS good one week and bad the next.

And you know what, I've never really thought about it that way before. It seems like we're kinda agreeing on the intangables now (and I have a feeling Andrew is right there, too). Intangables do effect the outcome of games, but now what we're disagreeing on whether or not the all the relavant intangable cause a tangable effect that DVOA will recognize.

BTW, I don't think anyone who has responded to you has treated you as a troll. You've brought up a topic that has led to a rather interesting debate (which is a good thing). Even in disagreement, you've brought something to the discussion. Unlike some people who just attack for the sake of attacking because they have some axe to grind.

268
by Andrew (not verified) :: Thu, 11/16/2006 - 12:32pm

Pacisfist Viking #263:

But the mood of a player, while it will show up in the stats somehow, is not consistent or predictable.

Yes and no. Sometimes certain players consistently have bad games or good games against the same teams over and over again, over a period of many years. Jake Plummer has played great against the Eagles for 10 years now, but he has yet to have a good game against Miami, at Kansas City or at Dallas.

I question the logic of that statement, but I understand the point you are making about intangible factors leading to tangible results.

When Todd Pinkston had a "lack of focus" against Ricky Manning Jr. in the 2003 NFC Championship game, the results were obvious on the field - no receptions, 2 or 3 interceptions off passes to him. Could it have been predicted? Well, Ricky predicted it before the game ...

By the way, I hate cliches. So I hope you’re not pinning cliches on me

Just some good natured ribbing!

269
by Oh, Mathematics (not verified) :: Thu, 11/16/2006 - 12:45pm

I definitely agree with some of what Pacifist is saying (and I am one of those who believes math can explain everything), but I'm not sure I agree with how he is applying it. "Intangibles" definitely affect the game (and in a quantifiable way) but they are most certainly non-predictive. Statistics say that Philly should beat Indianapolis. But suppose that Dallas has more heart/confidence/&c than Indy this week, and ends up making them look like practice squad scrubs. Indy bounces back with their most preparation of the season, and just puts up an incredible showing against Philly, making key defensive stops when necessary and the offense rolling on all cylinders. Indy wins the game, outplays Philly (maybe not even by very much, as measured by score or DVOA), and the board is flooded with Indy trolls who want to get out their "told ya so's" (which of course leads to Indy losing their next 5 in FOMBC fashion).

So Pacifist, Indy wins the game, due to the fact that they are the better team in that game, which is due to "intangibles". Are they the better team overall? Who knows really, intangibles have a huge role in football games (as you've said) and teams generally fluctuate pretty wildly from game to game. However, the most important question I think is whether or not we can learn something from this performance. Does Indy's recognition of their mortality against Dallas and their subsequent increase in work ethic/confidence/heart/&c to show their critics wrong have any predictive value for future games? The absolute best we can do in order to make these kinds of predictions is to analyze the team's entire body of work (as DVOA does) and compare them to that of the team they are playing. That doesn't mean that Houston couldn't easily jump up to a 21 point lead on Indy based on some flukey plays and kill Indy's confidence and win the game. Trying to predict which teams will have those intangibles from week to week is going to require that you enlist the help of a psychic (or at least the dog whisperer).

FWIW I want nothing more than for Indy to trounce Dallas and put up no fight whatsoever to the Iggles.

270
by B (not verified) :: Thu, 11/16/2006 - 1:32pm

So a team with great intangibles would play well but be inconsistent. Also, they would most likely play their best games against tough opponents, because those are the games where they have to prepare, and would bring their grit, pluck, heart and determination. So the team with the best intangibles should be Jacksonville, right? This actually explains a lot.

271
by Benjy\\\'s Cousin (not verified) :: Thu, 11/16/2006 - 1:49pm

#4 - no - they won't give a hoot. 2 games up over NO, and 2.5 games over NYG/SEA for the #1 seed in the NFC? Why would they care if DVOA ranks them #1 or #2 or #10?

With Philly's schedule (@ Indy, @ Was, @ NYG, @ Dal, vs Ten, vs Atl and vs Car), they are going to struggle to make the playoffs.

272
by chris clark (not verified) :: Thu, 11/16/2006 - 2:00pm

Without stirring the pot too much, just let me point out that DVOA is a metric derived from a summary. Not only does DVOA omit intangibles, it omits many tangibles. To be more clear DVOA is a number (or set of numbers Off DVOA, def, ST, weighted, etc.) that try to quantify the information contained in the play-by-play. If I understand correctly, DVOA is tuned to correlated to long term PF/PA numbers. To do that is emphasizes certain information contained in the play-by-play and de-emphasizes other information. However, it doesn't inject information that isn't in the play-by-play. Thus, since the play-by-play doesn't include certain important and valuable information, e.g. the formation, who was blitzing, who blocked whom, who whiffed on a tackle, etc. That information is not in the DVOA either. Those are all tangible pieces of info that aren't factored into DVOA. Those are all factors that are important in how successful a play was.

As a result, DVOA (but not VOA?) also omits other bits of statistics, e.g. who recovered a fumble. Not because they aren't important factors in the game, but because without similar detailed tangible information on the formation and missed blocks et. al., one cannot predict how sustainable that statistic is. It may be sustainable and it may not be. However, the information contained in the play-by-play doesn't allow you to predict it reliably.

Therefore, if you watch enough PHI and IND games, you may have insight into which of their characteristics are not captured in the play-by-play summary and thus know whether one teams DVOA actual over or under estimates their future success. I don't have that knowledge, although I have my suspicions.

Unfortunately, such insight is likely to be team-specific (and perhaps season specific) and not something that Aaron will ever be able to quantify into DVOA in a satisfactory and justifiable manner. That insight requires subjective judgement, because it doesn't unversally apply to all teams in a uniform manner.

Curiously, I saw the long-term statistical predictions (that got posted on this site last week and roundly panned) as an attempt to do that. I think the presentation of the data as being cyclic caused everyone to miss the fact that it was trying to account for long-term team specific trends.

Let me make this concrete by applying a cliche to a specific team. I'm not picking on that team as I don't know the facts.

Perhaps, this year PHI is good at finding ways to snatch defeat from the jaws of victory. That might be classified as the "conventional wisdom". Given that this isn't the first year that PHI has done that, perhaps one might think that the DVOA measurement is overstating PHI's probability of winning, because DVOA isn't accounting for that intangible "fact".

For another illustration, let me refer back to the famous play a few years back where Marvin Harrison caught the ball and went down without being touched by a Bronco defender and subsequently got up to score and the Colts won the game. I do't believe there was any way to predict that from preceding game play-by-play summaries. However, being a Broncos fan at the time, I was not surprised, because the team often played just inconsistently enough to lose to other top teams. You didn't know which play it would occur on, but you had a perpetual fear of it happening in any given game. Can one quantify that? I doubt it. Was it real? It happened didn't it.

Could one predict it, and if so how? That's the real question. Attempting to do so gives you stats like Farve has never won in a dome stadium where the opponent had a bye the week before. Fact, yes (well, I don't know about this case, I just made it up). Relevant, depends on if there is a causal relationship and not just a correlation of events in the past.

To some extent stats are only as good as they are convincing (and reliable). An unbelievable but reliable stat is useless, because you won't believe it. A believable but un-reliable stat is bad, because you will believe it and you will be wrong.

And, the judgement one applies is knowing how much one believes any given stat compared to ones saying something different.

273
by Ray (not verified) :: Thu, 11/16/2006 - 2:14pm

Pacifist Viking, please don't feel too persecuted. Your stance is completely reasonable, and everyone here should recognize that. But you're running into some people who love to debate this stuff, and so long as you keep up your side, they will too. Given that you've been completely respectful in all your comments, I doubt anyone here will have hard feelings towards you, and certainly no one will think you a troll. Please don't get scared away by the 'believers' here. I think you'll fit in just fine.

274
by chris clark (not verified) :: Thu, 11/16/2006 - 3:10pm

re: 249

"profit taking" is actually a bit of jargon that describes a specific set of actions (selling stocks that have run-up over a short period of time). It may not be meaningful in the explanatory sense or when looking at the words at face value. However, if I heard any one of your phrases on a stock market tv program, I would understand what they were describing. BTW, no I don't "take profits" everyday, just when I think a stock is likely ahead of it's curve. That represents me as a stock market investor, gambling on what I think the future performance of that stock is likely to be. Personally, I try to buy on "profit taking" days and sell on the day before. However, that's me gambling on what I think other investors think.

Bringing this back to football. Certain cliches encapsulate knowledge about football, that's how they became cliches. It's just like certain cliches encapsulate info about the stock market. However, in both cases, the cliches only impart partial knowledge.
The same goes for stats.

You know more about a team when you are told that "it finds ways to lose" than you did before you were told that fact, just not a lot more. You also know more about a team with the best DVOA in the NFL than you did without that info. However, it isn't enough to tell you what the team's W/L record is though.

275
by Andrew (A.B.) (not verified) :: Thu, 11/16/2006 - 3:23pm

I think it's pretty clear that DVOA does factor in intangibles. It's right there in the chart, under the heading "Var.".

276
by Andrew (not verified) :: Thu, 11/16/2006 - 3:56pm

chris clark #274:

BTW, no I don’t “take profits� everyday, just when I think a stock is likely ahead of it’s curve.

But the market as a whole does take profits every day. Reducing it to yourself is like Chad Johnson saying, damn, I had just 2 touchdowns in 8 games. The NFL is really lacking in touchdowns to wideouts this year!

277
by chris clark (not verified) :: Thu, 11/16/2006 - 4:37pm

re 276:

No, that is not "profit taking". Yes, people do make profits in the market everyday and some people sell everyday, but "profit taking" means a drop in the price of the stock after several days of gain. That does not happen everyday.

I do not mean that when I sell a stock I am "profit taking", although I may be cashing in on some profits. I am only doing so when my actions (in concert with other investors) cause the price of a stock to drop after it has gone up for several days.

"Profit Taking" is just jargon (a cliche term). If you speak the jargon, it makes sense. If you use the jargon term to mean something else, you are abusing the term, and potentially confusing your readers who may expect the precise jargon meaning, and go "what are you talking about?"

Confusing the term "profit taking" with the making of profits, is like taking the acronym DVOA apart and saying what do the words mean. DVOA means a specific stat the FO calculates and by extension the related stats which FO also calculate. If I used DVOA to mean W/L records or the Aikman efficiency index, I would be misusing the term, even if I could come up with an argument that those numbers were somehow "defense-adjusted values over average".

I think the point you are trying to make is cliches get overused to the point where they are meaningless (the pejorative meaning of the word cliche). That is true. Just because a pundit uses a cliche to describe a situation doesn't mean the pundit has done the underlying analysis to verify that the cliche is applicable.

But, this has little to do with football.

278
by chris clark (not verified) :: Thu, 11/16/2006 - 4:55pm

re 275:

Do you think the variance in the chart is sufficient to explain the difference in PHI's DVOA and W/L record? If so, then the intangibles are accounted for. If not, then they are not (in this case).

279
by Pat (not verified) :: Thu, 11/16/2006 - 5:07pm

Look, I really didn’t intend to start an argument in which I’m questioning everything. Really, I didn’t. All I’ve tried to do is explain why it is reasonable for a person to be skeptical of a statistical system.

Of course it is. There are things the statistical system doesn't measure. Drops by wide receivers. Blown coverages by individual players. Injuries. Point of completion. It also can't measure the performance of plays that didn't happen, as well.

But in the case where you disagree with a statistical measure, you should be able to identify a shortcoming, and then no one would disagree with you.

The problem with just being skeptical because of an opinion about a team is that the measurement is unbiased, whereas your opinion almost always is. Using opinions to look for a flaw is fine, but basing an entire criticism on it is a bit odd.

280
by stan (not verified) :: Thu, 11/16/2006 - 5:55pm

Hey Viking!

I don't know if this is what you mean by "intangibles", but it definitely affects games and cannot be quantified by any stats based system -- that is the ability of some players and coaches to adjust based on scouting and reading opponents during a game. Perhaps this can be considered part of the "matchup" problem.

Some teams and players are more easily "read" than others. Team A or B or C's coaches and receivers may not pick up on a coverage tendency or tip-off by team X. Team D has a coach or player who regularly picks up on the tip-off. So he or his team performs better against team X than we would be able to predict based on the results of all the other games.

This isn't luck. It is simply a skill we cannot measure. And it affects the outcomes of games all the time.

There are all kinds of ways that offensive players tip off plays (alignment, finger tip pressure, licking fingers, speed out of the huddle, length of time the QB speaks in calling a play in the huddle, etc.) Coaches get in play-calling rhythms and habits.

The teams and players who are good at picking this stuff up will sometimes win games they have no business winning.

281
by Pat (not verified) :: Thu, 11/16/2006 - 7:16pm

but it definitely affects games and cannot be quantified by any stats based system — that is the ability of some players and coaches to adjust based on scouting and reading opponents during a game.

Why wouldn't that show up in a stats based system? If a team plays better as a game goes on, it'll show up in DVOA. If they play better versus one team (or a few teams) than others, it'll show up in their variance.

So he or his team performs better against team X than we would be able to predict based on the results of all the other games.

That wouldn't show up in a predictive way. It would show up in a retrodictive way. You can't say, for instance, that the Colts are better at doing this, because if they were, they wouldn't be playing their opponents so close. You could have said, for instance, that the Falcons will decline over the season because other teams will figure out the weird offenses they were using (and you would've been right). But now that teams have, and have adjusted, the Falcons rightly fall in DVOA as well.

282
by chris clark (not verified) :: Thu, 11/16/2006 - 8:25pm

re 281:

I'm not sure if one (and only) team does this, it would show up in DVOA. Take for example a weird team that fell behind 7 points in the first half consistently and then consistently won by 3 points (with all the related stats moving the same way). Some effects would show up in DVOA, i.e. the teams 2nd half DVOA would be better than its first half. However, would the total DVOA for the team be good? Not necessarily. For example, if the team scored that way on 8 consecutive Sundays against this years teams of CHI, OAK, SD, TEN, PHI, HOU, NYG, and SF it would not have a top DVOA because the opponent adjustment for the bad teams would drag the DVOA down more than the DVOA adjustment for beating the top teams. and, although the team would be 8-0, and have beaten all the top DVOA teams, it would not be highly ranked. And, if the VAR isn't opponent adjusted, it wouldn't even have high variance. It would just be this weird GUT/SKATE team.

What's worse, if the team did so by consistently breaking very long runs for touchdowns, its DVOA could be quite low, because very long runs tend to be "flukey" and thus are selected against in the DVOA statistics. DVOA tends to reward teams that do things that win the "field position" battle. If some team isn't that good at field position, but is very good at consistently (but still only occassionally) springing the big play, that team will be under estimated by DVOA. Their actual W/L record will be better than what DVOA predicts.

I thought the article on rushing yards made that point very well. Most rushes are for very short yards and it is a few long rushes per game that bring the average length stats up. However, DVOA cuts off rushing years at a certain point, because after a certain length, length of run is not consistent (too many outside factors interfere with getting good consistent statistical measures).

Thus, one can see that it is possible to construct a hypothetical team that would "just win", but have a relatively poor DVOA. However, until we see such a team win the SB, I don't suggest that Aaron change DVOA to rate such a team better. After all, it would probably make predicting the results for other more "normal" teams worse.

And that's the point here, DVOA gives us some sense how teams that win games the "normal" way (by moving the ball up the field in more consistent chunks) are likely to fare. Outlier teams are less well predicted. The cause for them being outliers need not be unsustainable (for them). However, if other teams can't follow the same pattern to win, then the outlier teams will never likely score well in DVOA.

DVOA is good at rewarding teams that play well in the way(s) that other teams have played well in the past. In that way it is conservative. If you have a unique new way to win, you need to prove it to us first, before we will rate it well.

283
by doktarr (not verified) :: Thu, 11/16/2006 - 10:17pm

RE: 282,

That hypothetical team would not be very highly regarded by DVOA, because according to all the historical data that we have, the way they have won is flukey and not sustainable. According to the historical trends, teams that put together a bunch of wins that way are no more likely to win games going forward than a team that lost about half of those games, playing roughly the same way.

Now, if you are positing a hypothetical team that could, against all historical trends, actually sustain this, then sure, DVOA will miss that team. But until someone can show some statistical evidence that wins like those have significant predictive power, there's no need for a change.

284
by cowfez (not verified) :: Thu, 11/16/2006 - 10:24pm

Just wanted to mention that the NFL gave the Chargers an extra sack midweek.

285
by Eddo (not verified) :: Thu, 11/16/2006 - 10:31pm

I've only discovered the FO website recently, and this is my first post. I do have some natural skepticism of DVOA, but appreciate and respect it as the tool it is. I have one specific question, which I will get to in a minute.
It seems that most of the detractors of DVOA like to point out specific results. Unfortunately, because a 16-game season, a small sample size can often cause people to not see the true benefit of DVOA, even as a predictive tool. I had a college statistics professor who used this analogy: let's say a weatherman predicts that tomorrow there is a 90% chance of rain, and then it is sunny all day. This does not make his prediction "wrong." In fact, a prediction that involved any probability that is neither 0% nor 100% cannot be judged to be "wrong"--rather, you must figure out how many times it rains on all the days he ever predicts to have a 90% chance of rain. If that number is close to 90%, he is a good weatherman.
Now, on to my question: I was wondering what leads to the conclusion that fumble recoveries are 50-50. Why is it that when a team recovers a higher percentage of fumbles, it is due to luck, but when they complete a higher percentage of passes than the league average, it is not due to luck? I remember hearing how during Bears' practices, Lovie Smith has the team fight for every loose ball (both fumbles and incompletions) in order to increase their ability to score defensive touchdowns. Maybe there is no correlation between this technique and the actual recoveries of fumbles during games. If so, I will understand.
(Please note: I am in no way claiming that Aaron is wrong in discounting fumble recoveries; I just want to know the reasoning behind it. Thanks.)

286
by Pat (not verified) :: Thu, 11/16/2006 - 11:50pm

And, if the VAR isn’t opponent adjusted, it wouldn’t even have high variance. It would just be this weird GUT/SKATE team.

Variance is opponent adjusted. It's the variance in DVOA, not the variance in VOA.

More likely, it's VOA's variance would be less than its DVOA's variance - because it's playing consistently, regardless of opponent. As Aaron mentioned recently in a blog post, there aren't any teams like that. The closest is Chicago's defense.

287
by Andrew (A.B.) (not verified) :: Fri, 11/17/2006 - 12:02am

Re: 278

There's a difference between intangibles and luck (e.g. an opponent who makes a 62 yard FG)

And don't forget the variance of their opponents (eg Jax being great week 8 and awful week 10).

288
by Pat (not verified) :: Fri, 11/17/2006 - 12:07am

Now, on to my question: I was wondering what leads to the conclusion that fumble recoveries are 50-50.

See here. There's just absolutely no correlation between a past ability to recover fumbles and a future ability to recover them.

289
by BillWallace (not verified) :: Fri, 11/17/2006 - 12:28am

re: 285
The reason is because there is a correlation between passing at a high % today, and passing at a high % tomorrow.

Aaron's research has shown that there is not a correlation between recovering high % of fumbles now, and recovering a high % tomorrow.... no matter who the coach is. It is technically possible that one exists, but Aaron's research suggest that it doesn't.

Therefore if a team passes for a high %, you can say that is a good passing offense, and they are likely to be one tomorrow. But if a team recovers a high % of fumbles, all you can say is they were lucky so far, and we have no idea what will happen tomorrow.

No one addressed the fact that some of the criticism has claimed that DVOA is not predictive. Despite that fact that it's based on being highly predictive. It is certainly more predictive than won/loss record.

290
by Eddo (not verified) :: Fri, 11/17/2006 - 12:56am

Re: 288-289.
Thanks for pointing me towards the evidence I was looking for. I hadn't exactly been doubting that fumble recoveries were random; I was just looking for proof.
And Bill, your point about the criticism that DVOA is not predictive is kind of what I was getting at with my anecdote in post #285. It seems that DVOA's critics can cite specific examples (read: games) where DVOA has "failed," yet they don't quite understand how probability plays a role in predictive ability. DVOA never guaranteed anything--it simply has given a better indication of which teams are likely to play better based on trends determined by empirical evidence.

I think what people like Pacifist Viking are arguing is that you can't take DVOA as an absolute judge of success. There, I tend to agree; DVOA is exactly right is showing that the Eagles have played better ON THE WHOLE than their 5-4 record indicates, and that the Colts have played worse ON THE WHOLE that their 9-0 record indicates. However, the beauty of football (or any sport, for that matter), is that the players still have to execute at the right times (to use a cliche, find the ways to win) in order to be truly successful.
In the end, the outcome of a game comes down to a case of the whole being greater than the sum of its parts. DVOA does an accurate job of showing which teams do the best/worst during the entire course of a game or season, but not which teams end up making those "special" plays that go a long way in determining who wins and loses. In this way, DVOA does a good job of showing which teams are likely to keep up their success and which teams are not--which is something Aaron and the rest of FO should be proud of.

291
by navin (not verified) :: Fri, 11/17/2006 - 2:34am

That San Diego sack scoring change reminded me that there was a scoring change in the Philly-Washington game. The second McNabb touchdown is actually credited to McNabb now. Earlier it was a completion and fumble, now it's ruled that the first receiver didn't catch it. That should bump Philly's rating up some more, maybe they might even pass the Bears.

292
by bengt (not verified) :: Fri, 11/17/2006 - 10:25am

I think Oakland's defensive rank of 32 in the Fox rankings is wrong for several reasons:
- the 49ers are 32nd as well. You cannot have 32nd twice in a field of 32.
- the Raiders are tenth defensively in the rankings on this page, which they AFAIR were last week as well
- there is no tenth ranked defense in the Fox rankings, and only one ninth.

293
by Oh, Mathematics (not verified) :: Fri, 11/17/2006 - 10:49am

RE: 291

Though it will probably be added, DVOA should absolutely not count that play. While I agree that Reggie Brown never caught the ball, the fact that Buckhalter was right there at the right time to catch the ball in stride and run it in for a TD? That was just the Iggles getting some of that luck they've been lacking.

294
by Kuato (not verified) :: Fri, 11/17/2006 - 10:59am

As an Indy fan, I just hope the wrath of the FO message board does not call its full revenge upon my team. As far as I can tell, it is not Indy fans who are involved in the argument (as it was with the Atlanta fans last year).

All of my friends and I agree with the idea that Indy has not been the best team, but have been fortunate enough to end up with the best record. Is it still possible for them to become be best team, sure, but right now I'm worried as hell about the Dal game this weekend. I hope there is enough tape on Romo now for the Colts to throw a wrench into his fast start.

295
by Wanker79 (not verified) :: Fri, 11/17/2006 - 11:14am

Re: 293

I absolutely agree that it was luck (although I thought, without seeing the replay, that it kinda looked like an attempted leteral). And having Philly bumped up to #1 in the Fox ratings would probably bring on a wave of idiots the likes of which we haven't seen since the Atlanta FOMBC. But if every single data point is based on the play-by-play data, that has to be counted as a completion for a TD.

296
by Andrew (not verified) :: Fri, 11/17/2006 - 12:24pm

Oh, Mathematics:

Though it will probably be added, DVOA should absolutely not count that play. While I agree that Reggie Brown never caught the ball, the fact that Buckhalter was right there at the right time to catch the ball in stride and run it in for a TD? That was just the Iggles getting some of that luck they’ve been lacking.

While it was lucky that Buckhalter was able to make the catch, it was not luck that he was in position to do so. The Eagles offensive attack is in part predicated on unblocked offensive players running to the ball. Buckhalter was racing downfield to make a block, and by doing this put himself in position to make a great play. Just like the Freddie Mitchell recovery out of the air of an LJ Smith pass that karomed into the endzone from a couple of yards out in the NFC Divisional game in 1/2005.

The Eagles offense deserves some credit for a scheme which permits these type of plays, even if the plays themselves are lucky.

297
by Oh, Mathematics (not verified) :: Fri, 11/17/2006 - 12:31pm

Give me a break. That ball could have bounced in any direction, and it happened to bounce into Buckhalter's hands. 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. To say that the Eagles offense deserves credit for it seems ludicrous to me. I am an Eagles fan, for the record.

I do, however, realize that the play MUST be counted in DVOA. I just hope that with the game charting project, freak occurances like this one can be removed from the metric.

298
by stan (not verified) :: Fri, 11/17/2006 - 12:32pm

285,

Eddo -- there actually is a proper technique for recovering a fumble and good coaches practice it. All too often we see players try to recover the ball by flopping their bodies on top of it. And usually the ball squirts out.

It is very similar to the problems receivers have when they try to catch a pass by trapping the ball against their chest or stomach rather than catching the ball with their fingers. Fumbles need to be grasped by the fingers and drawn into the body which is sideways on the ground in the fetal position (this avoids having the ball squirt free from below a player who has several others jump on top of him. The ball is then protected by hands grasping each end of the ball with arms crossed and the ball centered to the stomach.

The random bounces of the ball are such that they overcome the stats. However, in practice you can toss a number of spinning or bouncing balls on the ground for players to recover and see that they get better and better at recovering them as they practice proper technique.

The only time it really makes a difference in a game is when a player is the first one to the spot and clearly has the make or break opportunity to recover the ball before others get there to poke, tear, or smash the ball free.

299
by Oh, Mathematics (not verified) :: Fri, 11/17/2006 - 12:32pm

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.

300
by Irishfan (not verified) :: Fri, 11/17/2006 - 1:00pm

Has anyone ever worked correlation coefficients for team wins in a season with DVOA for a season. I know on a season by season basis the correlation with DVOA in that season and each team is .85 or there abouts. I just wonder is there some teams who have a lower correlation and it is getting hidden by the overall trend – ie does DVOA consistently underestimate 49ers wins or overestimate eagles wins, for example? I don’t know how to calculate correlation coefficients so cant do this exercise – I also wonder will the teams with the highest average variance have the lowest correlation – anybody want a nice Friday afternoon project?