Writers of Pro Football Prospectus 2008

Most Recent FO Features

LuckAnd12.jpg

» Scramble Over/Unders: The Souths

Mike and Tom's tour of the league gets musical! Listen to the strangely up-tempo Blinded by the Luck, the haunting melancholy of The Schiano Man, and the endless march of the Saints.

16 Dec 2008

Week 15 DVOA Ratings

by Aaron Schatz

After a week of close games and upsets, Philadelphia is back on top of the Football Outsiders DVOA ratings. Conventional wisdom is coming around to agree with DVOA's constant season-long support for the Eagles, but they still may not even make the postseason. It's also important not to fetishize the idea of which team is "number one." After Philadelphia's win over New York knocked the Giants down from first to fourth, we're left with four teams so close at the top that we practically have a four-way tie. The two hottest teams in the NFL rank fifth (Pittsburgh) and sixth (Carolina). When the playoffs finally get here, it really is anybody's ballgame. The 2008 standings are a lot similar to the 2006 standings. That year also had no dominant team, with seven teams at 20% DVOA or higher through 15 weeks.

Going back to 1995, there are only two seasons where the best team as of Week 15 had a lower DVOA than Philadelphia does this year: 1997 (San Francisco at 30.4%) and 2003 (Kansas City at 29.8%).

While the top ten teams in regular DVOA are the same as the top ten teams in weighted DVOA, we can see some trends. Carolina and Indianapolis both have clearly improved over the past few weeks. Lower down, Houston and especially Kansas City are getting stronger and offering fans hope for 2009. On the other hand, Green Bay, Chicago, Washington, Arizona, and Denver have all faded a bit in the second half of the season. Buffalo has really dropped off. The Jets are sort of interesting -- their weighted DVOA is higher than their total DVOA because of bad losses in Weeks 2-3. Our short-term memory thinks of them as fading because of their losses to Denver and San Francisco, but weighted DVOA gives primary weight to the last eight weeks and that period also includes their strongest games, particularly the win over Tennessee and the 47-3 blowout of the pathetic Rams.

* * * * *

Clearly the defining characteristic of this season is the imbalance between the divisions. The NFC West and AFC West are abysmal, and the AFC East has three mediocre teams that have put together 9-5 records in part because they've picked on the teams from those two western divisions. The Patriots may become only the second team in history to go 11-5 and not make the playoffs, and the other one (the 1985 Broncos) played when there were just two wild cards.

I decided to try a little experiment: What would this season look like if the eight NFL divisions were balanced? We put together a "pretend" NFL by taking the 16 teams in each conference and matching them up based on current DVOA rating. For example, one division consists of the AFC teams currently ranked first (Baltimore), fifth (Miami), 12th (Cleveland), and 16th (Oakland). I assigned each division to play another division in their own conference and a division in the other conference. The final two games were based on last year's records. For example, Baltimore would have ranked second if these four teams had been a real division last year, so they play against AFC Division 4 (BUF, HOU, IND, NE), NFC Division 2 (DAL, NO, NYG, SEA), and the two teams that would have finished second in AFC Divisions 2 and 3, Jacksonville and Denver.

There are two versions of the simulation. The first simulation includes an adjustment for "luck" based on how each team's actual win total differs from the Pythagorean projection. If the divisions were balanced, that doesn't suddenly mean that Philadelphia wouldn't have spent the first half of the year blowing close games with bad short-yardage runs, it doesn't mean that Denver wouldn't have gotten away with some close wins, and so on. The second simulation is only based on DVOA, with no adjustment.

With balanced divisions, the playoffs would include some of those good teams that are going to be left out in the real world. In both simulations, the AFC playoff teams are the same: Baltimore, Tennessee, Pittsburgh, and Indianapolis, with Miami and the New York Jets as the wild cards. The NFC divisions go to the New York Giants, Carolina, Philadelphia, and Tampa Bay, with Atlanta and either Minnesota or New Orleans as wild cards. In each simulation, Arizona, Denver, and New England are stuck at home the majority of the time. You can also see where the ability to win close games (or, alternately, the serendipity to win close games) would affect the postseason races. In the "luck adjustment" version, Baltimore and Miami are basically tied atop AFC Division 1; in the other simulation, Miami is clearly in second place. The Dolphins also show how last year's finish can affect schedule, even though it only changes two games. In our simulated league, Miami plays Cincinnati and Kansas City as its two final games. Despite balanced divisions, they still have a very easy schedule.

* * * * *

For Premium subscribers, we have a new -- and very commonly requested -- view now available in the DVOA database, which will show you how DVOA would have looked through any week of any season back to 1995. Each week's data is based on opponent adjustments as they looked at the time, so Week 6 would have opponent adjustments at 60 percent strength and based only on the first six weeks, not the entire season. Now you can follow along with the "Worst DVOA Ever" watch or, if you prefer, figure out which teams had the best offense through Week 15, or the best defense through Week 9, or anything else you are interested in.

UPDATE 11:30pm EST: All stats pages are now updated. I have no idea why the FUTURE SCHEDULE columns came out weird, so I just removed them. They don't really mean much anyway with just two games left.

* * * * *

These are the Football Outsiders team efficiency ratings through 15 weeks of 2008, 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 to consider all fumbles, kept or lost, as equal value. These ratings also include opponent adjustments. SPECIAL TEAMS DVOA is adjusted for type of stadium (warm, cold, dome, Denver) and week of season. WEIGHTED DVOA is adjusted so that earlier games in the season become gradually less important. It better reflects how the team is playing right now.

As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.

To save people some time, please use the following format 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
WEIGHTED
DVOA
RANK
W-L OFFENSE
DVOA
OFF.
RANK
DEFENSE
DVOA
DEF.
RANK
S.T.
DVOA
S.T.
RANK
1 PHI 31.5% 3 30.3% 1 8-5-1 10.0% 13 -20.0% 4 1.5% 14
2 BAL 30.3% 2 30.1% 2 9-5 3.4% 19 -27.0% 1 -0.1% 20
3 TEN 30.0% 4 29.6% 3 12-2 8.8% 14 -19.4% 5 1.7% 13
4 NYG 29.4% 1 29.4% 4 11-3 19.8% 3 -6.3% 8 3.3% 6
5 PIT 24.6% 5 24.1% 6 11-3 -0.6% 21 -26.1% 2 -0.9% 23
6 CAR 20.2% 6 24.9% 5 11-3 15.4% 6 -3.0% 9 1.8% 12
7 TB 16.7% 7 15.1% 8 9-5 1.2% 20 -15.6% 6 -0.1% 19
8 IND 11.9% 10 15.6% 7 10-4 18.0% 4 4.1% 14 -1.9% 24
9 MIN 10.9% 16 12.5% 10 9-5 -2.4% 22 -20.7% 3 -7.5% 32
10 NO 10.5% 8 13.8% 9 7-7 19.8% 2 8.8% 20 -0.5% 21
11 GB 10.1% 9 6.7% 13 5-9 10.6% 12 1.9% 12 1.4% 16
12 ATL 9.3% 11 8.3% 11 9-5 17.7% 5 10.7% 22 2.3% 9
13 CHI 8.7% 15 3.8% 16 8-6 -2.5% 23 -8.4% 7 2.9% 7
14 DAL 5.5% 19 3.9% 15 9-5 8.7% 15 -0.9% 10 -4.1% 28
15 MIA 5.0% 14 4.4% 14 9-5 14.7% 8 3.4% 13 -6.3% 31
16 WAS 4.7% 12 -1.4% 20 7-7 8.7% 16 0.9% 11 -3.1% 26
TEAM TOTAL
DVOA
LAST
WEEK
WEIGHTED
DVOA
RANK
W-L OFFENSE
DVOA
OFF.
RANK
DEFENSE
DVOA
DEF.
RANK
S.T.
DVOA
S.T.
RANK
17 NYJ 4.5% 18 7.4% 12 9-5 4.9% 18 4.3% 15 3.9% 5
18 ARI 4.0% 13 0.1% 18 8-6 15.2% 7 7.9% 17 -3.3% 27
19 SD 1.5% 17 -3.0% 21 6-8 12.6% 9 12.0% 24 0.9% 17
20 NE -2.6% 21 -1.0% 19 9-5 11.3% 10 16.6% 27 2.7% 8
21 HOU -3.7% 22 2.7% 17 7-7 10.8% 11 16.4% 26 1.8% 11
22 JAC -4.2% 23 -4.9% 22 5-9 6.9% 17 10.6% 21 -0.5% 22
23 DEN -5.9% 20 -9.4% 23 8-6 19.9% 1 21.1% 30 -4.7% 29
24 CLE -13.1% 24 -10.1% 24 4-10 -10.6% 27 6.6% 16 4.2% 4
25 BUF -15.2% 25 -22.6% 27 6-8 -7.6% 25 13.8% 25 6.2% 1
26 SF -19.0% 26 -23.0% 28 5-9 -15.2% 28 8.4% 18 4.6% 2
27 SEA -25.0% 27 -21.4% 26 3-11 -10.3% 26 17.0% 29 2.3% 10
28 CIN -28.0% 30 -26.7% 29 2-11-1 -16.8% 30 8.7% 19 -2.4% 25
29 KC -29.0% 29 -17.9% 25 2-12 -6.9% 24 16.6% 28 -5.5% 30
30 OAK -29.9% 28 -32.5% 30 3-11 -23.1% 31 11.2% 23 4.4% 3
31 DET -44.0% 31 -39.4% 31 0-14 -16.2% 29 29.3% 32 1.4% 15
32 STL -49.0% 32 -45.1% 32 2-12 -23.8% 32 25.3% 31 0.1% 18

  • NON-ADJUSTED TOTAL DVOA gives performance without adjustments for schedule strength, fumble recovery luck, and weather/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.
  • VARIANCE measures the statistical variance of the team's weekly DVOA performance. Teams are ranked from most consistent (#1, lowest variance) to least consistent (#32, highest variance).


TEAM TOTAL
DVOA
W-L NON-ADJ
TOT VOA
ESTIM.
WINS
RANK PAST
SCHED
RANK VAR RANK
1 PHI 31.5% 8-5-1 29.9% 9.9 3 0.8% 15 16.1% 14
2 BAL 30.3% 9-5 29.4% 10.2 2 3.1% 7 17.1% 19
3 TEN 30.0% 12-2 32.1% 10.3 1 -3.6% 24 9.6% 5
4 NYG 29.4% 11-3 29.6% 9.9 4 0.6% 19 22.7% 28
5 PIT 24.6% 11-3 21.3% 9.7 5 4.4% 5 6.3% 2
6 CAR 20.2% 11-3 20.4% 9.0 6 -0.7% 22 16.5% 16
7 TB 16.7% 9-5 15.7% 8.4 9 0.8% 17 15.6% 12
8 IND 11.9% 10-4 14.5% 8.8 7 1.1% 13 18.2% 24
9 MIN 10.9% 9-5 8.1% 7.8 12 2.3% 9 17.1% 18
10 NO 10.5% 7-7 9.1% 8.2 10 1.6% 10 13.0% 8
11 GB 10.1% 5-9 4.4% 7.8 13 3.8% 6 16.8% 17
12 ATL 9.3% 9-5 6.4% 7.9 11 2.5% 8 20.5% 27
13 CHI 8.7% 8-6 5.3% 8.5 8 1.4% 12 10.5% 6
14 DAL 5.5% 9-5 7.0% 7.3 17 1.4% 11 32.3% 31
15 MIA 5.0% 9-5 20.5% 7.5 16 -8.5% 30 17.4% 20
16 WAS 4.7% 7-7 1.7% 7.3 18 0.8% 16 6.3% 1
TEAM TOTAL
DVOA
W-L NON-ADJ
TOT VOA
ESTIM.
WINS
RANK PAST
SCHED
RANK VAR RANK
17 NYJ 4.5% 9-5 14.1% 7.7 14 -10.4% 32 17.9% 23
18 ARI 4.0% 8-6 7.4% 7.2 19 -4.3% 26 19.8% 26
19 SD 1.5% 6-8 5.0% 7.6 15 -3.7% 25 15.9% 13
20 NE -2.6% 9-5 3.6% 6.8 21 -7.7% 29 19.5% 25
21 HOU -3.7% 7-7 -9.4% 5.9 24 4.7% 4 11.2% 7
22 JAC -4.2% 5-9 -4.6% 7.0 20 0.8% 14 8.2% 4
23 DEN -5.9% 8-6 -4.6% 6.2 22 -4.7% 28 23.2% 29
24 CLE -13.1% 4-10 -21.6% 6.0 23 9.4% 2 14.2% 10
25 BUF -15.2% 6-8 -6.3% 5.3 25 -9.8% 31 13.2% 9
26 SF -19.0% 5-9 -14.7% 5.2 26 -4.4% 27 6.6% 3
27 SEA -25.0% 3-11 -21.6% 4.5 27 -3.1% 23 17.4% 21
28 CIN -28.0% 2-11-1 -39.7% 3.9 28 13.7% 1 14.4% 11
29 KC -29.0% 2-12 -26.8% 3.8 29 0.3% 20 16.1% 15
30 OAK -29.9% 3-11 -29.5% 3.7 30 -0.3% 21 33.0% 32
31 DET -44.0% 0-14 -47.8% 2.6 31 7.7% 3 17.5% 22
32 STL -49.0% 2-12 -49.0% 1.8 32 0.7% 18 23.4% 30

Worst DVOA Ever Watch

For all the talk about how there are so many really awful teams this year, the fact remains that there are only two historically bad teams: St. Louis and Detroit. The gap between Oakland (30th) and St. Louis (32nd) is equivalent to the gap between, for example, Pittsburgh (5th) and the Jets (17th). Detroit may be the team that finishes 0-16, but according to DVOA, the Rams are worse -- now the second-worst team in DVOA history. The Lions' near-win over the Colts vaulted them up a few spots on the list of the worst teams ever, although their defensive DVOA actually got worse in this game and still ranks as the worst of all time.

In retrospect, the Rams' 34-14 win over Dallas in Week 7 is just incredible, Brad Johnson or no Brad Johnson. The Rams' DVOA in that game is 64%. Their 19-17 win over Washington was a bit lucky, and the Dallas win is the only time all year St. Louis has put up a DVOA over -10%. The Rams have six games with DVOA below -70% and nine with DVOA below -40%. By comparison, the Lions have two positive games (both losses to Minnesota, 12-10 and 20-16) and two other games that are negative in single digits (27-23 loss to Chicago in Week 9 and this week's loss to Indianapolis). The Lions have four games below -70% and seven below -40%.


WORST TOTAL DVOA
AFTER WEEK 15
  WORST DEFENSIVE DVOA
AFTER WEEK 15
2005 SF -61.4%   2008 DET 29.3%
2008 STL -49.0%   2000 ARI 28.5%
1999 CLE -48.4%   2008 STL 25.3%
2000 ARI -47.4%   2000 MIN 24.4%
2000 CLE -46.0%   2002 CIN 23.9%
2004 SF -45.4%   2004 MIN 22.9%
2000 CIN -45.0%   2004 STL 22.7%
2008 DET -44.0%   2001 MIN 22.4%
2003 ARI -43.0%   1998 CIN 22.2%
2002 ARI -41.6%   2003 ARI 22.1%

Posted by: Aaron Schatz on 16 Dec 2008

137 comments, Last at 19 Dec 2008, 11:25pm by Pat (filler)

Comments

1
by JMM :: Tue, 12/16/2008 - 4:57pm

Pitt is clearly ranked too low because Big Ben knows how to close. Wins, losses and head to head battles is better than this. All da wey 2 da supperbowl whoohoo!

39
by Anonymous Too! (not verified) :: Tue, 12/16/2008 - 6:29pm

Wins are clearly overrated in Football. Ask any Patriot fan whether they'd rather have 19-0 or the all-time DVOA crown, and they'll probably need you to remind them what 19-0 even refers to. Aaron's warning about not fetishizing DVOA is too little, too late I fear, in our DVOA-crazed world.

80
by Wanker79 :: Wed, 12/17/2008 - 10:27am

Actually, Aaron was saying that the 1-32 ranking position (not the DVOA values) shouldn't be of much concern because the top 4 spots are basically a dead heat. And if you've ever payed attention to anything Aaron's ever written, you probably should remember numerous occasions when he warns people that DVOA is just one tool we can use and not the definitive truth. But none of that would fit into the little narrative that seems to run through basically every post I've seen from you. So I guess it's understandable if you've blocked those details out.

89
by JMM :: Wed, 12/17/2008 - 11:50am

"Wins are clearly overrated in Football."

Not sure I agree with this. Many in the NFL think the point of Football is winning.

If you meant its use as a predictor, well yeah, that is why I used the classic Lionsfan format. I do believe 4th Q play might be undervalued in DVOA land.

Any of the top 6 can be considered legit contenders for a SB appearance and win.

117
by The Ninjalectual :: Wed, 12/17/2008 - 11:47pm

"Footballs are clearly overrated in Football."

Is poster drunk?

122
by Heathnyy5 :: Thu, 12/18/2008 - 1:54am

Um, the rankings have nothing to do with "closing games." Ben Roethlisberger, by the way, may have won a ring at Super Bowl XL, but he had one of the worst performances of all-time for a winning QB in the game. The MVP went to Hines Ward, and some still think if not for terrible officiating, the Seahawks would have been champs. I don't know if I go that far, but I do say that calling Big Ben a closer/winner is not entirely accurate. Of all the young QBs that are now doing so well in the league (Roethlisberger, Eli Manning, Philip Rivers, Tony Romo, and Jay Cutler; with respect to Aaron Rodgers, Matt Ryan, Matt Cassel, and Joe Flacco, who have not been around for long enough, to Carson Palmer for injury, and to Drew Brees due to the fact that while he IS still under 30-years-old, he's not necessarily a member of this group) I would think Manning is the one to take when your team is down late in the game, and you need a drive to win. Manning's performance in Super Bowl XLII is not my only reason for thinking so, he has had a knack for the late drive to win or tie the game his whole NFL career thus far. For instance, he got his first win in his rookie season that way (a late drive to beat Dallas with a TD in the closing seconds), and had another big win that way in his first full season (against the Denver Broncos, who that year went to the AFC Title game).
In comparison, Roethlisberger has come up short in some of his biggest games, including the Super Bowl, and his first two playoff games after the 2004 season, when he was bad vs. the Jets (and lucky the Jet kicker choked the game away), and then against the eventual champion Patriots. Romo's hiccups in big games are well documented, and even though I think people are too harsh against him when they bring it all up, he has developed somewhat of a tendency to make mistakes in the least opportune moments. Rivers (though he was indeed hobbled and playing without LaDanian Tomlinson) was unable to lead the Chargers to a single TD drive in the AFC Title game last year, when the Chargers defense was holding the Patriot offense in check, Rivers was unable to get the job done. The other QBs I named haven't yet been tested in the biggest games yet (Carson Palmer was injured so quickly into his first and only playoff game, that it's impossible to make a decision on him in such situations); they haven't yet had to come through in the clutch in a moment where failing to do so would cost their team the playoffs or the championship.
Ben Roethlisberger, in Super Bowl XL, was lucky enough that his team did not need a contribution from him to win the game, as he was not up to the task in what was absolutely (up to this point) the biggest game of his career. The Giants, however, could not have won Super Bowl XLII without Manning leading the Giants on two TD drives, the last one a drive that began with less than three minutes on the clock, and resulted in the game-winning TD.
While it is impossible to really measure on QB against another in any easy fashion, as they have different systems, different teammates, and everything else (for instance, if Joe Montana was sent to Miami in a trade for Dan Marino, it is impossible to say how Montana would perform as a Dolphin, or Marino as a 49er), the preference of which QB one would want in the closing minutes of a game with everything on the line is mainly done by feel and past performance. Right now Eli Manning is not only head and shoulders ahead of Ben Rothlisberger in this measurement, but the only current QB at or above Manning's level is the one he beat to win XLII, Tom Brady. However, since Eli Manning is the only one healthy right now, it's hard for me to see how anyone could choose Roethlisberger or anyone else over Eli Manning if they had the choice between them in such a situation.

Now he belongs to the ages.

124
by JMM :: Thu, 12/18/2008 - 12:16pm

"Um, the rankings have nothing to do with "closing games."'

I know, that is why I used it. See the suggested complaint format in the article. It specifically requires a non-related subjective rational.

As to the rest of your post, try some white space and a couple of carriage returns. It looks like you want to make a case for Eli vs Ben. Fine with me. As a Steeler fan, I'm OK with Ben and best of luck with Eli.

2
by Bobman :: Tue, 12/16/2008 - 4:58pm

"It's also important not to fetishize the idea of which team is "number one."

Damn.

Well, there goes that picture of McNabb in thigh-high latex boots....

8
by Mike B. In Va :: Tue, 12/16/2008 - 5:19pm

Most.
Disturbing.
Post.
EVAR!

25
by Danish Denver-Fan :: Tue, 12/16/2008 - 5:50pm

EVAR...

Eaglesadjusted-Value-Above-Replacement. Sombody should look into this stat...

3
by Yuri (not verified) :: Tue, 12/16/2008 - 5:00pm

Future schedule has some positively wild numbers. DAL @62%? I think this bug has happened before as well.

6
by tally :: Tue, 12/16/2008 - 5:14pm

Might have forgotten to divide the total opponent DVOA by the 2 weeks remaining in the season when averaging them. Dallas does have to play #2 Baltimore and #1 Philly.

10
by Staubach12 :: Tue, 12/16/2008 - 5:20pm

No, that seems right. They close the season against the #1 and #2 teams (combined DVOA is about 62%).

23
by FireOmarTomlin :: Tue, 12/16/2008 - 5:41pm

"FUTURE SCHEDULE lists average DVOA of opponents still left to play this season, "

4
by zlionsfan :: Tue, 12/16/2008 - 5:01pm

Joe Barry, Rod Marinelli, and Matt Millen. Ladies and gentlemen, the architects of one of the worst defenses in history!

At least the constant success by the opposing offense means the clock runs more. Makes the games go faster. Plus no one really feels the need to confuse the Lions, so you don't see a lot of shifting or motion or weird formations or anything. Easier to chart.

Sam Bradford?

18
by BucNasty :: Tue, 12/16/2008 - 5:36pm

A lot of repetition in those worst ever lists. Good thing the data only goes back to 95 :)

5
by Ali Haji-Sheikh (not verified) :: Tue, 12/16/2008 - 5:14pm

If there are 20 teams better than the Texans in the NFL I'll Fu'umatu-Ma'afala myself

30
by DrewTS (not verified) :: Tue, 12/16/2008 - 6:03pm

They're much better now than they were when they were 0-4. But those games still count. For what it's worth Weighted DVOA puts them slightly higher.

7
by Bobman :: Tue, 12/16/2008 - 5:18pm

Just an interesting look at six playoff teams above (I hope the formatting works):

... Off rank ... Def Rank ... ST Rank

Team A ... 13 ... 4 ... 14

Team B ... 14 ... 5 ... 13

doesn't get much closer than that, does it?

Team C ... 3 ... 8 ... 6 (mondo consistent)

Team D ... 6 ... 9 ... 12 (mondo consistent)

Team E ... 21 ... 2 ... 23

Team F ... 4 ... 14 ... 24

A/B are Phil/Tenn. Looks like a mirror-image SB?? But Tenn is stumbling and may have injury issues. Phil, as noted above, may not even make the post-season. The Giants (C) are incredibly consistent across their units, but seem to be fading. Carolina (D) is strong in all phases and is coming on strong. Pitt (E) has what looks like a very suspect O (I didn't know that). Can the D compensate? It has, so far. Indy's (team F) O is somehow back in the top ranks, and their D is better than most observers would guess. They are hot at the right time and their health is returning, though injuries are still a problem. ST is poor by comparison with the other teams above, but stellar for Indy.

Balt has that amazing D and are hitting on all cylinders, but their O and ST are not scaring anyone.

I would not be too surprised to see Indy, Pitt, and Car in the championship games. Giants, too, if they regain form?

28
by FireOmarTomlin :: Tue, 12/16/2008 - 5:55pm

interestingly... could it be the Steelers mediocre Special Teams will prevent them from achieving the "Great D, horrible O" SB win of the 2000 Ravens and 2002 Bucs? The O's are eerily similarly bad, and the D is about as good overall, though more balanced than either of those great D units. (I fully expect the Week 17 Clowns game to put the Steeler D to -29 or -30% total DVOA)

2000 BALT
DEF 1 -30.0% 2 -30.7% 2 -21.7% 6 -41.7% 1
OFF 21 -7.1% 27 -11.9% 23 -22.3% 25 8.3% 7
ST 3 7.2% 11 9.9% 2 15.4 4.1 0.5 3.9 18.4
TOT 2 30.2%

2002 TB
DEF 1 -33.6% 2 -29.3% 1 -55.8% 1 -7.6% 7
OFF 21 -2.5% 12 -2.0% 19 6.2% 14 -13.6% 27
ST 8 2.9% 27 1.6% 11 4.2 9.3 -0.6 1.4 2.7
TOT 1 34.0%

2008 PITT
DEF 2 (for now ;) ) -26.1% 2 -26.8% 1 -29.5% 1 -21.4% 2
OFF 21 -0.6% 22 -0.9% 22 -1.1% 21 -0.1% 18
ST 23 -0.9% 19 -1.8% 24 3.3 6.2 -9.6 1.9 -6.2 -1.6%
TOT 5 24.6%

38
by Mike Kurtz :: Tue, 12/16/2008 - 6:25pm

I think a key difference in that Pittsburgh's offense has the potential to be good. It's all about how well the offensive line holds up, and thus far it appears to be completely random. If the line holds up, however, PIT has a pretty good offense, so I think that separates it from TB02 and BAL00, which were miserable because they had no weapons.

That said, that's a really, really massive "if."

41
by Anonymous Too! (not verified) :: Tue, 12/16/2008 - 6:39pm

Agreed, but it's not really the line that has been bad of late. They are pretty solidly mediocre but not that bad. The offense as a whole has been inconsistent because of occasional breakdowns on the O-Line, drops, bad throws/decisions, inconsistent and injured running backs, etc.

Calling the Steelers special teams mediocre is also kind of misleading. They have a good kicker, surprisingly wonderful coverage units (until this past week), a horrible punter, and dreadful return teams. It's a weird mish-mash.

53
by Mystyc :: Tue, 12/16/2008 - 7:29pm

The amusing thing is, that weird mish-mash almost perfectly matches the rest of the team: the coverage units are almost as good as the defense, so the other team has to fight for pretty much every inch on every play, but the return units are as shaky as the offense, leaving the Steelers with no way to reliably move the ball forward. Even the punter can't get the ball down the field.

29
by panthersnbraves :: Tue, 12/16/2008 - 5:57pm

Thanks for taking the time to do the format. It made me take off my homer-vision glasses (if only for a moment).

9
by thestar5 :: Tue, 12/16/2008 - 5:20pm

"The Patriots may become only the second team in history to go 11-5 and not make the playoffs, and the other one (the 1985 Broncos) played when there were just two wild cards."

Aren't there still only two wildcards???

Also, nice to see Dallas beat down the Giants. After all the flak I got for saying the first DAL-NYG game would have been close had Romo played and these were two evenly matched teams, it was nice to see that validated on the field. Maybe I wasn't raiderjoe without the mispellings, tundrapat?

19
by tuluse :: Tue, 12/16/2008 - 5:37pm

Yeah, there are only two wildcards now, but there is an extra division winner compared to 1985. For a while though there were 3 wildcards teams with 3 divisions.

20
by Fake Zero (not verified) :: Tue, 12/16/2008 - 5:38pm

FO worded this poorly, but they are correct (in spirit).

In 1985 there were 3 divisions + 2 wildcards = 5 playoff teams. Today, with an extra division, there are 6 playoff teams.

The switch to 6 playoff teams actually occurred before the 4 division format when the NFL went to 3 divisions, 3 wildcards in 1990.

87
by dryheat (not verified) :: Wed, 12/17/2008 - 11:45am

Wait...I don't have a dog in the fight, but weren't the Giants missing their top RB and top WR -- elite players at their positions? Why then is the second game more of an indicator to you than the first game? The Giants don't get the benefit of injuries to key players? Yes, I know they almost certainly won't be getting Plax back this year, but he was a significant loss who wasn't adequately replaced.

11
by TerryTate :: Tue, 12/16/2008 - 5:25pm

The Packers, 1-6 in their last seven games, have fallen off a "little bit." I'm glad it could apparently be worse.

62
by DisplacedPackerFan :: Tue, 12/16/2008 - 8:56pm

The Packers are ridiculous. They've still outscored their opponents by 32 points. The Loins and Bears could both beat them by 16 points and they would be even on points scored and points allowed and end up 5-11. I realize that points scored and point differential are only tangentially FO stats but I've got to wonder if anyone knows the most positive point differential by a losing team. The "average" game for the Packers right now "should" be 26.5 to 24.2. Small margin for sure, but they look like a winner based on that.

Of course when you look at how craptacular they have been in short yardage and the defense deciding to not play the last 2-5 minutes of a game, things change a bit.

I'm not trying to say they are anything but an average team at best, but even some of the traditional indicators point to them being "better than their record".

7-7 Washington has been outscored by 35 points. The 5-9 49ers (same record as GB) have been outscored by 46 points. The 8-6 Denver Broncos have been outscored by 40 points, they are the only team this year with a winning record that has given up more points than they've scored.

The quick explanation to this, since the Packers have only had 2 losses by more than 10 points (11 point lose to the Cowboys and the 22 point beat down by New Orleans is that they aren't 'clutch'. They don't 'rise to the challenge' in close games.

The interesting discussion comes in the why. Since they've had several late losses where they had the lead with 5 minutes or less to go, you want to blame the defense. I still point to short yardage situations for the offense though. The offense stalls out way too often in 2nd or 3rd and less than 3. The defense often responds but there are too many times where 3 or 4 points are not put on the board because of a lack of a conversion. Ryan Grant can't find a hole on the rare occasions when the offensive line makes one. They are one dimensional offense which is part of the reason why they have the short yardage issues.

Where are the big issues? The lines. The offensive line does a passable job in pass protection (though some games they've been horrendous with that too), even with defenses trying harder to get pressure since they aren't worried about the run. But the run blocking just isn't there. The defensive line can't generate pressure or stop the run. The loss of Jenkins early in the year was HUGE to this team. Of course now you have morale issues I'm sure. Guys aren't playing 100% anymore. It's a shame too because I kinda wanted to see an 0-16 team but the Lions are going to be the Pack in that game now.

12
by TerryTate :: Tue, 12/16/2008 - 5:28pm

Being fairly new here, I have a question. How does one gain the ability to look at premium content?

49
by justme_cd :: Tue, 12/16/2008 - 6:57pm

the online store

13
by thestar5 :: Tue, 12/16/2008 - 5:29pm

Also, I disagree with your methodology of using DVOA to sort out who would be the wildcards in the balanced divisions. For example, Dallas and New Orleans have had virtually identical schedule strengths, yet Dallas is 9-5 but NO is only 7-7. So I don't see why they would be a wildcard over Dallas. I think Dallas is ranked so low because of that one horrific Rams game, and that awful game brings them down way farther than it should. It would be like if whenever you lost by 20 or more points, it counted for 2 losses instead of just 1. I don't think one game should carry so much weight, especially when you only have a sample size of 16 games.

71
by RickD :: Wed, 12/17/2008 - 12:56am

Yes, that's how DVOA works. It doesn't believe in ignoring outliers.

A lot of teams get overrated and a lot get underrated. And then when one that seems to be overrated wins a couple games, the DVOA guys say "see, you made fun of Philly but they've won two games in a row!", ignoring that we also made fun of how high Arizona was, how high the Jets were, etc.

DVOA is an interesting statistic but it's hardly the end-all, be-all. I think its component parts (for offense, defense, and special teams) are far more valuable than the sum of the three. That Pats have a very good offense and a terrible defense. The Jets are mediocre at both. And that's why the Jets are thought to be a better team?

111
by Pat (filler) (not verified) :: Wed, 12/17/2008 - 7:41pm

ignoring that we also made fun of how high Arizona was, how high the Jets were, etc.

Why wouldn't they ignore it? The three situations are totally different.

They pointed out Philadelphia because the level at which they're playing doesn't match their record. They kept playing at that level, and their record is starting to match their level. If they finish 10-5-1, having them in the top 5 is not going to be that surprising.

Arizona's DVOA did match their record, roughly. They were 7-3, ranked right around where other 7-3 teams were. They did not continue to play at that level, and their record tanked.

I don't get the Jets comment, as the Jets DVOA has hovered around 5-10% the entire time. They've moved around in the ordinal rankings, but that's just due to other teams moving around. If you go back far enough, they've been moving down steadily, but that's just the opponent adjustments kicking in and steadily yanking the team down while they play an easy schedule.

The comment regarding the Eagles is only meant to point out that DVOA tends to be quicker at pointing out level of play than wins/losses. It doesn't mean that DVOA can predict how a team will continue to play.

88
by dryheat (not verified) :: Wed, 12/17/2008 - 11:49am

Well, then you would have to throw out each teams' worst performance, and I'm not sure what added benefit that would have. And probably each teams' best performance too, and then we're talking about losing 1/8 of the small data set.

14
by Anonymouse (not verified) :: Tue, 12/16/2008 - 5:29pm

what about best DVOA never to make the playoffs? looks like Bal might be a contender.

107
by spacab :: Wed, 12/17/2008 - 5:23pm

I'm assuming you mean that Philly is a contender...

15
by Key19 :: Tue, 12/16/2008 - 5:30pm

The ascension of Dallas to a somewhat realistic DVOA ranking has begun. I think they beat the Ravens next week and I think Week 17 against Philadelphia will be one for the ages. If Philly can win this week as well, I see no reason why DAL/PHI is not the Flex Game for Week 17. It would be a game for the ages. Should be a lot of fun and I can see it going either way if the Eagles are impressive again this week.

31
by Fourth (not verified) :: Tue, 12/16/2008 - 6:05pm

The Baltimore game should be a great one. Philly too. "One for the ages" is pretty rare so I dunno about that. Also, NBC has flexed 4 Fox games in a row now, so something tells me the week 17 flex will be Miami at NYJ assuming both win this week. It might not be one for the ages, but it will be for the AFC East and the 3 seed vs not making the playoffs.

42
by Mike Kurtz :: Tue, 12/16/2008 - 6:40pm

NBC also has to make the request 13 days before the game. There was an exception for this coming week, but I have to imagine that's the rule, or CBS and FOX would be up in arms.

59
by zdneal@yahoo.com :: Tue, 12/16/2008 - 8:11pm

Per MMQB this week the NFL decides the schedule for Week 17, not the networks.

64
by tuluse :: Tue, 12/16/2008 - 9:50pm

I thought that FOX and CBS were allowed to "protect" a certain number of games each week too.

82
by Rocco :: Wed, 12/17/2008 - 10:56am

They are, but not for the last week of the season. I think they can each protect up to 5 games for Weeks 10-16. I imagine CBS protected Steelers/Titans (since the rest of their lineup is pretty lousy this week), so Giants/Panthers was flexed instead.

16
by Todd (not verified) :: Tue, 12/16/2008 - 5:32pm

What happens if we remove the Cowboys games without Romo? I'm sure they wouldn't be that low. Such an analysis might be appropriate given that FO has already determined that the drop-off from BJ to Romo was the largest since compiling their statistics.

90
by dryheat (not verified) :: Wed, 12/17/2008 - 12:00pm

Probably the same thing that happens if you remove the Giants' games without Brandon Jacobs, the Bengals games without Carson Palmer, the Bills' games without Trent Edwards, the Eagles' games without Bryan Westbrook, the 49ers games without Shaun Hill, and the Bucs' games without Jeff Garcia....they would rise in the rankings.

Injuries, suspensions, etc. happen in the NFL. It seems unreasonable to expect the FO team to re-create the entire season using each team's optimal lineup, or to only count games when each team had its full complement of players.

17
by Brian (not verified) :: Tue, 12/16/2008 - 5:35pm

I don't think Pythagorean estimations are valid for football. In baseball, scoring is independent, meaning that the winner's score does not correlate with the loser's score. In football however, scores for winning and losing teams correlate significantly (above r=0.4). So no matter how you adjust the Pythagorean exponent, the win estimations are going to be very biased.

24
by thestar5 :: Tue, 12/16/2008 - 5:45pm

Sorry, not that I think you're wrong at all, but I don't understand what you mean. How is it possible to not have the winner's score correlate with the loser's score since the winner's score must be higher?

26
by Staubach12 :: Tue, 12/16/2008 - 5:52pm

Because some games are high scoring, and the losing team goes pass happy to keep up--moreover, the winning team makes it easy for the other team to score by going into a prevent defense.. Some games are low scoring, and the winning team grinds it out and keeps the score down. There are also game conditions that have more bearing in football than baseball, since baseball is a summer game and they don't play baseball in the rain.

The scores in football would correlate less if winning teams didn't play prevent defenses or try to run out the clock in games. Something like that can't happen in baseball.

27
by thestar5 :: Tue, 12/16/2008 - 5:54pm

Okay, thanks for the clarification. It still seems like this would apply to baseball though, albeit to a lesser degree. A team way ahead would put in mop-up pitchers instead of their set-up men/closer, and maybe rest their best hitters too. Maybe its not significant but it seems like this still applies.

Also, sorry about the triple post, not sure how that happened.

125
by armchair journe... :: Thu, 12/18/2008 - 4:01pm

"something like that" absolutely happens in baseball. the MLB equivalent to a prevent defense is finishing 8-1 games with a mediocre middle reliever while you rest your set-up man and closer. just like in football, high-scoring baseball games generally mean soft defense/pitching.

_______________________________
armchair journeyman quarterback

72
by RickD :: Wed, 12/17/2008 - 1:04am

"being higher" is not a correlation.

Let's say my local track team sent its five best sprinters against the Jamaican national team. We would lose all five races, but there would not be any correlation involved. You wouldn't expect their fastest guy to be more likely to be matched up against our fastest guy for any reason.

Competition in baseball is similar, since baseball is basically two different pitcher vs. batter competitions. One could think that sometimes, a hitting team plays more aggressively when way ahead or way behind, but I'm not surprised to see that there is little correlation in baseball.

But as for football, high scoring games have a certain flavor to them that low-scoring games do not. High scoring games are more likely to have fast scoring drives that do not give the defense time to rest. In baseball, having a high-scoring offense does not increase the pressure on the pitcher. They still only have to get 27 outs.

78
by thestar5 :: Wed, 12/17/2008 - 2:36am

Thanks for clearing that up.

112
by Pat (filler) (not verified) :: Wed, 12/17/2008 - 7:51pm

"being higher" is not a correlation.

In this case, you're not entirely correct. If you correlate "winning score" vs "losing score", you actually will see a correlation even if there isn't one at all. It's just selection bias because the winner's score is higher.

I'm assuming that the original poster didn't make the mistake of correlating "winner's score" vs "loser's score", and actually correlated "score 1" vs "score 2." If I'm wrong, yeah, the parent poster is actually right.

Draw two random numbers, from 0 to 10. Take the higher number in each pair, plot "higher number" vs "lower number", and you'll find a correlation, even though they're completely random. If "higher number" is 1, "lower number" is zero. If "higher number" is 10, "lower number" will be 5, on average. If "higher number" is 6, "lower number" will be ~3 on average.

Poof, correlation.

If, however, you plot "first number" vs "second number", that all goes away, obviously.

121
by Pat (filler) (not verified) :: Thu, 12/18/2008 - 12:15am

... Of course, I could've just done the correlation myself and seen if he made the mistake. Which I think he did. A multi-year regression (93-96) found the correlation between scores, when properly randomly selected (done by duplicating the data set mirrored - it's the easiest way), to be on the order of an r = 0.05 - 0.10, nowhere near the r of 0.4 that the original poster mentioned.

If you screw up, however, and do "winner vs loser", you get a correlation of 0.5, strikingly close to what the original poster had.

So yeah, I think the proper answer is that the original poster just screwed up the statistics.

21
by thestar5 :: Tue, 12/16/2008 - 5:40pm

Sorry, not that I think you're wrong at all, but I don't understand what you mean. How is it possible to not have the winner's score correlate with the loser's score since the winner's score must be higher?

22
by thestar5 :: Tue, 12/16/2008 - 5:40pm

Sorry, not that I think you're wrong at all, but I don't understand what you mean. How is it possible to not have the winner's score correlate with the loser's score since the winner's score must be higher?

66
by Pat (filler) (not verified) :: Tue, 12/16/2008 - 10:18pm

I don't think Pythagorean estimations are valid for football.

Of course they're valid. The fact that they work, and have a pretty high correlation, kinda proves that.

In football however, scores for winning and losing teams correlate significantly (above r=0.4).

That's just scoring pace. It's about what you would expect, given that the standard deviation for number of drives in a game is about 15-20% of the mean (it's like 12+/-2 or so, if memory serves). It's why a pace-free stat would be, and is, better (that's what Estimated Wins is for). Basketball has the same problem, as does any sport with a decent scoring rate and clock.

So no matter how you adjust the Pythagorean exponent, the win estimations are going to be very biased.

"Very"? Not really. They'll have a significant amount of error, but not really bias. Total points is just going to be (points/drive*drives/game), and so the difference between the variation in points/drive and total points is just going to be on order of the same variation in drives/game. Adding in average points/drive, you're talking about an error on the order of 15-20% or so. The bias is just going to be convoluted with points/drive spread, so it'd be like 5-10% or so.

128
by Rich Conley (not verified) :: Fri, 12/19/2008 - 11:39am

"Of course they're valid. The fact that they work, and have a pretty high correlation, kinda proves that."

Eh, they're poor tools. Teams with great defenses consistently under perform Pythagorean wins as compared to teams with great offenses. The methodology has serious flaws.

137
by Pat (filler) (not verified) :: Fri, 12/19/2008 - 11:25pm

That might have something to do with the fact that it's based on two numbers. Pythagorean wins are a first-order estimation. Of course it has flaws, but it also has the least assumptions. You take two numbers from each game. That's it.

In any sport where the scoring is semi-independent and follows a Weibull (or Gaussian-like) (and it is semi-independent in football), the Pythagorean theorem is going to "kinda work."

Pythagorean wins don't really work in baseball, either, since the run-scoring environment is different for each game (depending on who the managers are pitching) but it's still a useful average.

32
by Scatman (not verified) :: Tue, 12/16/2008 - 6:05pm

hmm...there is something fishy going on with the OL stats......there ain't no way Atlanta has given up 26 sacks....14 by my (and the NFL's) reckoning.........

34
by FireOmarTomlin :: Tue, 12/16/2008 - 6:10pm

No, there's something wrong with your reading comprehension.

That 2way table is color separated for a reason.

http://www.footballoutsiders.com/stats/ol

33
by kevinNYC (not verified) :: Tue, 12/16/2008 - 6:10pm

The Giants offensive numbers have dropped them like a rock the past 2 weeks.

I do have to admit seeing the Eagles at the top of this list bothers me quite a bit.

35
by Joe T. (not verified) :: Tue, 12/16/2008 - 6:16pm

NY Bretts is clearly ranked too low because Brett Favre is just having fun out there, and Brett Favre just wins. My great-aunt making her pro-bowl selections based solely on the cuteness of the players' butts is way better than this. LOL, Rams are at botom of dung heep, makes me ROFL. Rams wil never win w/ gold pants, need real QB like Jared Lorenzen. IMO Frisco fans need to STFU, there team is crap.

85
by Tundrapaddy (not verified) :: Wed, 12/17/2008 - 11:26am

It's almost like Raiderjoe discovered a alias.

100
by Bobman :: Wed, 12/17/2008 - 3:58pm

I was thinking they are cousins... and everyone in the family is named Joe.

36
by NHRevsFan (not verified) :: Tue, 12/16/2008 - 6:18pm

How can you go about isolating the luck factor in the playoff odds of the Jets vs Pats? Or more precisely, how much of their edge in playoff odds can be assigned to that OT win over the Pats, given that their DVOA rankings are otherwise pretty similar?

The Jets have slightly better special teams and more consistently average offense and defense, while the Pats have a better than average offense and worse than average D. Netted out, though, there's not a hell of a lot of difference between the 2 teams.

The Jets won that OT game, though, and the Pats lost, and it pretty much came down to the flip of a coin, since it's entirely possible, even probable, the Pats would have driven and scored just as readily as the Jets did.

73
by RickD :: Wed, 12/17/2008 - 1:10am

Well, without the coin flip, the Jets would be a game behind, losing all the tiebreakers, and pretty much dead in the water.

I don't think the playoff odds care so much about the games in the past as much as they do about the games in the future.

Personally, I think the Jets are playing like crap and if they don't lose at Seattle they will get throttled by the Dolphins in Week 17.

37
by jimm (not verified) :: Tue, 12/16/2008 - 6:22pm

Special Teams for the Vikings have obviously hurt them. Average special teams play and the Vikings are roughly the same level as the top 5 teams.

Longwell must be near the bottom of the league in KO distance. There never seems to be a Viking within 20 yards of a punt returner on any Kluwe kicks and his touchback rate must be one of the worst as well. Beyond that the TD returns on punts and blocked FGs (7) I would suspect has a large element of bad luck.

If they can play any where near even on special teams I really don't see that any team in the league is appreciably stronger.

40
by BlueStarDude :: Tue, 12/16/2008 - 6:29pm

Aaron,

I don't understand why you still haven't discussed the Dallas DVOA with and without Romo. In 2005 when Roethlisberger missed a few games it was a major discussion point. We know that the drop-off from Romo to BJ was more than the drop-off from Big Ben to Maddox, so I don't get it. This is asked about in the comments every week. What gives?

43
by The Powers That Be :: Tue, 12/16/2008 - 6:47pm

I'm not sure why they won't discuss it (well, I have my ideas on why), but the numbers are roughly as follows (not acccounting for how eliminating those 3 games would change everyone else's averages too, but we do what we can):

Dallas DVOA, minus the Brad Johnson Experience:
Total DVOA: 21.82%
Offense DVOA: 20.99%
Defense DVOA: -5.61%
ST DVOA: -4.75%

That would put Dallas 6th, behind PIT and ahead of CAR, which sounds just about right for the season (of course, they would seem to be on the upswing now). Surprisingly the Offense number would top the NFL, while they'd be 9th in Defense.

44
by thestar5 :: Tue, 12/16/2008 - 6:52pm

Thank you, it'd be nice if we could see what their playoff odds would be too. I have no idea why this is never addressed or discussed. Instead, every Cowboy win is a combination of mistakes and bad luck by the opposing team apparantly.

45
by Matt W (not verified) :: Tue, 12/16/2008 - 6:52pm

Actually by DVOA the drop-off from Romo to Johnson isn't much bigger than the drop-off from Roethlisberger to Maddox at the time -- about 75% for Romo/Johnson, 69% for Roeth/Maddox.

50
by Love is like a bottle of gin (not verified) :: Tue, 12/16/2008 - 7:03pm

One possibility is that the tone of the Dallas fan comments has pissed him off. Really not trying to be snarky and not pointing the blame stick at you or anyone, just making an honest guess.

63
by BlueStarDude :: Tue, 12/16/2008 - 9:15pm

I'm pretty sure Aaron's above not looking into something just to spite some fan base. It probably just hasn't occurred to him. I guess I could also ask myself: if I really care so much why don't I email Aaron since I know he doesn't have time to read through the comment streams anymore?

74
by RickD :: Wed, 12/17/2008 - 1:12am

What an odd comment to make.

Aaron, why aren't making exactly the same kind of comment that you made in some obscure column 3 years ago? What, are you chicken???

135
by cjfarls :: Fri, 12/19/2008 - 5:32pm

The most obvious reason I can think of, is that if Dallas makes the playoffs (or even just before Philly/Dallas), this would be a GREAT feature article.

Why waste such a great story in throw-away comments in a regular article... you need to think like like a news editor, not just as a whiny fan.

46
by vesini :: Tue, 12/16/2008 - 6:54pm

Maybe this is consistent year-over-year, but has anyone noticed that eight of the the top ten teams are all in the top ten for defense, but in the teens (or worse) on offense? The lone exception is the NY Giants, but they've been an exception all season long, 'cause they have been in the top ten for all three stats (O, D, and ST), but is there some correlation to this?

Are we back to traditional bobble-headed wisdom - Defense Wins (DVOA) Championships?

Discuss!!! ...

vesini, who did not use the proper stats, and is dead.

48
by vesini :: Tue, 12/16/2008 - 6:56pm

Sorry, forgot Carolina in that comparison, who along with the NY Giants is 6th on offense, but still, that defense stuff is inconcievable, right?

I couldn't resist!

vesini, who did not use the proper stats, and is dead.

70
by Bobman :: Wed, 12/17/2008 - 12:28am

Vizzini?

You fell victim to one of the classic blunders! The most famous is never get involved in a land war in Asia, but only slightly less well-known is this: never go in against a Sicilian when death is on the line!

83
by vesini :: Wed, 12/17/2008 - 11:01am

Uh, yeah ... that's kind of the point ... a smart guy who didn't understand simple things (like game theory) got killed because he thought he was so smart he couldn't be fooled ... the mispelling denotes the lesson, you see?

And yet, not one comment on the Defense stuff, huh? What's the deal, FO'ers? Are we all now subscribing to STANDARD FOOTBALL LOGIC (or as I like to call it, Wade Phillips logic) ... WHY ARE ALL THE DEFENSIVE TEAMS AT THE TOP OF DVOA??????

vesini, who did not use the proper stats, and is dead.

101
by Bobman :: Wed, 12/17/2008 - 4:09pm

Well, I'd think that anyone with the temerity to call Aristotle, Plato, and Socrates "Morons!" would know how to spell his own name. But I digress....

The simplistic take on the high ratings of strong D teams is that it's been an off year for top-flight QBs. Say guys like Manning, Brady, McNabb, Ben R, Romo, Hasselbeck, Palmer, Bulger, etc al give their team's O ratings a little boost at the same time they erode the D ratings of their opponents. The rsult is that a lot of the top D teams are a little lower and a lot of the top O teams are a little higher.

As I said, very very simplistic. But think about P. Manning from 2004-2007 and Brady in 2007--each time they play full-throttle, they chisel away a bit at the D'd DVOA and bulk up their own. This affects 14 teams or about half the league by year's end (their opponents and their own team, less the div rivals they play twice). Add two elite QBs to the mix, and many of the D's in the league lose a bit, plus their teams climb a bit.

What teams have had QB dropoffs and what teams have had improvements if you average the past few years and compare them to this year. They may equal out, shredding my theory, or it may be that there is a big net dropoff in QB play.

47
by jimm (not verified) :: Tue, 12/16/2008 - 6:55pm

this isn't necessarily the thread for this comment but Jared Allen is a backup in the pro bowl to Peppers and Tuck. He has clearly better stats

He's ahead of them in sacks 14.5 to 12.5 and 12. He is top 3 in both hit and hurries according to an article here a week ago. Neither Tuck or Peppers is in the top 8 or so listed. Tuck has the edge in tackles (60-48) and he had a Int TD return but Allen had 2 safties.

On the other hand I don't think Pat Williams is anywhere near the force he was in the previous two years and he made it.

51
by Love is like a bottle of gin (not verified) :: Tue, 12/16/2008 - 7:05pm

The Pro Bowl mans absolutely nothing. Stop pretending it does.

60
by zdneal@yahoo.com :: Tue, 12/16/2008 - 8:14pm

People commenting on the Pro Bowl aren't evil. Stop treating them that way.

75
by Love is like a bottle of gin (not verified) :: Wed, 12/17/2008 - 1:18am

I wasn't saying he was evil I was trying to help him. Trust me i cared about the Pro Bowl for 2 or 3 years too. After I realized it was hopeless the whole process is much more enjoyable (in that it can be safely ignored).

This year it looks like they actually did a half decent job.

52
by Thok (not verified) :: Tue, 12/16/2008 - 7:18pm

In retrospect, the Rams' 34-14 win over Dallas in Week 7 is just incredible, Brad Johnson or no Brad Johnson. The Rams' DVOA in that game is 64%.

Not that incredible. Haslett is a horrible coach, except for his ability to get his players really pumped up for 2 or so games a year, which probably adds about 30& DVOA by itself. After that, his players stop listening to him.

Seriously, somebody ought to hire him as a play-off motivation coach and use him just to inspire a team to a conference championship and Super Bowl rout.

54
by jimm (not verified) :: Tue, 12/16/2008 - 7:31pm

Gin Bottle guy - don't tell Antoine Winfield that - he just commented that it was the one of the best days of his life.

But you are right it doesn't mean much in terms of evaluating the best players.

55
by Reinhard (not verified) :: Tue, 12/16/2008 - 7:32pm

"The NFC divisions go to the New York Giants, Carolina, Philadelphia, and Tampa Bay, with Atlanta and either Minnesota or New Orleans as wild cards"

I don't understand... there are only two NFC Divisions between Giants and Philly, Carolina and Tampa

56
by Eddo :: Tue, 12/16/2008 - 7:40pm

That was in reference to the simulations where the divisions were realigned to be more balanced.

57
by jimm (not verified) :: Tue, 12/16/2008 - 7:46pm

The wide open throw it all over the field type teams like Indy and NE of previous years aren't particularly good teams in 2008.

NO, Arz, NE, Den, GB are all good at throwing the ball but aren't very good teams.

Balt, Pitt, Tenn, Minn, and Tampa are all very challenged passing offences, yet are all top 10 teams.

58
by Aaron M. (not verified) :: Tue, 12/16/2008 - 8:05pm

Check out how bad Denver is (#23), and they are going to win the AFC West. That's just pathetic. The Chiefs are only 2 spots behind them at 25 in weighted DVOA, although there is a significant gap in the numbers 9.4 to 17.9. After #24, that's quite a drop off. At least the Chiefs are the BEST of the WORST!

61
by Will Allen (not verified) :: Tue, 12/16/2008 - 8:21pm

Regardless of how the Vikings finish with their run stopper injured, they have outpaced my projections for them, given their difficult schedule. Their quarterbacking has been every bit as awful as I expected, and their special teams hideous beyond belief. This speaks to how extremely good they have been on the line of scrimmage, and to why Adrian Peterson would be a very good pick for MVP, which is usually a choice I'm indifferent to. No, I'm not interested in what Peterson's FO stats look like, because those stats don't capture context. This team would be about 5-9, at best, with even a good running back, say, Brandon Jacobs, playing in place of Adrian Peterson.

119
by The Ninjalectual :: Thu, 12/18/2008 - 12:08am

What about with a good running back, say, Chester Taylor?

65
by chubbypuppy (not verified) :: Tue, 12/16/2008 - 9:50pm

It's been covered by ammek and others but the Packers defensive line issues go WAY beyond Cullen Jenkins.

--trading Corey Williams
--expecting KGB to return 100 percent from a knee injury
--expecting Justin Harrell (2007 number 1 pick) to do SOMETHING

And Jenkins was nagged all of 2007 by injury. Not a shock he got hurt. Again.

It's the OFFENSIVE line that is the concern because other than Jason Spitz these guys have stunk more than they have shined. Clifton is slipping. Tauscher had slipped before getting hurt. Colledge is ridiculously erratic. Toss in Rodgers lack of pocket radar and it's no surprise the sack totals have gone up.

But let's be clear again for some of the slower audience members:

--Rodgers wasn't the issue this season
--lack of "safety depth" wasn't the issue this year

It was no pass rush.
It was horrible punting
It was incessant penalties

And bad coaching. Both in game prep and within the game. The Packers offense REPEATEDLY starts slow putting the team in catch up mode. And then at key moments Mike McCarthy and Co. have made WHACKSH*T CRAZY decisions highlighted by playing for a 50 yard field goal attempt with 2 minutes left and at the opponents 40 yard line. And 2 timeouts. Seriously. You can't make this stuff up.

Sad thing is that with the offensive tackles on the brink of collapse next year's indicators for being better (based on this season's points differential) will almost certainly be wrong. Clifton's lateral movement continues to erode and who knows if Tauscher will even be back between injury and free agency.

Thompson gutted the line in 2005 letting Rivera and Wahle walk and has never really fixed it.

Better buckle down Teddy boy. You got it right with Rodgers. But you are looking to get everything else WRONG. And that ain't gonna work......

67
by 57_varieties (not verified) :: Tue, 12/16/2008 - 10:35pm

How did the Steelers beat the Ravens, and outgain them, yet lose ground in DVOA?

69
by jonnyblazin :: Tue, 12/16/2008 - 10:48pm

Special teams, probably. The Steelers best special teams play was a fumble recovery and runback for 40 yards, but thats a negative for DVOA. Plus the Ravens had several very good returns. This probably makes up for the offense/defense disparity, where the Steelers outplayed the Ravens but not by that much.

76
by RickD :: Wed, 12/17/2008 - 1:24am

Yes, how did that happen indeed?

Well, the Ravens outplayed the Steelers for 57 minutes and then inexplicably relaxed long enough to let the Steelers score a game-winning TD. DVOA probably thinks the first 57 minutes outweigh the last three.

77
by troycapitated p... :: Wed, 12/17/2008 - 1:54am

Re: The Ravens outplayed the Steelers for the first 57 minutes.

It would be more accurate to say that for the first 57 minutes, the difference in the game was the disparity in special teams play- the Ravens with good punting and kick coverage versus the Steelers with punts traveling longer than previous games, but coverage that was less solid than usual. The performances of the 2 offenses and defenses for the first 57 minutes were basically equal. The Steelers had actually gained a few more yards, but due mostly to the difference in starting field position had not gotten as many points.

Then, the last drive happened and tipped the balance.

94
by DGL :: Wed, 12/17/2008 - 12:53pm

To reprise my reply to the same comment in the Audibles thread:

Before the last PIT drive, Baltimore had 187 net yards and Pittsburgh had 220. Baltimore had 11 first downs and Pittsburgh had 12. Flacco was 9-26 for 100 yards, Roethlisberger was 16-29 for 157 yards. Baltimore had 112 rushing yards and Pittsburgh had 89. Baltimore was 4-14 on third down and Pittsburgh was 6-15. Baltimore had the ball for 27:56 and Pittsburgh for 28:57. That sounds awfully even to me, not like either team manhandled the other.

Although I see that you've changed the verb from "manhandled" to simply "outplayed", I still don't see it.

105
by jonnyblazin :: Wed, 12/17/2008 - 5:08pm

Don't forget that the Steelers fumbled 3 times and the Ravens 1 time. If you are going to use stats, don't only use the ones that back up your argument and ignore the others!

68
by jimm (not verified) :: Tue, 12/16/2008 - 10:39pm

Case in point Will - the Vikings get 6 quarters in the last two games of decent QB play and as a result they score 52 points.

If Jackson can play any where near the level he has in his last two games their offence will be dramatically better. The throw he made on the 2nd TD to Berrian was the best throw I've seen Jackson make yet. He was on the money on the first deep ball to Berrian as well it was just defended well.

It sure is nice to watch a really good defence. The defence has held opponents to less than 20 pts in 12 of 14 games. 19 TD's allowed in 14 games - three of which originated inside the Viking 11 yard line after turnovers by the offence.

I think you could make an argument for Jared Allen as the MVP. This team would be no where without him.

79
by steelberger (not verified) :: Wed, 12/17/2008 - 7:59am

The Ravens are the 2nd ranked team, yet they are 2-5 against teams with winning records. I'm not saying the system is broken, but there is certainly something off.

81
by dg (not verified) :: Wed, 12/17/2008 - 10:46am

The DVOA analysis appears to have broken down this year in a big way given the disparity of records or, as you point out, quality of wins, to DVOA rankings. I'm waiting for a better explanation from the FO crew.

84
by vesini :: Wed, 12/17/2008 - 11:13am

So, if a team has a horrible record, like the Lions, they have no chance to win any game? The point of DVOA is to determine how the team is performing across all other teams. Oddly enough, this might mean that a team's record is not a reflection of its performance, but of other factors, like random events or (god forbid) COACHING!!!

Look, I'm an Eagles fan, and I know that the Eagles are a good (not great) team, yet they're 8-5-1, right? But, on a play-by-play basis versus all 31 other teams, their overall performance (on average - Value Above Average) is higher than anyone else's (after defensive adjustments).

So, the reason thy're not 14-0? COACHING!!! The coach seems to either 1) not understand coaching until December or 2) plays possum with all the other teams and screws up his game plans and clock management until December (how's that for a conspiracy theory, Tanier???).

Random events, like fumble recovery and long field goals always being made against you also play a small part, but how can anyone predict those, right?

Hope this helps.

vesini, who did not use the proper stats, and is dead.

95
by DGL :: Wed, 12/17/2008 - 1:00pm

As others have pointed out, a higher DVOA is no guarantee of victory. Looking just at PIT-BAL, you have two teams that are very close. If they played 100 times, maybe Baltimore would win 55 and Pittsburgh 45 - but the scenario where Pittsburgh wins both games is by no means unusual. And indeed, the two games came down to an overtime FG and a last-minute touchdown (which reasonable people can argue should not have been called a touchdown, meaning the game would likely have gone to OT like the first game). I don't see that as "broken DVOA", I see it as "statistical margin of error".

97
by steelberger (not verified) :: Wed, 12/17/2008 - 2:17pm

Your margin of error argument looks at only the Balt-Pitt games. You ignore the fact that Baltimore has not beaten anyone good this year.

They have played 7 games against teams with winning records. They lost 5 of them.

They have beaten Miami (who's winning record even DVOA says is a product of it's schedule) and Philly (the bipolar team that tied the Bengals the week prior).

My point is that I think blowout victories over such powerhouses as the Raiders, Texans, and Bengals shouldnt outweigh the fact that they cant beat a playoff caliber team.

104
by jonnyblazin :: Wed, 12/17/2008 - 5:01pm

except they have beaten two playoff caliber teams...

so which playoff caliber teams have the Steelers beaten? NE is an average team, not as good as Miami. Dallas isn't as good Philly either, and Philly pounded the Steelers into the ground in case you don't remember. Ravens and the Steelers both beat the Skins if you want to qualify them as playoff caliber, but I wouldn't.

The difference separating the Steelers and Ravens in the standings are the two games they played against each other, which were pretty close. The DVOA difference is due to the fact that the Ravens don't mess around with average to below average teams and simply blow them out. Steelers won games by 1, 5, and 4 points against crappy teams like SD, JAC, and CLE.

The Ravens have a good formula for beating bad teams. Punish them with the run, beat them with the bomb. It doesn't work against good defenses because the running game is efficient but not explosive, so 5-6 yard carries become 3-4 yard carries, and Mason and Clayton can beat CBs 1 on 1 when bad teams move 8 into the box but not when they are playing 2 deep.

106
by dg (not verified) :: Wed, 12/17/2008 - 5:14pm

Thanks for the comments. I don't think I'm stupid, so I think I get the idea behind DVOA. My problem is that it assumes that all plays count equally, which they do not. If you took the offensive DVOA or whatever adjustments you wanted to make for the final drives--those required to beat the Ravens and the Cowboys--I think you'd find that the Steelers offense finds fourth and fifth gear. You can call it coaching or quarterbacking or whatever, but there are certain plays in a game that matter more and are not accounted for in this modeling exercise. You assume that because Baltimore does better than 31 teams on average 55% of the time while Pittsburgh only does it 45% of the time then that means that the Ravens beat the Steelers 55 out of every 100 games. I say that the Steelers might beat the Ravens 85% of the time because not all of those plays are equally representative. A team with intangibles (coaching, QB, heart, whatever) can throw the stats out the window because key plays are not normally distributed. I am not saying that this is the disconnect that explains the greater than historical discrepancy between W-L records and DVOA in this particular year, but it is a nagging doubt I've always had about the assumptions underlying the predictive value of DVOA. I'm no statistician or quant-jock, so I could be wrong. Would welcome feedback from brighter minds...

109
by Staubach12 :: Wed, 12/17/2008 - 6:28pm

DVOA does not give equal weight to every play (at least as I understand it). It recognizes that certain plays and certain points in the game are more crucial than others.

110
by jonnyblazin :: Wed, 12/17/2008 - 6:37pm

I think its a question of sample size. No doubt the Steelers have had some clutch drives this season, I'll admit that. Lets say the Steelers in the divisional round are trailing by 4, have the ball with 2:00 to go, and fail to score. Would that disprove your theory? Because you are suggesting that, given the evidence this year, the Steelers are definitely better when it matters. So how many instances of failing in the clutch would it take to disprove that notion?

114
by dg (not verified) :: Wed, 12/17/2008 - 8:16pm

I'm not suggesting that Pittsburgh definitely is better in the clutch, but I am raising it as a possibility. When asked what the difference was between the multiple failed drives earlier in the Dallas and Baltimore games and the final, decisive drives, multiple Steelers noted a sense of urgency, which is not easily captured in the stats but flies in the face of an assumption that all plays are the same. Now I'm not good enough at stats to know how many of these critical, differentiated plays need to be in a sample size to say anything meaningful about it. But my comments are not really about the Steelers but about DVOA and the FO methodology. If DVOA does account for these critical plays in a game, like third down or fourth quarter drives, then how does it do it? It's really a black box to me. Hence my original post...

115
by jonnyblazin :: Wed, 12/17/2008 - 8:37pm

well, I believe the premium content divides DVOA up by quarter, and there might be a "late and close" metric if you think that'll measure clutchiness. I don't have premium so I can't be certain though.
I think there was also a "secret sauce" analysis done awhile ago to measure which regular season metrics best correlate to postseason success. I don't remember a lot of that, but I think defense is a greater indicator than offense.

118
by Pat (filler) (not verified) :: Wed, 12/17/2008 - 11:52pm

I'm not suggesting that Pittsburgh definitely is better in the clutch, but I am raising it as a possibility.

One of the concepts which is ridiculously common in sports analysis is "leverage." An at-bat has high leverage if there are multiple runners on base, as opposed to the bases being empty. A play has high leverage if it's third down, rather than first down. As in, the outcome has a big change on the progress of the game. Much more so than other plays.

This is what you're talking about when you think of "important" and "unimportant" plays.

You have to then ask "do teams play better in high-leverage situations than regular situations?" Sure. Of course some do. But do they continue to play better, or do they regress to the mean? In general, teams tend to regress to the mean. They tend to play the same on high and low leverage plays.

So while the game doesn't treat all the plays the same, in general, if you evaluate the team's performance, you should.

Let me be a bit more concrete:

When asked what the difference was between the multiple failed drives earlier in the Dallas and Baltimore games and the final, decisive drives, multiple Steelers noted a sense of urgency

What you're talking about here is how the Steelers won, not why they won. They won because they performed better in a high-leverage situation. This is not a guarantee they will do so again.

It's kindof like a fumble recovery: when a fumble happens, people jump, scramble, and heck, sometimes alert, quick thinking players hit the ball away, and finally one person ends up with it. You can say "that team just wanted it more," which is true. But it tells you nothing about how they will perform the next time. There is no "want-meter." The ball bounces around, someone makes a play, and it happens. Sometimes we say "luck," but we really mean "unpredictable." It's like when two equally-matched players play chess: someone has to win, but it doesn't mean that one player's better. It just means that they played better at that moment.

Put yet another way: the problem with saying that "Pittsburgh is just better in the clutch," is that I can then say "well, then, why doesn't Pittsburgh just play that way the rest of the time?"

126
by dg (not verified) :: Thu, 12/18/2008 - 5:00pm

Nice post, Pat. I am not a statistician, but an investor. So I am perhaps out of my depth here. But I do also believe in probabilities and mean reversion, which in my experience generally operate like gravity. You see some investors doing well for a while and then reverting to average performance, and you see companies posting fat margins for a while and then mean reverting. But then you have notable exceptions, less than 5%, that defy gravity by showing persistence in their investment record (Buffett) or sustainability of supra-normal margins (Coke). I've read about championship caliber athletes that do this as well, seeming to defy the odds by being clutch much more often than not (or than average). You say that they all regress to the mean, but you don't offer quantifiable proof that they do. Did Michael Jordan have the same percentage success rate on clutch shots as the average basketball player? Was John Elway's comeback drive success rate the same as the average QB? I don't have the data either to say for sure, and it could be that we confuse high frequency with high success rates, but anecdotal evidence I've seen seems to contradict what you are alleging. Now, of course even those great players do not sustain that level all of the time, since it probably would be too exhausting. But they do seem to be able to call it up at those important moments with more regularity or consistency than the average player. This is not to say that they have magical powers to will the errantly bouncing ball to fall their way, but they might put themselves in a zone in which the bouncing ball appears to move more slowly and thus is seized or they remain relaxed enough under pressure to prevent the ball from bouncing out at all. The difference between Tiger Woods and a very good golfer is that the latter will throw up and feel faint under the pressure of the money shot, while Tiger will hit it without his heartbeat going up at all. This is what I mean by clutch, and I definitely believe that if the game consists of unequally important moments, then a team that rises to the occasion in those moments more often is exceptional even if the aggregate statistics say otherwise. Of course, I'd love to see whether you can use stats to bear this out or not. And, again, I'm still not sure that DVOA does this or attempts to.

127
by jonnyblazin :: Thu, 12/18/2008 - 8:42pm

Hmmm, I think the problem here is you are comparing Elway, Jordan, and Woods to the average player. I think we can safely say that these guys are probably in the top 5 for their position/sport of all time, as documented by their regular season stats. Was Jordan really that much better in crunch time than he was in the first quarter? Not really. At all times he was the best player on the floor, so it follows that he'd be the best player during crunch time as well.

If you can find an average player who somehow raised his play time and time again in the clutch (Robert Horry?), then maybe you can say they are clutch. But there is definitely a sample size issue there. Usually great players do well in the clutch because they are better than everyone else.

129
by Rich Conley (not verified) :: Fri, 12/19/2008 - 11:53am

"If you can find an average player who somehow raised his play time and time again in the clutch (Robert Horry?), then "

IMO, you're misunderstanding "clutch". Clutch players don't elevate their game in high leverage situations, they just stay level while everyone else is feeling the effects of pressure. Tiger woods doesn't putt better in a playoff hole, he just doesn't screw up.

130
by Mr. Clutch (not verified) :: Fri, 12/19/2008 - 2:10pm

It is difficult / impossible to measure the psychology inside a player's head or a team's collective mindset, so it is hard to do statistical analysis on it. Based on my own guess, I think being clutch is over-played but also is real.

If you believe you are clutch, you may be more confident and behave in ways that make winning more likely. Conversely, if you believe you are a choker, you may be less confident and be more likely to fall apart in key moments.

But there's no way to measure what was really going on in the player or team's heads. As a sports fan, we look at the results and infer the reason. You saw Jeter make some great play in the clutch, so he seems clutch. But his stats in the post-season are basically the same as the regular season. Is he clutch or not? I say no.

As a Steelers fan, I see Ben regularly orchestrating 4th quarter comebacks over his career. I certainly have faith in him at the end of the game, and I think he's clutch. But I can't prove it. And even if I could, the future doesn't perfectly predict the past.

133
by A Whore (not verified) :: Fri, 12/19/2008 - 3:39pm

That one's easy.

Pittsburgh doesn't "play that way the rest of the time" because the first 55 minutes of any given football game are a good time to establish their worst-in-the-nfl running game.

123
by Jerry :: Thu, 12/18/2008 - 3:05am

Which is the better offense? One that scores on three drives in the first half and coasts in the second, or one that has six unsuccessful drives before pulling out a win in the last minute?

120
by The Ninjalectual :: Thu, 12/18/2008 - 12:12am

Historically, "blowout victories over such powerhouses as the Raiders, Texans, and Bengals" are better predictors going forward than close games against good teams.

132
by A Whore (not verified) :: Fri, 12/19/2008 - 3:35pm

True, but the Steelers won those games by an average of 24 points. They just haven't played many teams in that category.

I think DVOA really loves em some Ravens because of the Philly game, but the Ravens O didn't do anything until what was basically garbage time.

86
by joenamath :: Wed, 12/17/2008 - 11:42am

interesting is that this year you could easily have 4 wildcards advance to the second round

102
by Bobman :: Wed, 12/17/2008 - 4:15pm

Excellent. Maybe they need a new slogan: "NFL > BCS"

91
by jimm (not verified) :: Wed, 12/17/2008 - 12:00pm

Tarvaris Jackson DVOA 11.3% - Gus Frerotte -7.0%. Keeping my fingers crossed that this shows real improvement as opposed to a small sample. Interestingly Jackson DVOA last year was -5.8% with essentially what of the service would seem a weaker offence minus Bernard Berrian.

I've been accused of cherry picking stats - but if you look at Jackson's performance over the last 11 or so games it is miles better than other QB's performed in the same system.

92
by jimm (not verified) :: Wed, 12/17/2008 - 12:13pm

Tarvaris Jackson DVOA 11.3% vs Gus Frerotte -7.0%. Keeping my fingers crossed that this shows real improvement as opposed to a small sample. Interestingly Jackson DVOA last year was -5.8% with what would seem a weaker offence (same player with one change Bernard Berrian replaces Williamson). Same line, same RB's and TE, same Bobby Wade, same everything, just Berrian exchanged for one of the worst receivers in the NFL and Jackson still had a better DVOA in 07 than Frerotte did this year.

93
by noahpoah :: Wed, 12/17/2008 - 12:20pm

Why highlight the weighted DVOA columns if they're not determining the overall rank? Also, why put the overall rank, unlabeled, to the left, with last week's rank to the right of Total DVOA, when all the other ranks appear to the right of the score determining the rank. Also agin already, the link to the DVOA explanation is broken (it's missing /info/ in before methods#dvoa).

Finally, it has become lore around these here parts, but I can't find any strong evidence of the assertion that, as you wrote in this week's Quick Reads, "Because our research has shown that fumble recoveries are random chance,...". What was the research that shows this? I'm curious how you define random, what the evidence is, and whether or not it depends on any of a number of possible contextual variables.

96
by Love is like a bottle of gin (not verified) :: Wed, 12/17/2008 - 2:12pm

I agree with the criticism of how the chart is laid out, I have been coming here for 4 or 5 years and I still find it a huge pain to read. You simply do not do the "left side, similar data range but different data right side, right side" thing in a good chart.

98
by Love is like a bottle of gin (not verified) :: Wed, 12/17/2008 - 2:20pm

After playing with some different formats I think the format of:

TEAM, DVOA, RANK, DELTA (write the data as +3 or -1 or -(helps distinguish it from other data)), WDVOA, RANK

That is about 10 times more readable.

As for the highlighting I would ditch it all together, it adds nothing.

99
by Anonimous (not verified) :: Wed, 12/17/2008 - 3:30pm

as for 'random fumble recovery',

i'm not going to google for the article,

but the argument was that while some teams have very high fumble recovery rates and some teams have very low ones,

there is no correlation with what the recovery rate is the next season for either of those teams. which is to say, fumble recovery rates appeared identical to what you would expect them to look like if each recovery was determined by a coin flip.

that said, they have admitted that certain types of fumbles have different recovery rates; botched snaps tend to be recovered by the offenses, receptions that are fumbled downfield tend to be recovered by defenses. but those metrics would require game charting to go into the formulas, and that's not what they want to do.

116
by Pat (filler) (not verified) :: Wed, 12/17/2008 - 9:39pm

I'm pretty sure fumble location is taken into account. Fumble recovery rates are regressed back to the expected fumble recovery rate based on type of fumble, not 50%.

103
by Billy-Boy (not verified) :: Wed, 12/17/2008 - 4:17pm

If the season had already ended, Atlanta, Miami, Philadelphia, New England, and Chicago would not be in the playoffs, but according to the FO's playoff odds each one of these teams has a better chance of making it to a conference title game, the Superbowl, or winning the Superbowl than does Arizona, and Arizona has already clinched a playoff berth. Chicago only has a 4.5% total chance of making the playoffs at all, and the Bears are still a better bet to win the Superbowl than the Cardinals.

If this were TMQ, it now would read "Ye Gods!" followed by a 1000-work explanation of how a seeded playoff is needed in the NFL.

108
by FavreFan (not verified) :: Wed, 12/17/2008 - 5:25pm

Another packer fan ranting...

Agree that the offensive and defensive lines are mostly to blame. Most safety tandoms in the league would give up just as many big plays if their d-lines produce 0 pressure all season. I believe that Thompson, after sitting on the sidelines the last few years in FAgency, will target and over-pay (do top free agents want to play in Green Bay?) for a top d-line free agent.

Based on available talent coming in through the draft, I would not be surprised if the team took an OT in the 1st round to replace Clifton. There looks to be solid depth at the top of the draft at OT, and this year's class indicates that it may be a position that can contribute immediately.

One other random note - I would really like to see the team use a lot of spread/hurry up next season. With Aaron becoming more comfortable, I'd love to see the team try to mitigate its o-line deficiencies by using the hurry up and wearing down the opposing defensive line. The team certainly has the talent at WR to do it.

113
by jimm (not verified) :: Wed, 12/17/2008 - 7:51pm

Before the season I predicted that there would be all sorts of talk about how the Packers OLine had dropped off badly in performance. The reason I suggested was that Rogers sack rate as an NFL Player was very high. While it is true Rogers sack rate wasn't very high 6%, it was still more than double the rate Favre had the year before (2.7%).

I suspect the change from Rogers to Favre had something to do with the line looking so bad this year.

131
by Flounder :: Fri, 12/19/2008 - 2:35pm

It's a little bit of both. Rogers definitely hangs on to the ball too long sometimes, but nevertheless the o-line has had a couple horrendous games, pass-protection wise. Since Tauscher's injury Tony Moll is the picture of inconsistency. He will play well for stretches, but when he gets beat, he gets bead BAD.

Certainly the o-line was never as good at pass-protection as the adjusted sack rate under Favre would lead one to believe.

134
by A Whore (not verified) :: Fri, 12/19/2008 - 3:59pm

I don't get a couple things.

The Steelers are 29th in yards per carry (only that high because of successful garbage time running), and 32nd at picking up 3rd and 1, and yet there are ten teams with worse rushing VOA than the Steelers?

The Steelers are 29th in net yards per punt while rarely pinning anyone inside their 20,

yet they are an above average punting team?

136
by Billy-Boy (not verified) :: Fri, 12/19/2008 - 6:05pm

If the playoff teams were decided by taking the top twelve teams in DVOA, and the teams were seeded by record, (DVOA as tie-breaker) this is how the playoffs would look:

(Note: Pit and Car are switched to keep AFC/NFC top seeds together for the "divisional" round, and teams advance in the bracket by DVOA matchup)

Round 1:

Game 1: 8 MIN vs 9 ATL
Game 2: 5 IND vs 12 GB
Game 3: 6 BAL vs 11 NO
Game 4: 7 TB vs 10 PHI

Round 2:

Game 5: 1 TEN VS 8 MIN
Game 6: 4 PIT VS 5 IND
Game 7: 3 CAR VS 6 BAL
Game 8: 2 NYG VS 10 PHI

Round 3:

Game 9: 1 TEN VS 4 PIT
Game 10: 6 BAL VS 10 PHI

Superbowl: 1 TEN VS 10 PHI

To me, this looks infinitely better than whatever will happen in the actual playoffs.