Writers of Pro Football Prospectus 2008

printable version

DVOA Analysis

Football Outsiders' revolutionary statistics that break down every single play of the NFL season


Most Recent FO Features

CMcCoy.jpg

» SDA/FEI Bowl Mash-Up Part III

Buckeyes or Longhorns? Refresh yourself with the wisdom of our college football experts before tonight's Fiesta Bowl, then use this thread for in-game comments.

18 Nov 2008

Week 11 DVOA Ratings

by Aaron Schatz

Some things just try my sanity. Take, for example, the Philadelphia Eagles. The good news this week is that the Eagles drop behind the undefeated Tennessee Titans. The bad news is that they don't drop any further than that, and they still stand out as a team where DVOA seems to bear no connection with actual wins and losses.

Or do they?

Two spots below Philadelphia, very close in DVOA, you will find Green Bay -- a team that has an even worse record than the Eagles do (5-5). Do these two teams have similar characteristics that might cause them to be overrated by DVOA? Actually, no.

  • The Eagles struggle to run the ball in short-yardage situations, ranking 31st on power runs. That's not a problem for the Packers, who rank ninth on power runs.
  • The Eagles have problems converting third downs in general. Their offense ranks 11th overall but 22nd on third and fourth down. The Packers, however, are actually better on third down. They rank ninth on offense overall, but fifth on third and fourth down.
  • The Eagles were struggling in the red zone earlier in the year, but they aren't anymore. Both the Packers and the Eagles have a red zone offensive DVOA within one percentage point of their total offensive DVOA.
  • The Packers are getting killed by the run -- 25th in defensive DVOA on first down, 32nd on second down -- which constantly gives their opponents nice third-and-short opportunities. Not an issue for the Eagles, who rank between fifth and eighth on every down whether you are measuring pass defense, run defense, or the combination of both.
  • Neither team has played a particularly hard schedule so far, and they rank ninth (Green Bay) and 10th (Philadelphia) in VARIANCE.

There's one particular DVOA split the two teams do have in common: both offenses struggle after halftime. The Eagles' offense ranks third in the first half of games and 20th in the second half. The Packers' offense ranks second in the first half of games and 18th in the second half. Other than that, it's hard to find a place where we could somehow tweak the DVOA formula based on a belief that the accuracy of our ratings for Philadelphia and Green Bay needs to be improved. (That's not meant to be obnoxious -- we're really trying to find a way to improve our ratings to fix the "Eagles problem," and it's damn hard.)

Here's what both teams really have in common: Their wins come by large margins and their losses by small ones. The Packers have lost three games by a field goal or less, and their largest loss is 11 points. All four Philadelphia losses have come by less than a touchdown. The Packers have outscored opponents by 65 points with +7 turnover margin. The Eagles have outscored opponents by 71 points with +3 turnover margin. Despite five losses, the Packers have only one game with DVOA below zero: their Week 4 loss to Tampa Bay. Despite four losses, the Eagles have only two games with DVOA below zero: their Week 2 loss to Dallas and this week's tie with Cincinnati.

The biggest difference between Philadelphia and Green Bay? Their chances of making the playoffs. Green Bay is part of a three-way tie for first place between three 5-5 teams, and ranks 25th in the NFL in future schedule strength. Our playoff odds simulation gives them a 67 percent chance of making the postseason. Philadelphia plays in a division where all four teams have winning records, and has the third-hardest remaining schedule. Our playoff odds simulation gives the Eagles a 41 percent chance of making the postseason -- actually, it gives them less chance than that, or will once we fix it. Apparently we forgot to give the playoff odds simulation the ability to recognize ties.

Speaking of the playoff odds, Tennessee's chances of going undefeated rose only slightly this week, from 4.8 percent to 7.6 percent. What makes this watch interesting is that their toughest opponents are the final two: Pittsburgh and (a now fairly healthy) Indianapolis.

Which team is the opposite of Philadelphia and Green Bay? Carolina, which is 8-2 but ranks just ninth in DVOA. The explanation here is pretty simple: schedule. The Panthers rank 29th in past schedule strength -- and first in future schedule strength. If you're looking for a test of how much the guys in Vegas read Football Outsiders, check the line when the Panthers go to Green Bay in Week 13.

* * * * *

NEW IN PREMIUM THIS WEEK: The Premium DVOA Database now includes Adjusted Line Yards, Adjusted Sack Rate, and all our other various offensive and defensive line stats going back to 1997. Right now, you can view each year; soon, we'll be adding views to see each individual team's history in these stats.

* * * * *

These are the Football Outsiders team efficiency ratings through 11 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 NYG 40.9% 1 41.9% 1 9-1 26.0% 1 -11.2% 7 3.6% 6
2 TEN 30.0% 3 30.4% 2 10-0 8.4% 13 -22.0% 2 -0.5% 22
3 PHI 25.2% 2 24.3% 3 5-4-1 10.6% 11 -14.8% 5 -0.2% 18
4 BAL 24.5% 4 23.0% 5 6-4 4.8% 17 -21.6% 3 -1.9% 25
5 GB 22.4% 7 23.5% 4 5-5 13.1% 9 -6.3% 9 3.1% 8
6 ARI 21.3% 5 21.3% 6 7-3 21.6% 2 -2.1% 10 -2.5% 26
7 PIT 20.3% 6 20.3% 7 7-3 -4.2% 23 -24.9% 1 -0.4% 20
8 TB 17.6% 9 17.4% 9 7-3 2.6% 21 -15.3% 4 -0.4% 19
9 CAR 16.2% 8 18.2% 8 8-2 4.5% 19 -10.1% 8 1.6% 14
10 IND 12.9% 11 15.7% 10 6-4 19.0% 4 4.9% 17 -1.2% 23
11 WAS 10.5% 12 8.0% 12 6-4 12.5% 10 -1.2% 11 -3.2% 28
12 NYJ 7.9% 13 10.0% 11 7-3 4.5% 18 1.4% 14 4.8% 4
13 ATL 7.1% 10 6.5% 14 6-4 16.2% 7 10.8% 21 1.7% 13
14 MIA 6.8% 15 6.6% 13 6-4 16.7% 6 0.5% 13 -9.4% 31
15 NO 4.6% 17 5.2% 15 5-5 19.8% 3 16.1% 26 0.8% 16
16 CHI 2.6% 14 -0.3% 16 5-5 2.9% 20 0.1% 12 -0.2% 17
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 DEN -0.9% 21 -1.3% 17 6-4 18.9% 5 18.2% 27 -1.6% 24
18 MIN -1.8% 16 -3.9% 19 5-5 -4.7% 24 -13.8% 6 -11.0% 32
19 SD -3.3% 18 -6.4% 22 4-6 16.1% 8 21.7% 29 2.2% 10
20 JAC -3.9% 19 -4.0% 20 4-6 7.8% 14 14.5% 25 2.8% 9
21 NE -4.2% 20 -3.6% 18 6-4 6.1% 15 11.6% 23 1.3% 15
22 DAL -5.4% 23 -9.5% 23 6-4 5.9% 16 4.9% 16 -6.4% 30
23 CLE -7.4% 24 -4.5% 21 4-6 -0.5% 22 13.1% 24 6.2% 3
24 BUF -9.4% 22 -12.4% 24 5-5 -5.8% 25 10.5% 20 6.8% 2
25 SF -11.8% 25 -13.1% 26 3-7 -14.3% 28 4.5% 15 7.1% 1
26 HOU -14.1% 26 -12.5% 25 3-7 8.9% 12 26.9% 30 3.9% 5
27 SEA -19.2% 27 -18.2% 27 2-8 -10.1% 26 11.3% 22 2.2% 11
28 CIN -28.2% 30 -28.2% 28 1-8-1 -18.9% 30 6.1% 18 -3.1% 27
29 OAK -30.0% 28 -31.5% 29 2-8 -26.9% 32 6.3% 19 3.2% 7
30 KC -37.0% 29 -33.0% 30 1-9 -11.3% 27 20.2% 28 -5.5% 29
31 DET -45.0% 31 -43.7% 31 0-10 -14.8% 29 32.3% 32 2.1% 12
32 STL -47.3% 32 -45.0% 32 2-8 -19.0% 31 27.8% 31 -0.4% 21

  • 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.
  • 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 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 FUTURE
SCHED 		RANK VAR RANK
1 NYG 40.9% 9-1 46.6% 7.8 1 -3.5% 26 13.2% 5 19.9% 25
2 TEN 30.0% 10-0 30.8% 7.6 2 -2.4% 22 -5.1% 21 3.2% 1
3 PHI 25.2% 5-4-1 29.2% 6.7 4 -2.8% 24 16.9% 3 12.2% 10
4 BAL 24.5% 6-4 23.2% 7.0 3 2.2% 13 3.7% 12 18.4% 21
5 GB 22.4% 5-5 17.8% 6.5 7 -0.3% 16 -8.0% 25 10.2% 9
6 ARI 21.3% 7-3 26.7% 6.6 6 -5.8% 28 -1.3% 17 12.6% 11
7 PIT 20.3% 7-3 19.5% 6.7 5 5.2% 6 1.9% 14 5.8% 2
8 TB 17.6% 7-3 20.6% 6.1 10 -1.0% 17 -10.1% 28 19.0% 22
9 CAR 16.2% 8-2 18.4% 6.3 9 -5.8% 29 18.3% 1 19.1% 23
10 IND 12.9% 6-4 9.1% 6.4 8 5.6% 5 -11.6% 30 20.4% 26
11 WAS 10.5% 6-4 8.6% 5.7 12 0.2% 15 6.3% 8 8.4% 8
12 NYJ 7.9% 7-3 18.5% 5.4 14 -12.3% 32 -0.9% 16 23.8% 28
13 ATL 7.1% 6-4 9.0% 5.0 17 -2.2% 20 -2.8% 19 24.8% 30
14 MIA 6.8% 6-4 15.9% 5.4 15 -2.5% 23 -20.4% 32 16.9% 18
15 NO 4.6% 5-5 7.8% 5.8 11 -3.0% 25 4.2% 11 7.6% 7
16 CHI 2.6% 5-5 -1.3% 5.6 13 3.6% 10 -8.0% 26 6.5% 4
TEAM TOTAL
DVOA 		W-L NON-ADJ
TOT VOA 		ESTIM.
WINS 		RANK PAST
SCHED 		RANK FUTURE
SCHED 		RANK VAR RANK
17 DEN -0.9% 6-4 2.0% 4.8 20 -4.5% 27 -11.1% 29 18.2% 19
18 MIN -1.8% 5-5 -6.4% 4.7 21 6.3% 3 4.6% 10 7.3% 5
19 SD -3.3% 4-6 -0.3% 5.1 16 -2.3% 21 -6.1% 22 14.7% 15
20 JAC -3.9% 4-6 0.0% 4.8 19 -1.1% 18 9.3% 6 6.2% 3
21 NE -4.2% 6-4 3.8% 4.9 18 -6.8% 30 -2.0% 18 21.8% 27
22 DAL -5.4% 6-4 -6.6% 4.4 23 6.0% 4 16.0% 4 25.6% 31
23 CLE -7.4% 4-6 -11.9% 4.5 22 6.6% 2 9.2% 7 18.3% 20
24 BUF -9.4% 5-5 -4.3% 4.1 24 -7.2% 31 -7.8% 24 13.3% 12
25 SF -11.8% 3-7 -12.1% 4.0 25 -2.0% 19 -7.4% 23 7.5% 6
26 HOU -14.1% 3-7 -18.2% 3.2 27 2.6% 12 2.7% 13 13.4% 13
27 SEA -19.2% 2-8 -23.6% 3.6 26 4.9% 8 -3.4% 20 16.5% 17
28 CIN -28.2% 1-8-1 -38.5% 2.5 29 10.7% 1 4.8% 9 16.1% 16
29 OAK -30.0% 2-8 -31.6% 2.6 28 1.5% 14 -8.4% 27 26.0% 32
30 KC -37.0% 1-9 -33.4% 2.2 30 4.1% 9 -13.0% 31 19.8% 24
31 DET -45.0% 0-10 -45.1% 1.9 31 2.7% 11 17.1% 2 14.4% 14
32 STL -47.3% 2-8 -50.3% 1.4 32 5.1% 7 1.4% 15 24.5% 29

Posted by: Aaron Schatz on 18 Nov 2008

by DZ (not verified) :: Tue, 11/18/2008 - 5:50pm

I'm glad I'm not the only one who occasionally runs into freaky formatting problems with charts.

by Aaron Schatz :: Tue, 11/18/2008 - 6:17pm

That wasn't a freaky formatting problem... it was a major issue with the Internet connection at Panera Bread on Route 9 in Framingham. Now fixed.

by glengarry (not verified) :: Wed, 11/19/2008 - 10:22am

FO Boycott!!

by Independent George :: Tue, 11/18/2008 - 6:30pm

Oh, those poor, poor Lions fans: 0-10, with the 2nd hardest schedule in the league coming up.

by White Rose Duelist :: Wed, 11/19/2008 - 10:09am

I don't know - if I were a Lions fan, I would be rooting for 0-16 by now. It seems like the perfect capper for the Millen era (it's not like the balance of the season isn't his fault). And if you're not going to win, you might as well do something memorable in losing.

by Key19 :: Tue, 11/18/2008 - 6:33pm

Fun fact: Dallas is rated #22 after beating the #3 (the Eagles), #5, #8, #11, #23, and #28 teams. They also lost to the #6 team in OT, and lost to the #1 and #32 teams when they had "Dallas Quarterback" at the helm.

Meanwhile, the Eagles are ranked #3 and have beaten #32, #27, #25, #13 and #7. They've also lost to #1, #11, #16, and #22 (Dallas), and tied #28.

Don't those number just seem a bit off to you?

by NewsToTom :: Tue, 11/18/2008 - 6:48pm

I'd feel bad about it if DVOA could reasonably distinguish between "the Cowboys are playing with their backup quarterback and don't feel like trying to win since he sucks so let's get destroyed by a crappy opponent" (STL) and "the Cowboys are playing with their backup quarterback, so let's play pretty good defense so we can beat a good team without our starter" (TB). Would it also be churlish to point out the Eagles are 3rd in the NFL in Points For - Points Against, while the Cowboys are 18th?

No, clearly the Eagles are ranked higher than the Cowboys because Aaron hates Dallas because he departed from his Pats love growing up to root for the Orange Crush only to be disappointed when they got smacked around in the Super Bowl, and therefore makes DVOA hate the Cowboys.

by Jean Sansterre (not verified) :: Tue, 11/18/2008 - 7:01pm

Of course, there's always more to it than that, which I'm sure you're aware. The drawback to the "my team beat teams of the following ranks" perspective is that it ignores the character of the wins.

Dominating the Texans 40-10 speaks very well of you, even if they're a bad team. Barely beating the Packers due to fluketastic factors that are unlikely to be repeated speaks less well for you, even if it would be a victory over the #5 team, not the #26 team.

As Aaron explains quite openly, Philly either blows the snot out of the opposition or loses narrowly. When you blow out another team, you're effectively demonstrating that this win isn't random, that if you played this team 10 more times, you'd probably win 7-9 of them. When you win by three, on the other hand, you're effectively demonstrating that, while things may have come down on your side today, over a ten game stretch it's anyone's guess how things would go. If this were not true, pythagorean ratings would not be so effective.

So yeah, it's not whether you win or lose, it's about how you play the game :)

by vinyltoupee (not verified) :: Tue, 11/18/2008 - 7:17pm

"When you blow out another team, you're effectively demonstrating that this win isn't random ... When you win by three, on the other hand, you're effectively demonstrating that, while things may have come down on your side today, over a ten game stretch it's anyone's guess how things would go. If this were not true, pythagorean ratings would not be so effective. So yeah, it's not whether you win or lose, it's about how you play the game."

Not so fast there pardner. The fact that the Eagles are not very good demonstrates the limits of this theory, as Aaron states above. Killing bad teams and losing to good teams is apparently a bit more complex. For example, maybe the bad team rolls over. Maybe (kinda like a bully) it's easy to have confidence to beat a bad team but harder to believe you can compete with a good team. In spite of what the numbers say, I think there's something to the previous poster's argument that cannot be easily dismissed.

by Jean Sansterre (not verified) :: Tue, 11/18/2008 - 8:18pm

It seems slightly tautological to use "The fact that the Eagles are not very good" as a defense for the position that the Eagles are overrated compared to the Cowboys. The very subject we are implicitly discussing is whether or not the eagles are good, it's slightly cheaty to use your conclusion as evidence of the truth of itself. Anyhow...

You appear to be making a separate argument from what the original poster proposed. His position appeared to be that you could deduce team quality from looking at the quality of team won and lost to. I countered by saying that this position, while true, overlooked the importance of the nature of the win/loss, which is the major reason why the disparity he, quite rightly pointed out.

Your position, as near as I can determine, is that the pythag isn't perfect because a team may have psychological factors corrupting the reliability of the data. This is certainly possible (though hard to prove, since the results are the same whether it is random or psychological.) However, you do not propose that this applies to all such teams, (which would defeat pythag with such teams were it true), you merely propose that it is true with the eagles, and that the possibility in general lends weight to the original poster's position.

Your first allegation may well be true, not being clairvoyant I cannot say what the cause of the Eagles' unusual pattern is. However, I feel as though your tag-team is misplaced, the original poster made no argument regarding causes of defeat/victory, no assertion that Dallas wins with a stout and robust psychology that plays each team equally.

As a rule, pythag tends to hold true, and it certainly has exceptions, as you have pointed out. That, however, doesn't take away from the fact that it is a rule that at least needs to be addressed, even if dismissed due to alleged exceptions to it.

by MJK :: Tue, 11/18/2008 - 9:07pm

Or it could be possible that there is a great deal more randomness in the game than can easily be accounted for, and that trying to determine which of two decent teams is better based on past performance is like trying to figure out whether a carpenter ant is heavier than a bumble bee using only a bathroom scale. It's possible, and some measurment is better than none, but there are a lot of fluky things that could mess with either reading...

And as to the psychological argument, I would counter by saying that some teams may roll over when they know they're bad and facing a good team, but other teams get their hackles up and play hard. Last year, for example, EVERYONE put up practically their best game of the season when they played the Patriots. And still lost.

by vinyltoupee (not verified) :: Tue, 11/18/2008 - 10:55pm

Nice vocabulary Sartre. Never use the simpler word when a more impressive one will do, huh. You must have degrees. I am the Other that defines your argument. (A little philosophy for you.)

In regards to your first paragraph, you lopped off the final part of the sentence "as Aaron states above." To be exact, Aaron stated, "The bad news is that they don't drop any further than that", implying that he agrees that the Eagles aren't very good. I said nothing about the Cowboys. I am making the argument that the Eagles aren't very good. Numbers don't define my viewing experience; they support it. I watch the games, and from what I see, the Eagles aren't very good. That is all.

by Love is like a bottle of gin (not verified) :: Tue, 11/18/2008 - 11:26pm

Actually there is not another word which you can slip in to replace tautological, stop being an idiot. Criticizing someones perfectly sensible sentences is the last refuge of someone who has lost an argument.

by vinyltoupee (not verified) :: Wed, 11/19/2008 - 12:10am

sure there is. Redundant. Would you like to argue as well?

by Eddo :: Wed, 11/19/2008 - 12:39am

That word - I do not think it means what you think it means.

by Vizzini (not verified) :: Wed, 11/19/2008 - 11:59am

Inconceivable!

by VarlosZ :: Wed, 11/19/2008 - 12:41am

"Redundant" means saying something, than saying it again. "Tautological" describes an argument whose conclusion is contained within it's premises (i.e. a circular argument). They're really not close in meaning, and his using "redundant" wouldn't have made any sense. His criticism wasn't that you were repeating yourself, it was that you accepted as a given that which you were trying to prove.

by Scott C :: Wed, 11/19/2008 - 12:46am

Although one of the meanings of tautological is the same as 'redundant' the other is not.

When applied to a logical statement, it has a more complicated meaning.

Because the concept of tautology is more complicated than redundancy, it is both redundant and not redundant, and yet only tautological.

by vinyltoupee (not verified) :: Wed, 11/19/2008 - 12:47am

Yes Scott. Some are describing the logical meaning of "tautological". I am describing its rhetorical meaning. It appears we come from different worlds my friends. See, I wasn't trying to prove anything about the Eagles. I was repeating what the author had stated, as I stated above. Schatz seemingly agrees that the Eagles are rated too highly as he acknowledges the limitations of his own system.

Do you (that's plural you) not realize how seriously you take yourselves, and how much fun it is for me to stick responses in your craw? Just to watch you respond? DVOA is fun and interesting. It is also flawed. The most fun part of this site to me is to try to disagree and watch the logicerati emerge.

by David :: Wed, 11/19/2008 - 4:54am

Parenthetically, I realise that this is feeding the troll, but nonetheless...

Do you (that's plural you) not realize how seriously you take yourselves, and how much fun it is for me to stick responses in your craw? Just to watch you respond? DVOA is fun and interesting. It is also flawed. The most fun part of this site to me is to try to disagree and watch the logicerati emerge.

No, mate - that just makes you a jerk

by Whatev (not verified) :: Wed, 11/19/2008 - 7:21am

There's nothing to realize, because there's nothing novel here. I see this all the time. Guy on an internet forum makes a bad argument, gets called out, realizes he's backed into a corner and decides he'd rather be a jerk than look like an idiot. But claiming you were trolling all along just makes you look like a jerk and an idiot.

Don't do that. You can concede that you're unable to come up with an adequate counterargument without conceding that your position is wrong. Most people will accept that. Trust me. I'm a terrible debater, so I should know.

by glengarry (not verified) :: Wed, 11/19/2008 - 10:32am

... It appears we come from different worlds my friends.

Ok, Senator, please leave your class-warfare garbage at home. It's a new day in America, and we smart folks can use words like tautology now without guilt or fear.

by AnonymousToo! (not verified) :: Wed, 11/19/2008 - 12:03pm

Wouldn't it make more sense to focus on the "logical" meaning of a term when you are discussing the logic of an argument? Just asking.

Though I don't think that was class warfare. It was a claim that he comes from a world of rhetoric and literary genius far more shiny and happy than the dank and cold world of logic where everyoe else here lives.

by Anonymouser (not verified) :: Wed, 11/19/2008 - 4:00pm

So, back to football?

FOOTBALL!!!

by RickD :: Wed, 11/19/2008 - 5:41pm

Redundant does not mean the same thing as tautological.

If I get a second fire extinguisher as a back up, in case the first one fails, how is that an example of tautology?

by armchair journe... :: Wed, 11/19/2008 - 9:16pm

my vinyl-topped friend, please scroll down the next time you look up word definitions. "redundant" is not the meaning being used here, it is the second usage that applies:

http://www.merriam-webster.com/dictionary/tautologous
_______________________________
armchair journeyman quarterback

by armchair journe... :: Wed, 11/19/2008 - 9:19pm

and i would do well to hit refresh before duplicating commentary. ;)

by David C (not verified) :: Tue, 11/18/2008 - 11:08pm

I'm wondering if all these complaints would be occurring if the Cowboys had lost to the Redskins this week.

A lack of injury adjustments is DVOA's major weakness, but the Cowboys case is extreme. The Cowboys arguably went from the best quarterback in the league to the worst quarterback in the league. Meanwhile, their top corner, a starting lineman, the back-up running back, and a key linebacker in the rotation were also out of the lineup. Of course that's going to throw a system that doesn't adjust for injuries out of whack.

On an unrelated note, a lot of teams improved their overall DVOA this week. So hooray for everybody! Except the Hawks, who now suck even more.

by Drew630 (not verified) :: Wed, 11/19/2008 - 1:53pm

While you do offer a good point in the limitations that DVOA has regarding injured starters, I tend to think of injuries in a different light. DVOA is measuring the performance of the playing members of a team, be it superstar QB or third string WR. DVOA shouldn't adjust for the talent level of the players on the field because that is exactly what it is measuring: how the talent performs on the field. If anything, I see that while Dallas has a potent lineup of first string talent, their injury replacements have been lacking when asked to perform and that is reflected in their ranking. That may seem unfair, but it can give a good perspective on the needed depth of the team.

by Rich Conley (not verified) :: Wed, 11/19/2008 - 6:29pm

The issue is whether you want DVOA to predict how good a team actually is.

Saying Dallas hasn't been very good without Romo is true. That doesn't mean that now, with him back, they are the team they were without him. If you take away the games without Romo, you're somewhat overrating the cowboys. If you leave them in, you're somewhat underrating them.

"I see that while Dallas has a potent lineup of first string talent, their injury replacements have been lacking when asked to perform and that is reflected in their ranking."

Thats utterly ridiculous. ANY team that loses a probowl quarterback is going to decline. NO ONE has that depth.

by RickD :: Wed, 11/19/2008 - 5:45pm

"Arguably" would be a key word there. Only a Cowboys homer would say that Romo was the best QB in the league.

by David C (not verified) :: Wed, 11/19/2008 - 10:31pm

Eh, makes for a better argument. Besides, when you think about it, it's not that huge of a stretch. It ain't like Tom Brady's around to automatically get all the gravitas. Romo's 5th in DVOA despite his wide receiving corps (especially Owens) and the offensive line both having a terrible year. Two different quarterbacks (granted, neither were all that great to begin with) played absolutely abysmally in the exact same situation. There's also a lot of parity in QB DVOA right now.

by White Rose Duelist :: Wed, 11/19/2008 - 10:32am

The Cowboys are clearly ranked too low because they beat the #3 (the Eagles), #5, #8, #11, #23, and #28 teams. Ignoring that they had "Dallas Quarterback" at the helm is way better than this. Don't those number just seem a bit off to you?

by Dales :: Tue, 11/18/2008 - 6:54pm

Woot! The Giants no longer have the hardest remaining schedule! Finally!

by NewsToTom :: Tue, 11/18/2008 - 7:01pm

I don't think this has been mentioned, but TEN is thus far on pace for a historically great pass defense DVOA. Looking at the previous years, the 2002 Bucs stand out by a country mile at -55.8%, then you have a clutch of teams in the -34-36% range. Unless I missed somebody, TEN would currently rank 6th best at -33.4% pass D DVOA. It's possible to run on them, particularly when Haynesworth isn't in the game, but they've really put the clamps on opposing passers this year.

by Lou :: Tue, 11/18/2008 - 7:51pm

can you imagine if they still had pacman?

by Verified (not verified) :: Tue, 11/18/2008 - 9:24pm

Seeing as how he's suspended again...I imagine they'd be 10-0.

by RickD :: Wed, 11/19/2008 - 5:53pm

Indeed, if they still had Pacman, they would have probably released him after his latest incident, and they would still not have Pacman.

by jimm (not verified) :: Tue, 11/18/2008 - 7:06pm

As a Viking fan I can honestly say this team is dull dull dull.

for what it's worth - The Vikings were a far better offensive team in 2007 than they have been in 2008. The only changes - Berrian replaced Wade as number 1 receiver and Frerotte replaced Jackson at QB.

Looking back now I think it was a bad move. Jackson lost to GB on the road and he made one horrible pass but he kept them in the game with his scrambling to that point. Losing to GB on the road according to DVOA isn't exactly horrible. Then they lose to Indy which it turns out is a good team. If Shiancoe could catch that game probably turns out to be a win.

This offence can't afford sacks - Frerotte takes far more than Jackson and as it turns out Frerotte makes more mistakes (ints) as well.

I'm not sure I buy Frerotte has been better than what you could have expected from Jackson.

by Tundrapat (not verified) :: Tue, 11/18/2008 - 8:26pm

That being said, Jimm -

I'd still take Thigpen (no, not 'Yancy') over either Frerotte OR Jackson. Neither is the answer.

For that matter, I'd also prefer a coach who recognizes that giving the ball to your Pro Bowl running back on 2nd-and-2 is a good idea. Even late in the 4th quarter.

Or a coach who realizes that special teams contribute at least 15-20% of the snaps in a game, so you can't just ignore their horrendous production.

by zlionsfan :: Tue, 11/18/2008 - 9:48pm

Dull? Sounds like a great deal. Where can I get some dull?

by Love is like a bottle of gin (not verified) :: Tue, 11/18/2008 - 7:15pm

If the system didn't give us output we were surprised by it wouldn't be a very valuable system would it?

by Fizzman :: Wed, 11/19/2008 - 11:41am

Hear, hear! DVOA is sometimes "obviously" wrong, and is by no means perfect. But it is often right about non-obvious things, too. If it just parrots back what we already know from win-loss record and scoring differential, then it is a colossal waste of time - but it doesn't.

by Key19 :: Tue, 11/18/2008 - 7:17pm

The Eagles get a lot of their stats from blowing out the three horrible NFC West teams they've faced. Otherwise, their point differential is nothing special.

My point is this: The Eagles have not beaten very many good teams, and have lost to most good teams they've played. Dallas meanwhile, has beaten (when healthy) nearly every good team they've played (the Cardinals OT thing was partially healthy, partially not since the broken finger came in the beginning of OT). They lost to Washington in a very close game and now have evened the score with them. I just don't see what else Dallas has to do to climb the rankings or what else Philly has to do to go down. Being 0-3 in the Division should be a good indicator of what the Eagles are. They lose to good teams and blow out bad ones. End of story.

by Love is like a bottle of gin (not verified) :: Tue, 11/18/2008 - 7:27pm

DVOA is not a merit based system. It is a mathematical model. The cream does not always rise to the top. It is just a simulation with some well understood weaknesses (not knowing that the huge drop in DAL QB play was irrelevant (maybe)) is one of them).

Life isn't fair and football seasons with tiny samples are even less fair. Personally I have no idea what Dallas fans are upset about? Wouldn't you rather be 6-4 and have negative indicators from mathematical models than 5-4-1 and have positive ones?

by MJK :: Tue, 11/18/2008 - 9:10pm

It's not even really a simulation. It's an accounting method. It's not actual propegating anything forward according to as set of governing laws--it's just counting how many times certain very carefully chosen events happened for each team, and comparing them to how often they happen on average. It's a method for gathering data that could be used in a simulation (like the playoff odds report), but not a simulation in its own right.

by Love is like a bottle of gin (not verified) :: Tue, 11/18/2008 - 9:52pm

I know it is not a simulation, I think I was just trying not to use mathematical model two sentences in a row and that seemed a good synonym....also I am very tired :(

by glengarry (not verified) :: Wed, 11/19/2008 - 10:38am

I don't see how Dallas, or any other team, should be granted fillips or caveats because their QB got hurt. Injuries are a part of football. Dallas had a terrible QB2. Their fault. End of.

Now, for freak outside injuries / boneheadedness (R. Collier, M. Vick) -- that sort of thing is at least worth an asterisk somewhere.

by Rich Conley (not verified) :: Wed, 11/19/2008 - 6:33pm

EVERY TEAM in the NFL has a bad to terrible QB2. If they weren't bad, they'd be QB1 somewhere.

by Alex51 :: Thu, 11/20/2008 - 10:30pm

EVERY TEAM in the NFL has a bad to terrible QB2. If they weren't bad, they'd be QB1 somewhere.

The 2006 Eagles, 2000 Rams, and late 1980s 49ers would like to have a word with you.

by AndyE :: Fri, 11/21/2008 - 12:07am

Heck, the first 10 minutes of the 2008 Patriots might end up having a word with you.

Or maybe not.

by Jean Sansterre (not verified) :: Tue, 11/18/2008 - 8:32pm

But you are mistaking the principle purpose of ratings such as DVOA. DVOA does not tell you who won and who lost, the standings tell you that quite easily. The purpose of DVOA (and much of statistics) is to extract the predictive data, information that is reliable to draw conclusions from. To appreciate DVOA, you must fundamentally respect the fact that the final score is only partially related to what the game says about the teams. If a team wins by 3 because they recover all of the 6 fumbles that occur, that goes down as a win for them, but the fumble recoveries mean next to nothing about how good that team is (or how good the team that lost is.) DVOA will look at that game and say flat out that the winning team was worse than the losing team (probably) because there were huge luck factors that helped the winning team. To say that a team's DVOA is low simply because of mere wins and losses is to miss the purpose of the system in the first place.

As a matter of further principal, you probably ought not use three games as the bedrock for your argument "Being 0-3 in the Division should be a good indicator of what the Eagles are." I mean, the sample size is so small that it's not entirely reliable. Come back when they're 0-6 in the division.

by vinyltoupee (not verified) :: Tue, 11/18/2008 - 11:00pm

Since we're having so much fun, I'd like to comment on this one too: "The purpose of DVOA (and much of statistics) is to extract the predictive data, information that is reliable to draw conclusions from." I've never seen a mission statement from FO on DVOA. Is this really true or your assumption? If true, it does a pretty poor job, as multiple times, DVOA has been shown to be about as consistent as Pete Prisco in its predictive qualities.

I'd venture it's main purpose is to make money for its creators. At that, it does a great job. Here we are.

That should give you something to reply to.

by Love is like a bottle of gin (not verified) :: Wed, 11/19/2008 - 10:05am

Thank You Aristotle.

by Love is like a bottle of gin (not verified) :: Wed, 11/19/2008 - 10:05am

Thank You Aristotle.

by jonnyblazin :: Wed, 11/19/2008 - 1:31pm

Viny,
I'm in 2 office pools, one picking against the spread, one is a confidence pool. Each pool has over 200 people in it. I use DVOA and injury reports to make my picks. I'm first in both pools. From my perspective, DVOA is pretty damn useful since I don't have time to watch all 32 teams every week. So if you have a shred of intelligence you can use DVOA to make $ for yourself as well.

by Tundrapat (not verified) :: Wed, 11/19/2008 - 2:01pm

I think you lost him at the 'shred of intelligence' part...

by The Ninjalectual :: Wed, 11/19/2008 - 2:53pm

But you had him at $

by Boston Dan :: Wed, 11/19/2008 - 4:32pm

Having won two pools using DVOA, injury and weather reports last season.

Currently in third place in both pools this season having missed an entire week due to general drunkenness at a wedding. And thanks to KUBIAK, DVOA, injury and weather reports, having the best fantasy football team in a big league.

I am a DVOA believer.

by RickD :: Wed, 11/19/2008 - 5:55pm

"DVOA has been shown to be about as consistent as Pete Prisco in its predictive qualities."

Oh?

And where has this been "shown"?

I'm aware of several cases where the predictive powers of DVOA have been amply demonstrated.

by Tundrapat (not verified) :: Tue, 11/18/2008 - 9:07pm

You're also falsifying excuses for the losses. The Cowboys were 10 points behind in the Arizona game, which they were barely able to cover to force the overtime (in which Romo's pinkie finger got hurt). They covered to get to OT, then lost on a special teams play.

And when we look at 'Special Teams DVOA'...oh, look at that! The only team worse than the Cowboys is my own Vikings!

Welcome to the wonderful world of losing on return TDs. Rather than talk up your Cowboys here, perhaps you should email Wade Phillips and let him know that 15-20% of a game (special teams) is a large chunk and needs some coaching.

In short - to 'climb the DVOA rankings', what Dallas needs to do is produce sustained drives that don't occur in garbage time. And beef up their special teams.

by Hummingbird Cyborg :: Tue, 11/18/2008 - 7:21pm

Philly's estimated wins according to DVOA is 6.7? Or just below seven wins, right?

That means that if they were 5-5, they'd be just two games lower than what we'd expect.

So, this is a case of the discrete nature of wins and losses working slightly against them.

Is this so different than the Denver Broncos who have 4.8 expected wins right now, but are 6-4?

And with so many teams being compared, don't you expect random chance to show a team or two to have different records than you'd expect even while most teams are about where you'd expect?

Maybe there is something to it happening to the Eagles more than other teams, but I think that the perception is less caused by a problem with the system as it is caused by the ramifications of a high profile team or two outperforming their rating all while another team underperforms theirs.

A slight underperformance by record combined with a slight overperformance combines to create a percieved substantial difference when the difference is really minute.

by RickD :: Wed, 11/19/2008 - 5:58pm

You could make this argument even stronger by saying a tie is "half a win", in which case the Eagles have 5.5/6.4 wins.

The reason people think the Eagles are overrated is because they have shown a knack for losing close games, and because they were just tied by the freakin' Bengals in a game where the Eagles generally looked like crap.

Perhaps the aspects of game-day coaching that Reid is terrible at (clock management, for example), simply are not measured by DVOA.

by young curmudgeon (not verified) :: Tue, 11/18/2008 - 7:28pm

"When you blow out another team, you're effectively demonstrating that this win isn't random, that if you played this team 10 more times, you'd probably win 7-9 of them."

Is this really the case, and is this really what DVOA wants to hang its hat on? I don't have the data (translation: too lazy to do the study that wouldn't even be that hard), but we do have plenty of examples of teams playing each other twice in one season. It would be interesting to see how many teams blew out their opponents, then lost fairly convincingly to them later in the year (Wasn't there a Bills 31 Pats 0, followed by a Pats 31 Bills 0 a few years back or something like that? I know that's anecdotal evidence, but it does happen.) Blowouts aren't necessarily solid demonstrations of superiority, sometimes even a good team lets a game just get out of hand. ("Momentum, blah blah blah")

On the other hand, there are definitely emotional/attitudinal factors involved. I honestly believe that the Giants good showing against the Pats last year in the regular season made them more confident in the super bowl. It's an interesting question.

by DaveInTucson :: Tue, 11/18/2008 - 8:49pm

I just ran a quick analysis of season splits where the point difference in both games is at least 14 points. Turns out it happens 3 or 4 times every year.

Check that. I just realized I had a bug in my logic. It generally happens once or twice a season (though for some reason, it happened 7 times in 1996).

Here's what I got for 2004-2007:

2004:
BAL 3, CLE 20 / CLE 13, BAL 27

2005:
MIN 3, CHI 28 / CHI 10, MIN 34
WAS 0, NYG 36 / NYG 20, WAS 35

2006:
CHI 26, GB 0 / GB 26, CHI 7

2007:
HOU 17, JAC 37 / JAC 28, HOU 42
KC 30, SD 16 / SD 24, KC 10

There haven't been any such splits so far in 2008.

ETA: The BUF 31, NE 0 / NE 31, BUF 0 split was in 2003.

I have a blog where I post objective, predicive power rankings of NFL teams.

by chris clark (not verified) :: Tue, 11/18/2008 - 8:52pm

Yes, blowouts both ways do occasionally happen. But, you need to read the article on DOMS, GUTS, and SKATES, which shows that big margins of victories do tend to indicate something, especially in predicting teams winning the SB. However, that's not all DVOA stats hang their hat on. DVOA is designed to measure which teams win by consistently above average play and discounting plays that tend to be more unquantifiable, e.g. very long runs from scrimmage. The team gets credit for the long run, but not as much as the team would from a series of above average runs yielding the same result. The theory is that if you always do above average, you will continue to do above average. If you sometimes do something spectacular, you may do it again, but probably cannot expect it on a regular basis. Of course in the nature of a football game, a spectacular play may be all that needed in a close game to transform a loss to a victory. And with only 10 games having been played, two victories could turn a 4-6 team into a 6-4 one (or vice versa). Thus, if you want rankings that score teams based on wins/losses, there are places to find them (www.beatpaths.com for instance). Use DVOA for what it is good for. The authors of this site spend plenty of time trying to determine ways to make their scale better. However, it isn't likely to diverge much from what it is now, as it captures what it is intended to capture pretty well (an objective appraisal of a teams strength based upon available measurements, the play-by-play summaries).

by Rich Conley (not verified) :: Wed, 11/19/2008 - 6:51pm

"Yes, blowouts both ways do occasionally happen. But, you need to read the article on DOMS, GUTS, and SKATES, which shows that big margins of victories do tend to indicate something, especially in predicting teams winning the SB"

No, it really doesn't. Its on the right track, but its got some huge logical fallacies.

by MJK :: Tue, 11/18/2008 - 9:23pm

I would assume what the original poster meant when he said "if you played this team 10 more times...", he actually meant "if this exact game happened in 10 identical parallel universes, you'd win in 7-9 of them".

Obviously over the course of the season things change, which can shift the balance of power. The biggest example of this is, of course, injuries--e.g. playing the Colts without Bob Sanders or Jeff Saturday is infinitely easier than playing them with. The Pats got blown out by the Steelers the year they went 15-1, but dressed ONE RB that game, and were also low on CB's, due to injuries. When they met in the playoffs that year, they were healthier and the outcome was different. There's also schematic things that change...a team will come up with something new (e.g. the Wildcat, Kordell "Slash" Stewart, the Pats crazy spread offense from last year) that will take other teams a few games to figure out. In the Bills-Pats example that you mention, they played in Weeks 1 and 17. In Week 1, Laywer Milloy had been released by the Pats and gone to the Bills a few days before the game, taking all his knowledge of the Pats defensive formations, offensive preparations, and locker room presence with him, and hence the Bills blowout. That game could have been played in 10 parallel universes, and the Bills probably win in all 10 of them, despite the fact that the Pats were a deserved 14-2 that season and the Bills much weaker (don't remember at the moment).

This of course highlights a huge problem with DVOA, along with EVERY other objective ranking methodology: you MUST assume that the team you're ranking has some reasonably constant value of "goodness" that you're trying to measure. You can't take into acount changes over time, or at least you can only take into accoun