Writers of Pro Football Prospectus 2008

Most Recent FO Features


» Futures: Maurice Hurst

A heart condition discovered at the combine has put the Michigan lineman's career in limbo, but Hurst had the best film of any defensive tackle in this year's draft class.

17 Jan 2006

Playoff Power Rankings II

by Aaron Schatz

This year, we are continuing with DVOA into the playoffs, so that we can continue to do the FOXSports.com power rankings. Those rankings with extended commentary are found here. The extended commentary this week includes a discussion of why a playoff loss doesn't knock Indianapolis from the top spot, and what Patriots fans have in common with Woody Allen.

Here's what you should know about the DVOA table below:

  • These numbers are weighted DVOA. That means that Weeks 1-5 are not included, while Weeks 6-11 are somewhat discounted.
  • The exception is the shaded column marked TOTAL DVOA, which represents DVOA with all 19 weeks equal.
  • However, unlike the standard DVOA numbers on this website, these numbers do not include certain portions of Week 17 games outlined in the FOXSports.com commentary two weeks ago. (These plays are not accounted for in the TOTAL DVOA column either.)
  • Teams which did not play in the playoffs are treated as if they had two bye weeks.

Last week someone asked for all 32 teams instead of just the playoff teams, to see what they looked like with the weights in weighted DVOA moved further along. So below you'll find all 32 teams. This is the last power rankings of the year. Starting in February we'll be doing monthly off-season rankings which will be interesting, because they will be the first Football Outsiders rankings ever based on our subjective opinions. Each month's rankings will come from a poll of the FO staff. (We're saving the statistical projections for the book.)

One more thing: We think we can solve the server overload thing next week, so bear with us.

* * * * *

To save people some time, we remind everyone to put their angry troll hatred into the official zlionsfan angry troll hatred Mad Libs form:

<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>

I'm not going to bother to run the whole DVOA explanation; if you are new to the website, you can read about it here. As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.

1 IND 39.3% 1 14-3 38.3% 1 33.7% 1 -8.6% 11 -2.9% 27
2 DEN 38.0% 2 14-3 33.9% 2 24.7% 4 -11.2% 7 2.1% 9
3 CAR 32.8% 5 13-5 21.7% 9 5.4% 9 -20.7% 3 6.7% 5
4 KC 32.5% 3 10-6 23.9% 7 26.0% 2 -9.5% 9 -2.9% 28
5 SEA 31.2% 6 14-3 29.7% 4 25.1% 3 -4.5% 14 1.6% 12
6 PIT 30.2% 4 13-5 30.2% 3 12.8% 7 -16.1% 4 1.2% 14
7 WAS 20.8% 7 11-7 22.3% 8 -5.4% 15 -23.3% 1 2.9% 8
8 SD 20.2% 8 9-7 26.4% 6 17.6% 6 -3.2% 16 -0.5% 21
9 CIN 17.0% 9 11-6 28.0% 5 23.3% 5 6.7% 24 0.4% 16
10 NE 15.7% 10 11-7 9.2% 13 12.7% 8 -1.5% 18 1.4% 13
11 NYG 13.9% 12 11-6 17.8% 11 0.5% 12 -13.2% 5 0.2% 17
12 MIN 9.8% 13 9-7 -6.9% 20 0.0% 13 -7.9% 13 1.8% 11
13 JAC 9.0% 14 12-5 19.7% 10 4.7% 10 -3.2% 15 1.1% 15
14 CHI 7.2% 11 11-6 11.1% 12 -13.5% 25 -21.8% 2 -1.1% 23
15 TB 3.9% 15 11-6 5.1% 14 -4.8% 14 -9.3% 10 -0.5% 20
16 BAL 3.4% 16 6-10 -6.1% 19 -11.6% 23 -13.2% 6 1.8% 10
17 MIA 3.3% 17 9-7 2.8% 15 -12.7% 24 -8.4% 12 7.7% 3
18 OAK -10.9% 18 4-12 -4.3% 18 -5.6% 16 -0.5% 20 -5.8% 31
19 ARI -11.3% 21 5-11 -16.5% 23 -7.7% 17 -2.1% 17 -5.7% 30
20 ATL -11.5% 22 8-8 -6.9% 21 4.0% 11 15.2% 30 -0.3% 19
21 PHI -12.1% 20 6-10 -3.0% 17 -29.7% 31 -9.8% 8 7.8% 2
22 DAL -12.8% 19 9-7 1.7% 16 -11.0% 21 -1.0% 19 -2.7% 26
23 CLE -17.6% 24 6-10 -15.4% 22 -17.7% 29 5.3% 22 5.4% 6
24 HOU -18.3% 26 2-14 -33.3% 31 -8.0% 19 19.6% 31 9.2% 1
25 NYJ -20.0% 25 4-12 -21.6% 27 -18.4% 30 0.1% 21 -1.5% 25
26 DET -21.7% 23 5-11 -19.1% 24 -7.9% 18 10.8% 26 -3.0% 29
27 STL -21.7% 27 6-10 -23.0% 28 -11.5% 22 9.5% 25 -0.6% 22
28 BUF -23.2% 29 5-11 -19.2% 25 -15.4% 27 14.5% 29 6.7% 4
29 NO -26.7% 31 3-13 -28.2% 30 -14.3% 26 11.1% 27 -1.3% 24
30 TEN -28.3% 28 4-12 -20.6% 26 -9.0% 20 22.4% 32 3.1% 7
31 GB -30.9% 30 4-12 -23.7% 29 -16.3% 28 6.2% 23 -8.4% 32
32 SF -54.8% 32 4-12 -65.9% 32 -43.2% 32 11.7% 28 0.0% 18

Here are the one-game DVOA ratings for the first round of the playoffs. As pointed out elsewhere, yes, DVOA believes that the Colts outplayed the Steelers, although it wouldn't have the Colts higher if you included the Polamalu interception and the rest of that Indianapolis drive.

SEA 34.8% 14.7% -22.4% -2.3%
WAS 15.3% -12.7% -25.0% 3.0%
DEN 32.4% 0.7% -20.8% 10.8%
NE 4.1% 3.2% -11.7% -10.7%
IND 43.5% 37.9% -8.9% -3.3%
PIT 29.7% 11.9% -16.7% 1.1%
CAR 46.6% 37.1% -1.3% 8.2%
CHI -24.5% -5.8% 15.8% -2.9%

Posted by: Aaron Schatz on 17 Jan 2006

121 comments, Last at 07 Feb 2006, 6:26pm by mike


by Crushinator (not verified) :: Tue, 01/17/2006 - 9:09pm

The Panthers aren't taking any prisoners these last few weeks, yeesh. They made the Falcons look like a pop warner team, they made the Giants look like the Falcons, and they just shredded the number 1 defense in the NFL on the road.

Lost in all the "Colts lost to the Steelers" talk, theres a Panthers team that DVOA-wise has outperformed every team in the playoffs two weeks running.

by Mshray (not verified) :: Tue, 01/17/2006 - 9:17pm

Pretty hard to believe that with all those muffed returns that the Seahawks had only the 4th worst ST DVOA last weekend. Is a missed FG really that much worse than a fumbled return?

by Crushinator (not verified) :: Tue, 01/17/2006 - 9:21pm

Re: The Bears scoring 21 points on the vaunted Panthers D

I've generally been of the frame of mind that when a QB change is made in game or a QB who hasn't played in a while comes in, his first game or two he'll play signifigantly better than his ability.

I think this has to do with film study. Opposing defenses spend all week planning for a certain QB and his nuances, signals, strengths and weaknesses and then another player comes in.

Rex Grossman is better than Kyle Orton, but there are 6 quarters of gametape of Rex Grossman from this year, and I think 4 games from the beginning of last year. That's very little information to go on - especially as players are constantly improving. The Panthers just didn't seem to know how Rex Grossman would play and I think that had more to do with him putting up 21 points than the Carolina Panthers D suddenly collapsing. Film study is a huge part of game prep.

Luckily, theres a lot of game tape of Matt Hasselbeck.

by Pio (not verified) :: Tue, 01/17/2006 - 9:28pm

What worries me as a Broncos fan is that without special teams our DVOA for the game comes down to barely above 20%, a good bit less than that of the Steelers. Most of that ST DVOA is probably from the two forced turnovers, neither one of which is likely to be repeated.

by thad (not verified) :: Tue, 01/17/2006 - 9:32pm

Good God what hapened to the Falcons defense?
Did anybody mention the fact that their defense was about to suffer a total meltdown in that shot to hell thread.?
I sure didn't see it coming.

by NF (not verified) :: Tue, 01/17/2006 - 9:39pm

I think Jerome Bettis gives the Steelers the edge in breaking out big runs. Why? The mere sight of him barreling towards them will cause the brains of the Denver defensive backs to freeze up. Or they'll mistake him for an offensive lineman.

by thad (not verified) :: Tue, 01/17/2006 - 9:42pm

re 4
Yeah without those fumble recoveries Denver is probably doomed.
Nah I am just messing with you. I thought their defense played very well against a top notch offense.

by Luz (not verified) :: Tue, 01/17/2006 - 9:50pm


ya, i've always felt the same way. when a new QB comes in during a game they usually play great. then when the start the next game they play a little worse... then a little worse the next game... and then the next game everyone realizes why the guys a backup.

a good example of that was the fitzpatrick kid from the rams this year. he came off the bench and lit up the texans (ok, not that impressive) and the peter kings of the world start firing up their articles. four weeks later the peter kings of the world are tearing the same kid apart, wondering "what happened?"

by Larry R. (not verified) :: Tue, 01/17/2006 - 10:00pm

Aaron & FO are so far off base with there power rankings they are worse than the officiating this weekend. Come on give me a break I was beginning to start thinking your power rankings have some merit! But now it’s ranking them on the team you thought should win not who was best. The Steelers beat the Colts in every part of the game! But best of all I am going to laugh my butt off when the Seahawks beat the Panthers. Go Hawks!

by andrew (not verified) :: Tue, 01/17/2006 - 10:01pm

Is this still without select games from the final weekend?

by Björn (not verified) :: Tue, 01/17/2006 - 10:13pm

Larry R. is clearly ranked too low because he is clearly very intelligent and understands the objective nature of DVOA. Comparing who has more hair on their ass is way better than this. Derpa derpa derpadee doo.

by DGL (not verified) :: Tue, 01/17/2006 - 10:28pm

Y'know, it's getting to the point where I can't tell the difference between the trolls and the troll-satires.

Good job, everyone. Smoke 'em if you got 'em.

by Larry R. (not verified) :: Tue, 01/17/2006 - 10:29pm

11.:: Björn ....You must be a Colts fan LOL

by Andrew (not verified) :: Tue, 01/17/2006 - 10:31pm

Re: Offensive collpase and Special Teams rebirth of Eagles in last half of season.

Wow! Just wow! Holy crap!

-29.7% offensive DVOA? That's like a 50-60 point swing from last season. Isn't that equivalent to 12 points per game or more?

I mean I knew McMahon was really bad, but wow! Was he one of the worst ever?

And is the importance of Akers to Eagles Special Teams any clearer than their sinking to #32 with him out and return to the top with him back?

by putnamp (not verified) :: Tue, 01/17/2006 - 10:32pm

Larry R., I'm a Seahawks fan, and am deeply saddened by the low quality of each comment of yours I've read.

by Paytonrules (not verified) :: Tue, 01/17/2006 - 10:34pm

Larry R also clearly understands that Seattle is at home, and given the generally accepted 17% home "fudge factor" the favorites this weekend - unless you are Don Banks.

by Don Banks (not verified) :: Tue, 01/17/2006 - 10:54pm

#16--Say what now? Of course, if the Seahawks had two quality playoff wins like the Panthers I'd think more of them. I'd say more, but I have to go talk to Peter King about whether or not the Patriots should be #1 in the Fine 15 after this Sunday, regardless of what actually happens in the games.

by ToxikFetus (not verified) :: Tue, 01/17/2006 - 10:59pm

Re 14:

Holy Crap! Does someone have a week-to-week graph of the Eagles special teams DVOA? From what I remember, their early season ST DVOA was orders of magnitude worse than anyone else.

by thad (not verified) :: Tue, 01/17/2006 - 11:01pm

You heard it here first:

Carolina 33
Chicago 31

With every single prognosticator saying 6-3 it has to be a shootout right?


:: Clod — 1/13/2006
See Larry R,
that is a good prediction.
So before you take DVOA to the woodshed maybe we could hear your take on the other game.

by Harry (not verified) :: Tue, 01/17/2006 - 11:06pm


You sure have a warped view of Pats fans. The freaks who e-mail you are not at all representative of most fans. As a resident of Newton, MA I feel pretty comfortable saying that most Pats fans I know are still thrilled with the team and have enjoyed every step of the ride. Most Patriots fans don't spend a lot of time worrying about the national media or "respect". The Denver loss was hard to take more for the way they lost than the fact that they finally lost (everyone knew a 3-peat was going to be insanely difficult). Even on sports radio I haven't heard much grumbling, mostly people are excited that the Pats still have such a strong core to build on for next year. Can't get any enjoyment out of it? Maybe you're thinking of Red Sox fans.

by Ron Mexico (not verified) :: Tue, 01/17/2006 - 11:10pm


Of course, actually reading the explanation for why the ratings are the way they are might help....

But who wants to do that? It's easier to just spout off random nonsense...

by the K (not verified) :: Tue, 01/17/2006 - 11:22pm

Indianapolis is clearly ranked too high because Peyton Manning is a choke artist. Skip Bayless' endless wisdom is clearly better than this, and OMG THE STEELERS GOT SCREWED AND STILL WON LOL!!!!! PEYTON U SUCK!!!!!!!

by rk (not verified) :: Tue, 01/17/2006 - 11:24pm

Re: 14
I think part of the rise in ST DVOA for the Eagles is the fact that it is weighted. So I think Weeks 1-5 don't count at all. That means quite a bit of crappy football is off the books.

by Return of the son of the FO Message Board Curse (not verified) :: Tue, 01/17/2006 - 11:35pm

I'd like to remind everyone with these new rankings out to PHE4R M3!@!#!@. Seriously, enjoy the rankings, don't take them too seriously or else. And that else is Atlanta missing consecutive winning seasons for the first time in franchise history. Think about this before you post guys...long and hard. 6-2 to 2-6...it can happen to you.

by kleph (not verified) :: Tue, 01/17/2006 - 11:38pm

i hope to goodness there is a gratis copy of the next PFP for zlionsfan. his angry troll hatred Mad Libs form is the greatest innovation to this site barring only DVOA itself.

by MAW (not verified) :: Tue, 01/17/2006 - 11:42pm

Last week someone asked for all 32 teams instead of just the playoff teams, to see what they looked like with the weights in weighted DVOA moved further along. So below you’ll find all 32 teams.

That was me, thank you.

by Catfish (not verified) :: Wed, 01/18/2006 - 12:22am

Re: 20 Aaron lives in Massachusetts and is a Pats fan himself, so I think he knows what you're talking about. Keep in mind, though, that he also has to wade through tons of emails from idiotic angry fans of {insert "snubbed" team here}. I'd imagine that would shift anyone's perception of a group of fans.

by Nolan (not verified) :: Wed, 01/18/2006 - 12:31am

Aaron I think you are a total IDIOT let me show you why, Here you explain why a team who lost 3 out of the last 4 games of the season is still the best team in the league:
"And yet, they still sit atop the FOXSports.com power rankings. Why?
The answer is fairly simple: our statistical system takes a much longer-term view of the season when judging which team is the best"
and then here is your lame brain explanation of why caralina is ranked so high:
"Seattle ranks below Carolina because the Panthers were so good against the Giants"
Well bone head which is it? I have-been saying that this system of yours is a joke ever since you had to cheat the numbers to get Indy moved up the power rankings back in week 6 (I think) and you said by the end of the season we would see that the numbers would validate this!!!!!! well it looks like Indy was no better than 6th or maybee eighth after all they had to have the NFL officials do everything they could to try and stay with the Steelers... So if after this ling rant you forget what my point is let me refresh your memory AARON IS A TOTAL IDIOT!!!!!!!!!

by Terry (not verified) :: Wed, 01/18/2006 - 12:46am

So since the NFL said they were wrong on the Polamalu interception, will it be factored into DVOA?

by Larry R. (not verified) :: Wed, 01/18/2006 - 12:56am

15.putnamp: At least I was nice with my comments on the power rankings.....

by Shelley (not verified) :: Wed, 01/18/2006 - 12:57am

Thanks, Aaron, for the rational media (do you count as "media?") view of the Pats. I agree with Harry that most fans are taking this loss in stride, but sports pages are trying to turn this into the Patriots Apocalypse. The Pats failed to do what no team has done before, in an era where forces are specifically aligned against such as feat, and they're suddenly a bunch of failures and has-beens? I don't see it. I also don't see the departure of Mangini causing a defensive collapse (any more than the departure of Crennel did--injuries did that)--in fact, I think Dean Pees might have been a better choice of DC in the first place. And to look on the bright side, this year Belichick is only two weeks behind on draft preparations, instead of six.

by Comrade Jason (not verified) :: Wed, 01/18/2006 - 1:00am

Uh ... #28 is a joke, right? With spelling that bad, I'm just assuming it has to be.

That being said, I know Carolina (or "caralina," if you live in a part of the world where your first cousin is a possible prom date) is ranked above Seattle in weighted DVOA, but the difference is only 1.6%. Is that even enough to be statistically significant, or should it just be considered a dead-heat?

by Jon M (not verified) :: Wed, 01/18/2006 - 1:01am

Re: #28
Wow, I was almost at the end of reading Nolan's post and wasn't sure whether I agreed with him about Aaron being an idiot, but then he tacked on that ninth exclamation point, and now I'm sold. If it had just been seven or eight I would've had to say that Aaron's not an idiot, but how can you argue with nine? And in capital letters, no less. Very persuasive. I think Fox has hired the wrong guy.

by Ruben (not verified) :: Wed, 01/18/2006 - 1:11am

Rock on, Aaron.

One thing that continues to concern me, though, about Carolina: They REALLY use Steve Smith to return punts and kicks...I mean, I thought maybe I was just seeing things in the few highlight reels I observed during the season, but...wow.

I mean, if he's their ONLY offensive weapon, how can they risk him with one of the (equally) unproductive and high-risk duty of punt returning?

What, did they cut He Hate Me this season?

by RowdyRoddyPiper (not verified) :: Wed, 01/18/2006 - 1:13am

Uhhh...Comrade Jason, they do know how to spell Carolina in the southeast, two of the states have it in their name, it's necessary.

With that out of the way, let's start to guess the team of Trolls that don't specifically state where their allegiance lies. Nolan presents a tricky case...he mentions the Colts not being able to keep up with the Steelers and having a higher DVOA as ridiculous. Could he be a Steelers fan?? I am praying to the FSM that he isn't....please oh noodly one, don't let him be a Stiller fan. Based on his unwillingness to learn the proper spelling of this Week's opponent I put my money on a Seattle fan. 147 straight days of rain does strange things to a man's though process and spelling.

by RowdyRoddyPiper (not verified) :: Wed, 01/18/2006 - 1:21am

Ruben, I agree on the PR duties for Steve Smith, thankfully they do not have him on KR duties...yet. I mean seriously, your only offensive weapon trying to field a punt while defenders rush unmolested toward him...yikes!

by Mike (not verified) :: Wed, 01/18/2006 - 1:34am

Punt returns are not nearly as dangerous as kicks. Punt returners have the oppurtunity to fair catch any kick or not return it at all. Steve Smith does a great job of not taking chances with punts. He only returns the ones where he has enough space to work with, and they also spell him with Chris Gamble returning punts from time-to-time.

And considering Smith is 4th in the NFL in punt return average, I'd say it is worth having him back there.

by Kibbles (not verified) :: Wed, 01/18/2006 - 1:35am

Re #28: Thanks for refreshing my memory, there.

Anyway, can we put to rest this whole "The Broncos didn't beat the Patriots, the Patriots beat themseleves" arguement? Denver finished with a 28.3% advantage in DVOA, despite playing a weaker opponent (New England got a bigger DVOA-boost from the opposition adjustment). Yeah, it's obvious that Denver CLEARLY didn't deserve to win that game.

Also, Denver's special teams DVOA went from 21st to 9th... in just one game. That has to be some kind of record.

by someone (not verified) :: Wed, 01/18/2006 - 1:38am

Re. your extended commentary and the Pats dynasty. The Pats may remain "a favourite" for 2006 but only insofar as at least 8 other teams in the AFC have the same shot at a 10-6 record. I think the fact is the situation has markedly deteriorated for the Pats in terms of their relative competitiveness against the rest of the conference in the last two years. Their 2001-4 run came with their two elite, franchise players on boosted rookie contracts, not to mention generally more cap room than they currently have, allowing them the flexibility to perform the free agency manoeuvres etc that they have done so well. The fact is however, Brady is now getting paid at $10m+ pa, and Seymour will want the largest DL contract in league history soon. This suggests the Pats are going to have to be even more skilled at the draft and free agency than they have been during the run, in a time where other teams are in a building cap situation. Deals like Corey Dillon worked fantastically in the very short term but have only just started to bite the Pats in the ass contract versus production wise, and there's no contributing second round pick on the roster as a result of a deal like that. Pats are still the favourite for the AFC East title next year and with their experience and Brady, yes they have a shot, but I can't get away from the impression that they are now in decline as a competitive force on a relative basis. I doubt they have the cap room to fix the obvious problems they have - defensive secondary, tackles, running game - that led to them flattering to deceive somewhat this year. Its just the way it goes. No-one is immune to the cap.

by Stevie (not verified) :: Wed, 01/18/2006 - 1:49am

Personally I find the troll satires hilarious. I LMAO!!!!

by Pat (not verified) :: Wed, 01/18/2006 - 1:55am

For those who remember, from Week 17,

so a team with below-average special teams has never won the Super Bowl since our measurements started in 1998. The Colts, Broncos, or Bears could be the first, but next year's best teams will probably be back to having good special teams.

Well, now, a team with below-average special teams will still never have won the Super Bowl (Denver was marginal - likely if you include the playoffs, even non-weighted, they'll be above average). Fire Mike Vanderjagt!

Seriously, though, I do wonder. I've always thought that special teams becomes increasingly important when you've got evenly matched teams, because a poor kickoff/punt is just free yardage, and a missed field goal is just a free turnover.

by murray (not verified) :: Wed, 01/18/2006 - 2:19am

Some questions:

1. If Bettis had not fumbled late in the fourth quarter of the Pitt/Indy game, would the Steeler's offensive DVOA have been much higher?

2. What was it about the Colts' performance that gave them such a high offensive DVOA for that game? Watching the game, it looked like the Colts' offense was getting creamed.

3. Any chance of getting the DVOA variance of the four remaining teams with their playoff performances factored in? I think that is an interesting stat.

4. I noticed in the charts from last week's AFC matchup preview that Pittsburgh's game-by-game DVOA has gone as high as 120% for a couple of games this season, while the Colts' gbg DVOA, while more consistent, never went higher than 100%, and then only for one game. Could this be the elusive "swagger" everyone is talking about? No? OK, just checking.

by NF (not verified) :: Wed, 01/18/2006 - 3:01am


In between Edge getting very little yardage, which is partly understandable given the Steelers defense, and Manning either getting knocked down or forcing throws so he wouldn't get knocked down, he was gunning the ball up the field in big chunks of yardage. I think the Steelers also had an edge on time of possesion, which distorts the perception. Also, I don't think the Colts offense really got going until the 4th quarter, when I said it was as if Manning turned into the Hulk after waving off the punt team.

"Punt team make Hulk angry! Hulk throw far! Hulk crush puny team!"

I think Manning went from choking to anti-choking in the 4th quarter. Manning stepped it up, but thought he could win the game himself, when his offensive line was failing to pass-protect well, and a couple of situations he should have handed off to James. I think there are two reasons Dungy didn't try to get Manning out of this bad pattern. First, Dungy claims the punt team wasn't supposed to go on the field in the first place, so Dungy wouldn't have realized there was a problem until into the 4th quarter. Secondly, Dungy had nothing hat worked other than letting Peyton go crazy passing. And he did, with only three plays in the 4th quarter where he didn't pass out of the shotgun, and only one running play. I think how the Steelers stopped him on the last two drives is that they realized that even with three timeouts Peyton was going to pass regardless what looks he got from the defense.

by Bobman (not verified) :: Wed, 01/18/2006 - 5:45am

RE #42, the Colts were freakishly consistent this year, and Pitt not so much. In my mind, it's like Indy was the student that consistently gets 90, 91,92 on tests. Then along comes the hungover/bored rocket scientist type, who might get a 75, then a 100, then an 88, and now as the season closes, a couple over 100 (with extra credit). Actually, their DVOA wasn't so high this weekend as their head to head matchups with Indy were almost unfair (as Aaron pointed out).

It mirrors the calm coach vs the volatile coach pretty well. Which would you rather have? As a Colt fan I am struggling with that: extremely successful and steady regular season vs being able to pick it up for the post season. (Dungy calls each game "just another game" and often says "we do what we do" which helps keep things steady and puts one loss or 13 wins, for example, into perspective.) Whereas Pitt clearly did not treat this as just another game and did some new stuff. Which (memo to Coach Dungy) appears to be required!

RE #33: Jon M be very funny. Hulk like funny. Nine marks funny. Convince Hulk.

Rowdy, we Seattlites say "rain schmain!" If you're nuts, you're nuts, and the unending gloom tends to overly calm some people down (see: coffee addiction, Starbucks HQ division) rather than make them excitable trolls.

Though you give me a great idea: An FO "all-pro team" of trolls. This would be the best (i.e. most extreme and ludicrous) trollmail extolling the virtues of each team while impugning DVAO, east coast bias (having lived 30 of my 41 years in NYC, I NEVER thought of Indiana as East Coast--if not for the Colts and some domer friends, I'd have never thought of it at all!), Aaron's heritage, IQ, hairy palms, etc. Of course, some of the best have been (redacted) redacted, so us regular users might not have even seen them.... And who really has the time? Not I. Plus you have a game to watch this weekend, while I'll just be sucking my thumb, weeping into my Manning jersey, and reading stats from MVP seasons gone by. (does that fit the cliche image of a bitter Colts fan with nothing meaningful to do in the post season? I'll have to ask Joey Porter. Oops, I forgot bribing the officials and whining about them.) Hey, good luck this weekend. That goes for everybody else with teams still alive.

by JC (not verified) :: Wed, 01/18/2006 - 6:42am

Hmmmm, where to start? I have faithfully followed your power rankings all year, occasionally I agree with them but unfortunately, most times I don't. Aaron, you've been hiding behind this "DVOA" system for some bizarre reason. I understand it's place in sports analyzing but as a "professional" sports analyst shouldn't you rely on other factors? I'm an NFL fan, yes I have a favorite team but I remain un-biased when I read over your rankings. You have made three glaring errors this season that have troubled me enough to finally post just to make me feel better.

1.) You placed the Colts at number 1 midseason even though your DVOA system ranked them at 5. Be consistent! If your going to let your thoughts enter into your rankings then you should reconsider your "system". This system seems to miss some important factors that play out at the end of the year. The Colts lost to the Chargers in a reasonably critical game. They (the Colts) played there last two games awfully soft when they should have tried to make a statement (given their history of falling flat in January). Also, there "amazing" offense racked up most of their points against under 500 teams. And finally, look at their division, they were a shoe-in for 6 of their 16 games. I certainly have respect for the Colts but after watching a few of their games I truly felt they were somewhat overrated - you made Peyton sound like he had a cape and big red "S" obviously it didn't quite work out that way.

2.) You jumped on the Bengals bandwagon and rode it until the wheels fell off. Carson Palmer IS awesome and Chad Johnson IS one of the best receivers in the league but of the games they played the end result was consistent; wins in 8 games against teams that FINISHED the season UNDER 500, 2 wins against an inept NFC North in the beginning of the year, 3 losses to teams that made the playoffs, 1 loss to a 500 team and 1 loss to an under 500 team. They earned their win in Pittsburgh but they certainly didn't earn their position in your rankings.

3.) You have been slighting the Steelers since week 4 because they don't place well in this system you have. I agreed with their drop in the rankings when they hit their slide. They had two very important losses that can sway one's beliefs but..... in their last 4 regular season games they outscored their opponents 115 to 33, all of which were decisive victories with the exception of the Detroit game. Now they have beaten your beloved Bengals and Colts but you still think they don't have it? Ben is consistently in the 120's week to week, the Steelers run because IT WINS GAMES - one very important stat to consider 162-1-1 when leading by 10 or more points. Why would any coach jeapordize or change a winning formula to rack up points when the only thing that really matters is the "W". Furthermore, review the Patriots / Steelers game when you have a chance. Ben marched down the field at the end of the game without hesitation when they needed him to tie it up. Unfortunately there was just enough time for the Pats to get far enough to get the field goal. The Steelers have started with a lead and held it, started behind and then took the lead and held it, and of course dropped a few (I won't get into the whole Maddox thing). Bottom line is; Ben doesn't need to throw as much as your DVOA system likes, the defense stops opponents within the red zone and holds them to FG's (also doesn't fit well into DVOA) and finally, they are playing consistent, near perfect football right now.

The DVOA system is fun for considerations but as a "professional" shouldn't you do our job and analyze? Leave the math to the mathematicians.

by Kibbles (not verified) :: Wed, 01/18/2006 - 8:39am

Re #45: First, Aaron isn't "high" or "down" on any team. Aaron doesn't overrate Indy, Aaron doesn't underrate Cincy. DVOA does. DVOA measures how each team does on each play, and says how that compares to the rest of the league. And DVOA says that Cincy was performing better per play than Pittsburgh. DVOA is simply a mathematical formula. DVOA does not understand that these numbers that it's running mean anything at all outside of its formula. It simply takes the numbers, and churns out the results. Aaron is simply the caretaker. He inputs the numbers, and he presents (and often interprets) the results for the rest of the world. HE isn't ranking teams, DVOA is. And because DVOA is just a mathematical formula, DVOA doesn't account for injuries, teams playing "soft", or the planetary alignments. Sometimes this leads to results that are very much out of whack. Aaron's job, as a caretaker, is to interpret these results (saying, for instance, that Pitt was ranked too low at midseason because DVOA doesn't understand the difference between Maddox and Big Ben), or in extreme cases, to say that obviously there is a flaw in DVOA and to fix that flaw. Again, this isn't bias, this is simply Aaron keeping DVOA running smoothly.

As for suggesting that Aaron simply turn into just another sports bobblehead, mindlessly spouting off cliches (teams with a 100 yard rusher usually win!) instead of actually analyzing and contradicting common wisdom (teams that are winning run to kill the clock, and are more likely to have a guy rush for 100 yards)... to that idea, I say boo. The world already has enough Troy Aikmans, enough Phil Simms, enough Peter Kings and Dr. Zs and Tastefully Named Gregg Esterbrooks. Aaron is a bright guy, but I doubt he could contribute ANYTHING that those guys haven't already covered... except for "the math", as you put it. He can provide a fresh and unique perspective on football based on a statistical breakdown of teams on a per-play basis that isn't being provided ANYWHERE. And here you suggest that he gives that up, gives up the reason he got hired to be a "professional" analyst in the first place, to be just another guy rehashing the same old trite cliches that I'm already getting from three different sources? I think not.

Finally, using the results to contradict what DVOA says... sure, hindsight is 20/20. That said, Aaron continually tweaks his system to maximize predictability. Yes, sometimes DVOA gets it wrong... but I daresay DVOA gets it wrong FAR LESS FREQUENTLY than those sports bobbleheads that you would turn Aaron into. Who predicted that San Diego would be a good team last season? Who predicted that Carolina and Green Bay would come back from virtual elimination and make a run on the playoffs last season, or that Washington would do the same this year? No one. No one, that is, except for Aaron and his DVOA. Just because DVOA gets some wrong doesn't change the fact that it gets MOST right, and indeed, gets far fewer wrong than any other football "expert" I have ever seen.

Remember, too, that while you might want Aaron to become Troy Aikman or Ron Jaworski, both Troy Aikman and Ron Jaworski have PERSONALLY praised Aaron's statistics.

by Ruben (not verified) :: Wed, 01/18/2006 - 9:19am

JC's compaint:

"Aaron doesn't want to offer Smash-Mouth commentary; he wants to make you think. Aaron doesn't just line up some obvious facts, shake some magic beans over them, and offer a few cliches; he wants to try to trick you. He wants it to be thinking commentary instead of football commentary. Know what I mean?"

by dryheat (not verified) :: Wed, 01/18/2006 - 10:46am

#37, as a former returner, I'm of the opinion that Punt returning is MUCH more dangerous than kick returning in reference to the health of the returner. True, the punt returner doesn't have to return the ball, but then he doesn't have a job for very long if he just fair catches or lets the ball go. The punt returner is expected to return the punt if at all possible, and that means often taking a big hit before he gets out of first gear.

The kickoff returner, OTOH, typically has thirty yards buffer between himself and the wedge buster when he catches the ball, and has the option of running away from him. Add to that the blocking is much better on a kick return, and that you're less likely to be injured while moving full speed, and kick returning is the much safer of the two.

I agree Smith shouldn't be near the return teams.

#38 Kibbles. I'm not one of the "Broncos didn't win, Pats lost" crowd, but by your logic, the Steelers didn't win, the Colts lost, as the Colts have a much higher DVOA for the game. The difference in DVOA, and in the game is Special Teams. Fine, Special Teams is very important. But I don't know how much credit the Broncos can take for Troy Brown muffing a punt, which was likely the single play of the game that caused such a wide margin in Total DVOA. If the Patriots muffed four punts recovered by Denver, Total DVOA would have been blown well out of proportion, and Denver WOULDN'T have done anything special to deserve it. You seem to be missing the fact (or maybe I don't understand it) that one team's mistakes are going to inflate the opponents' DVOA as much as they hurt their own. In other words, to my eyes you're finding the effect and holding it up as the cause. Or am I just way off in understanding special teams DVOA?

by admin :: Wed, 01/18/2006 - 11:14am

Heh. I don't need to defend myself when I've got Kibbles to do it for me, and quite accurately at that.

Lots of good questions here, I know I owe people another mailbag, I'll do one in the "dead week" between the conference championships and the Super Bowl. Quickly, dryheat is right that what is good for one team is bad for the other, with two exceptions: field goals, and kickoff distance. (Kicking team is measured by net, return team only by return.) Muffs and fumbles on punts are treated the same by the system, though that's hopefully going to be upgraded.

As far as the "Did the Pats lose, or Denver win" question, or the question of Pats fans blaming officials, I was REALLY angry at the Asante Samuel PI call, until Brady threw the interception. That was the game-changing play, and it was a stupid decision caused in large part by great Denver pass pressure. From then on, it was clear the Pats were getting outplayed by Denver, despite Denver's inability to run the ball. But for a while there, it looked like the Pats were coming back and Denver had stalled, and that's where the frustration comes from.

Fumbling is usually not dropping the ball, either -- some Denver player had to knock it out of Faulk's and Hobbs' hands. The Brown muff was the only one that was just a bad Pats play.

by DGL (not verified) :: Wed, 01/18/2006 - 11:37am

Now, now, let's be tolerant. I think the only slip in JC's logic is one of historical perspective, which is to be expected from the wider viewership that FO gets due to the Fox deal, and his reaction is far, far more reasonable than many who show up here having first been exposed to FO from Fox. So let's not ridicule.

JC, if you haven't yet read it, I'd suggest you read the What Are We Talking About Here? article to get some background on the origin of DVOA. The key fact is that FO is football analysis based on mathematical analysis of statistics. The mathematical analysis came first; the talking-headism of the Fox deal came years later, and if it wasn't for the unique perspective that a mathematical analysis brings to it, I suspect the talking-headism never would have happened. (I can imagine Aaron telling Fox, "Y'know, this DVOA stuff was interesting, but I think next year I'll do my power rankings more like Peter King." And Fox saying, "Not on our website you won't.")

Many of the objections you raise about DVOA are shortcomings that Aaron has acknowledged, and I suspect will be experimenting over the off-season to find improvements to address.

End-of-season affects? Well, Aaron spent several paragraphs in the Final 2005 DVOA Ratings discussing this very issue, and tossing out ideas about the most accurate way to address it, though none that could be implemented and tested in time for this season's ratings.

Strength-of-schedule adjustments (behind the high ratings of both the Colts and Bengals): Your point seems to be that the strength-of-schedule adjustment is underrated (since the Colts and Bengals had a lot of success against bad teams, contributing to their high ratings). Aaron tweaked SOS the other way in midseason, but the results came out very close both ways. Another area where Aaron has acknowledged there's room for further experimentation.

"Slighting the Steelers": Actually, over the last four weeks of the regular season that you cite, the Steelers DVOA went up by 9.7 points (from 16.4 to 26.1), and WDVOA went up by 12 points (from 10.1 to 22.1). Over the two playoff rounds, DVOA went up another 4.1 points (to 30.2) and WDVOA up another 8.1 points (to 30.2). I would say that DVOA has reflected the Steelers late surge. But WDVOA (and more so, DVOA) has a memory, and doesn't forget that the Steelers threw away the Baltimore game, got run into the ground by Indy, and lost a shootout to Cinci in weeks 11-13. Denver, by contrast, hasn't looked bad since Week 1.

There's been some discussion of how much DVOA "likes" passing versus running, going back to the Great LHC Debate -- it was hypothesized that teams with a strong running game don't rise up as high in DVOA as teams with a strong passing game. Personally, without having ever seen the algorithms, I don't think that DVOA "likes" either one better -- DVOA "likes" plays that pick up first downs and touchdowns, regardless of how they're achieved. And I think it doesn't give as much credit to the 40-yard bomb as it would to three 14-yard runs on consecutive first downs.

I do think that DVOA can be misleading when a team goes into a "prevent offense" -- run the ball, chew up clock, and if you score, great, but the real objective is to use up time and mimimize the risk of a turnover. DVOA assumes (reasonably) that the objective of an offense is to put points on the board, and credits plays that contribute to that objective. But sometimes, that's not the primary objective of an offense, and I don't think DVOA accounts well for this. Since the Steelers' preferred strategy for winning games seems to be get up early and pound, pound, pound, they often get into the prevent offense when they're successful -- which I think penalizes them in terms of DVOA.

I'm sure if you have any suggestions on how DVOA can take this into consideration, Aaron would be happy to hear them.

by Scott de B. (not verified) :: Wed, 01/18/2006 - 11:41am

I doubt they have the cap room to fix the obvious problems they have - defensive secondary, tackles, running game - that led to them flattering to deceive somewhat this year. Its just the way it goes. No-one is immune to the cap.

Those problems were due to injury, not lack of cap space. Healthy, the 2005 Patriots would have been better than the 2004 Patriots. I don't see anything that would stop them from going 14-2 next year, though of course, like any team, they could perform worse.

by Scott de B. (not verified) :: Wed, 01/18/2006 - 11:44am

I do think that DVOA can be misleading when a team goes into a “prevent offense� — run the ball, chew up clock, and if you score, great, but the real objective is to use up time and mimimize the risk of a turnover. DVOA assumes (reasonably) that the objective of an offense is to put points on the board, and credits plays that contribute to that objective. But sometimes, that’s not the primary objective of an offense, and I don’t think DVOA accounts well for this.

Isn't DVOA normalized for time on the clock and score? So running a prevent will hurt your DVOA only if either a) other teams with a big lead don't run the prevent, or b) other teams with a big lead run the prevent better than your team.

Now, you could argue that running a good prevent is less predictive of, say, performance in the playoffs (where you are not likely to run up as huge of a lead as against a regular-season patsy), but I'm not sure that makes a whole lot of difference in the end.

by Harry (not verified) :: Wed, 01/18/2006 - 11:54am

#49, If you get a chance look at the replay - Brown is hit by a Bronco before the ball hit the ground after he called for a fair catch. Had the refs been paying attention this should have been a penalty on the Broncos, but the Broncos got the "muffed" punt and it became essentially the "game-over" play. It was clearly not the Pats day.

by pawnking (not verified) :: Wed, 01/18/2006 - 11:54am

I have yet to be convinced of the PREDICTIVE VALUE of DVOA in any one given game. DVOA is great at spotting season long trends, and great at pinpointing why a team beat another, but I'm having realy trouble seeing much ability to say "Team A will beat team B because of XYZ"

Am I merely misunderstanding what DVOA is all about, or am I correct?

by dryheat (not verified) :: Wed, 01/18/2006 - 12:07pm

#53..I remember the same thing happened in a Steelers' game not too long ago, where the returner called fair catch, was bobbling it and was hit, resulting in a change of possession. The play was flagged for fair catch interference. I was wondering if it would be called. Although I don't think it should've, it has been called before.

by zip (not verified) :: Wed, 01/18/2006 - 12:10pm


Aaron makes a lot of references in game previews to things like "Team X ranks 30th in the league against tight ends, look for player Y to have a big day." I always assumed that this was based on DVOA.

Also, think about the difficulty of saying "Team A will beat Team B because of X" -- for example, "Denver will beat New England by forcing turnovers" or "Carolina will beat Chicago by moving the ball consistently for the entinre game." I think getting that last part right -- the reason is something that is considerably harder than predicting a winner.

by zip (not verified) :: Wed, 01/18/2006 - 12:13pm


The key difference was that in the Steelers game the punt returner bobbled the ball into the air, meaning he still had a chance to catch it, whereas Brown dropped it and had no chance of catching by the time he contacted.

I'm not sure how the rule is worded, but it seems like the spirit of the rule was correctly enforced between these two plays.

by young curmudgeon (not verified) :: Wed, 01/18/2006 - 12:16pm

Re 51: "I don’t see anything that would stop them from going 14-2 next year" Wow! I think the Patriots are going to be good, probably very good, but you are saying that the most likely result by far is 14-2? For any team, I'd anticipate a couple of losses just on general principles (competitive balance, a few bad bounces, a couple of key players have an off day, a bad call at a critical time, having clinched the division and resting the starters, etc.).
I understand confidence and magic beans, but you really don't see ANYTHING?

by RowdyRoddyPiper (not verified) :: Wed, 01/18/2006 - 12:26pm

JC, just to address a few of your points:

The last 4 opponents were CHI, MIN, CLE and DET...combined record 31-33, not exactly a murderer's row. Blowing 3 of those 4 teams out of the water is good, but not great. The fact that they were getting torched by Joey Harrington was a source of great consternation for me...I was worried.

DVOA is also a rate, it measures success per play, not aggregate success. If Pittsburgh is running successfully than being run heavy will not have an adverse impact on DVOA. The problem is, Pittsburgh has not run as successfully this year as they have in the past. I've watched almost every snap of Steeler football this year and I have to tell you, there have been a lot of 2 yard runs and some 50 yarders mixed in. DVOA punishes that and game results do as well. DVOA also rewards teams for holding opposition to field goals. If someone is kicking a field goal rather than scoring a TD, odds are that you've stopped them from converting a 3rd down. This is rewarded in DVOA.

Please take the advice that is posted before every section and READ how DVOA actually works. DVOA does consider the situation that each play is run in...that's kind of the point. If the Steelers are rushing with a two TD lead, that is accounted for. If the Steelers want to have a higher DVOA they need to have more success than the average team in a given situation running the ball. As someone who has watched them...a lot...this didn't occur that frequently in the regular season. In fact if we were running the ball successfully we should never have had to convert two 4th downs to kill the clock against Indy. Not that I'm upset we made the conversions, but we shouldn't have been in that situation in the first place.

Kibbles, one quibble. I don't think that characterizing Aaron or FO in general as caretakers is fair. They have actually specified and implemented a model for valuing team performance. The model equations were not lying around waiting to be discovered, it took a lot of work to specify, implement, test, re-specify, re-implement, re-test, lather, rinse, repeat. Sorry, it just sucks when someone thinks you're the guy that just pushes the button and the button fell from the sky. Rant of a sometimes overlooked model builder ended ;)

PS. See you in the Beard Bowl.

by B (not verified) :: Wed, 01/18/2006 - 12:53pm

Pats 2006 Schedule: NFC North (I see this as 4 wins. Lions, PAckers and Vikings arn't getting any better, and the Bears willbe worse). AFC South (3 wins), Den Cin (1 win). Plus 4-2 in the division (this is pretty reasonable, Jets and Bills will struggle with new coaches, but Miami should be better) and that's 12-4. Now, if they find a way to beat Denver or Indy, that's 13-3 and should be enough for a 1st round bye. Of course, there are major questions for thier roster next year. First question is how will Rodney Harrison recover from his injury. Second is how much do McGinnist/Bruschi have left in them. Third is who will replace Corey Dillon as the every down back. It seems to me that if they can draft a good running back and inside linebacker, re-sign Seymour and Givens, they would be the team to beat again.

by JeffS (not verified) :: Wed, 01/18/2006 - 1:06pm

re: 60

It is fairly aggressive to propose a 13-3 or 14-2 season already for next year. I agree that the Pats should be the favorites to win the East, but a more reasonable Vegas over/under on wins for the season would be 11 maybe even 10.5 - 4-2 in the division is probably a good estimate, but 5 of the other 10 games are against 2005 playoff teams. A 3-2 record there is a good guess. So then they'd win out the remaining five games? No lapses? No close losses? With the benefit of the doubt, 80% is a decent win rate. That's 11-5. That's probably not a first round bye. I like the optimism, but color it with a bit of reality.

by pawnking (not verified) :: Wed, 01/18/2006 - 1:10pm

#60, I'd have to say that signing a quality RB is priotiry #1 for the Pats. Their system works best with a good RB. Without one, they cannot win a Superbowl, IMHO.

by Pat (not verified) :: Wed, 01/18/2006 - 1:14pm

I have yet to be convinced of the PREDICTIVE VALUE of DVOA in any one given game.

Yah, it's not designed to predict one team to win over another. I think that's a bit beyond any ranking or rating system right now - at least, to do better than about 60-70%, which is what naive systems can do anyway (heck, *random* picks get you 50%).

Of course, you do have to realize that football isn't entirely predictable anyway. It is a game, after all. Heck, baseball metrics can't predict the outcomes of baseball games - but they can predict the outcomes of baseball seasons.

So it is more predictive than naive predictors on a season-long trend. Just look at Atlanta, which DVOA very strongly said "look, guys, sorry to say, this team isn't doing that great" at midseason when they were 6-2. Who would've picked them to not have a winning season then?

by Steve Z (not verified) :: Wed, 01/18/2006 - 1:24pm

Re: #50
I do think that DVOA can be misleading when a team goes into a “prevent offense� — run the ball, chew up clock, and if you score, great, but the real objective is to use up time and mimimize the risk of a turnover. DVOA assumes (reasonably) that the objective of an offense is to put points on the board, and credits plays that contribute to that objective. But sometimes, that’s not the primary objective of an offense, and I don’t think DVOA accounts well for this. Since the Steelers’ preferred strategy for winning games seems to be get up early and pound, pound, pound, they often get into the prevent offense when they’re successful — which I think penalizes them in terms of DVOA.
So far as I can tell, DVOA appears to be agnostic about the strategic options available to a team and about the role game time plays in a football game.

It assesses teams on their abilities to advance the ball down the field, to prevent their opponent from advancing the ball down the field, to score and prevent points. The results of this analysis are then used to estimate (via the Forest Index and the Pythagorean formula) the number of wins (and losses) the team in question would have. This kind of analytical strategy, one that focuses on piecemeal success, works well in baseball because baseball teams have, in principle, an infinite amount of time to score runs. Because baseball games can last forever, baseball teams are not required to manage a clock. They can score as many runs as they can. This is not the case in football. Teams are limited in their ability to score (or prevent) points by the game clock. This fact requires football teams ‘to manage the clock’ — that is, to treat game time as a variable it must control in order to win games.

As DGL has suggested, the Steelers deploy a ‘prevent offense’ strategy when they achieve a large lead (>10 points). That is, they mostly abandon the forward pass in order to run the ball, use up game time and thereby deny the opposing offense the ability to score. Since their opponents are well aware of their strategic choice and prefer winning to losing, they can and do stack defenders on the line of scrimmage in order to stop the Steelers run-driven march down the field and control of the game clock. It would appear that the Steelers’ (read: Cowher’s) strategy works, even though this strategy seemingly won’t maximize the Steelers’ DVOA and its estimated wins. For instance, given a 10-point lead and the ball in the early part of the fourth quarter, Cowher’s strategy would greatly value a 10-minute drive composed of 3.4 yard runs which ultimately netted a lost fumble on the opponent’s one-yard line. I doubt that DVOA would value this kind of outcome as much as Cowher and the Steelers’ would. Cowher and the Steelers would value this kind of outcome because they believe, correctly, that the game clock (the forward movement of time) is their ally when they have a large lead.

by Opiwan (not verified) :: Wed, 01/18/2006 - 1:29pm

Re: 54

Well, I don't know what exactly you're looking for, but there were several times this season I won my office pool (winners only, no spread) basing my decisions on DVOA instead of the "random intuition" approach used by my peers. I can say with certainty I'm pretty resented because of my success this season, and I owe it all to DVOA and the "home field fudge factor". Is that the kind of predictive relationship you're looking for? Who wins?

by Vash (not verified) :: Wed, 01/18/2006 - 1:36pm

57: It's correct he bobbled the ball into the air, but the replay showed the officials made a questionable call because the ball was several feet beyond his reach when he was hit.

by J (not verified) :: Wed, 01/18/2006 - 1:41pm

I have yet to be convinced of the PREDICTIVE VALUE of DVOA in any one given game.

Using only total DVOA to predict games can be miss leading. There are other stats associated with total DVOA, which can give insight into game to game matchups.

Comparing total DVOA can give you an idea, but using all the data is best.

I have defended this site on a number of occasions. I have found using playing cards (careful with word choice - do not want to lose message to cyberspace). For you who understand cards...A-K is a relatively good starting hand. A starting hand of A-K vs. K-Q has about 81% chance of winning. However, if you take that same A-K against 10-J suited, the chance of winning goes to around 63%. SAme hand, but matched up against different starting hands reduces the chance of success.

The same is true with total DVOA. One team's total DVOA might be higher than their opponent's, but after taking into consideration the other stats (maybe higher DVOA team plays TEs very poorly and the other team has best TE stats) will give a more accurate reading as to the probability of victory.

On any given Sunday, any team in the NFL can beat another...that is why they play the game. However, the probability of a very weak team beating a very strong team is low. It is all probability, not certain.

by Pat (not verified) :: Wed, 01/18/2006 - 1:42pm

The results of this analysis are then used to estimate (via the Forest Index and the Pythagorean formula) the number of wins (and losses) the team in question would have.

Forest Index actually utilizes performance in certain situations more - performance in close situations, red zone offense/defense, etc., which means that prevent defense giving up points is not going to be that poorly rated.

Though I do think that clock management is a big deal, the thing is that teams can't control the clock arbitrarily. They have to hold onto the ball. So the offense can't be entirely below-average, for instance, because that gives the ball back to the opponent - and that's poor clock management, anyway.

So in some sense, the two are related. A team that has consistent success gaining first downs and bleeding the clock is going to have a high DVOA, and they're going to bleed the clock out. If a team just stops gaining first downs entirely, they're not going to have a high DVOA, and hey, they're not going to bleed the clock out either.

by Steve Z (not verified) :: Wed, 01/18/2006 - 2:01pm

Re: #68
I do agree that good teams will perform well, generate high DVOA scores, better than average estimated win values and will manage the clock quite well. I suspect the differences in strategies will show up at the extremes: Team A is like the Steelers: Get a large lead, then run the ball to hold it. If successful, Team A will generate good scores on the FO metrics. But, it might actually win more games on the field than the FO metrics would predict because of their clock management strategy. Team B is a high-powered offense. Unlike Team A, it’s strategy subordinates direct clock management (deny the opponent time to score points) simply by scoring so many points that most offenses won’t have enough time to outscore them. I believe an offense of this type would produce higher offensive values than an offense like Team A’s. Yet, Team A could, when called upon, adopt and be successful Team B’s strategy. In fact, Team A uses the score a ton approach when building its lead. It then switches to the offensively conservative clock management strategy to secure the win. The FO metrics seem to underrate a team like Team A. At least, that is my belief, although I do not think DVOA wholly misses out on the strategic value of Team A’s approach. I do believe that DVOA does partially account for Team A’s clock management. It just might not properly value this strategy.

by J (not verified) :: Wed, 01/18/2006 - 2:04pm

DVOA does not take into account clock management very well.

For example, this Sunday the Steelers went for it twice on 4th down and short. This means the did not convert on third down, bad for DVOA. However, each 4th down they converted took around 2 minutes off the clock.

Then, the Steelers were on the 37ish yard line, 4th and 5-6. They punted, a punt that netted only 17ish yards due to the touchback.

This drive for the Steelers was relatively successful. They took valuable time off the clock.

However, according to my understanding of DVOA, this drive would not have been rated very well...mainly due to failed 3rd down conversions.

by Crushinator (not verified) :: Wed, 01/18/2006 - 2:05pm

Aaron has said it before - DVOA is at its best when looking at match ups and breaking it apart, rather than comparing two teams overalls.

Sometimes, a team that's ranked lower can beat a better team just because they match up well. If a team thats ranked number 1 struggles vs strong Tight Ends and has NO strong tight ends in their conference and they're suddenly playing a team with one, that can be taken advantage of.

The Steelers ranked as poor defending the pass to the RB but that wasn't something Indy exploited.

The most valuable stats IMO are DVOA run defense and pass defense vs DVOA run and receiver numbers - Something thats proven to be pretty accurate (Like Berrian and the Panthers)

by Pat (not verified) :: Wed, 01/18/2006 - 2:13pm

For example, this Sunday the Steelers went for it twice on 4th down and short. This means the did not convert on third down, bad for DVOA. However, each 4th down they converted took around 2 minutes off the clock.

Uh, yah, but converting it on 4th down is good - very likely, *very* good. So 3rd and 4th down averaged together gets you a positive DVOA for the series. If they would've missed it on 4th down, then that would've been a negative DVOA for the series (and bad for clock management and field position as well).

by J (not verified) :: Wed, 01/18/2006 - 2:16pm


Exactly, people need to look at all the stats, not just total DVOA.

"The Steelers ranked as poor defending the pass to the RB but that wasn’t something Indy exploited."

Interesting you wrote this. I could not believe the Colts did not dump the ball off to edge, esp. with so much pressure coming from the OLB.

Also, Tunch Illkin, showed Indies last drive...post Bettis fumble. On their last two plays before the missed field goal, they threw it deep into the endzone. It was clear that on the first play Edge was wide open, and he would have easily picked up 8-10 yards. The second play was the same, but I think it was TE on a very short route, but he to would have picked up 8 - 10 yards and a first down.

by J (not verified) :: Wed, 01/18/2006 - 2:25pm


See, my understanding of DVOA is growing. I thought it was bad not convert on 3rd down. 4th down conversion are certainly good, but a "successful" team should not be in a situation to have to convert 4th down.

Aaron, maybe you could give us this drive's numbers.

by Sara (not verified) :: Wed, 01/18/2006 - 2:28pm

#34 - He Hate Me is still a Panther, but he's been battling injuries. Not sure if they IR'd him, but for all intents and purposes he's out for the year.

by pawnking (not verified) :: Wed, 01/18/2006 - 2:35pm

Please note I am not complaining about DVOA in general, I became a convert some time ago. However, the best test of the accurracy of any model is it's predictive powers. Thus, those fantasy owners who going by projected FO stats loved Matt Hasselbeck, they hated Dominick Davis.

Regarding the preview of the Indy-Pitt game, I was going to say that there wasn't much there. Then I actually re-read it. So much for my memory. Let's look at some specific quotes:

"The danger for Indianapolis is that Pittsburgh can pressure Manning more and worry about running back Edgerrin James less. For the second straight year, James slowed down in December."

It certainly looked like Pitt didn't worry much about James. And whether Manning continued to throw because of huberis or because he knew James was slow is a matter of speculation, but ceratinly the Pitt defense acted like they weren't worried about him. And apparantly they were correct.

"To win this game, the Steelers need to reverse their usual run-first mindset. We know they can do it: Roethlisberger averaged 10.4 net yards per pass in the first quarter this season, two yards higher than any other quarterback."

In the first quarter, the Steelers threw a lot and moved the ball easily on the Colts. They then ran it and had trouble doing much.

Also, the "trend" DVOA showed Pitt as much more closely matched with Indy than the overall season would indicate.

Overall, it seems that the play by play charting experiment FO used to predict the Steelers' blueprint to beat the Colts was fairly accurrate. It correctly speculated that James would not be much of a factor, and correctly projected the Steelers' winning game plan of throwing rather than running would be successful.

Terefore, I have to reverse my earlier stand, and apologize to Aaron and Co. I am looking forward to the upcoming game previews, with an eye of how accurrate their predictions will turn out to be.

by Pat (not verified) :: Wed, 01/18/2006 - 2:43pm

See, my understanding of DVOA is growing. I thought it was bad not convert on 3rd down. 4th down conversion are certainly good, but a “successful� team should not be in a situation to have to convert 4th down.

The thing you have to realize is that not converting on 3rd down *was* bad. You're trying to justify it by saying "well, but they got it on 4th down, so it all worked out" - which is true, but that's a lot like saying that rushing for no yardage on 1st down is okay if you convert 2nd and 10. No, it was still bad - it's just that the converted 2nd and 10 was really good. That doesn't mean they weren't in a bad position at 2nd and 10.

Likewise, not converting the 3rd down was bad. Having 4th and 1 is a worse position to be in than 3rd and 1, so that play was a net loss. And I guarantee that Bill Cowher didn't tell Bettis "hey, go in there... and lose a few, okay? I wanna burn a bit more off the clock so we'll get it on 4th down."

by Kal (not verified) :: Wed, 01/18/2006 - 3:21pm

One of the best predictions that DVOA made all year was Green Bay beating Atlanta this year. DVOA had shown repeatedly that Green Bay was losing primarily because of luck, and Atlanta was winning because of it. And that proved out well. Washington's success is another great one. Denver's success over the Pats is another. Chicago's ability to throw to Berrian.

DVOA predicted all the wild card matchups correctly and predicted 3 of the 4 division games - and in all of them, gave a fairly precise recipe for how a team would have to win if they did win. I wouldn't be harping on DVOA so much if it didn't actually, like, _work_.

Where it could use help is predicting future success of certain positions for fantasy football, but I'm willing to go on a wait and see kind of thing. Yes, I'm a bit miffed about the Kevin Jones thing, damnit. :)

by Ted Max (not verified) :: Wed, 01/18/2006 - 3:31pm

I want to add my own bit to #76:

In my view, the best test of a model is not just whether it is highly predictive, it's whether it makes good predictions WHILE being simpler than the reality it describes. Statistically, it's easy to create a perfectly predictive model: Just throw in the same number of variables as you have outcome measures, and tada, perfect predictability. But that model is JUST as complicated as the reality, so it doesn't actually give us any knowledge we didn't already have.

That's why I am not a huge fan of the "let's bolt another thing (injuries, coaching styles, off-field distractions, etc.) onto DVOA" school of thought. If DVOA is so complicated that it's equally as complicated as the game it is describing, what's the point?

It's like the old joke about a guy making an actual-size model of the earth. What would that model show you that you can't see just by looking at the earth? Models and maps are useful because they are selective, not because they are all-inclusive.

Balancing accuracy with simplicity takes a lot of skill, and I think the FO folks have done a good job of striking that balance.

by JC (not verified) :: Wed, 01/18/2006 - 3:39pm

I'm back to comment on the subsequent comments following my little rant earlier. I'd like to thank all who replied, you provided some excellent insight into this system and I agree with all of your comments - thank you for respecting my interpretations.

I also have faithfully watched every Steeler snap this year and would agree that we had a shaky season - even when our starters played. I spent many Sunday afternoons nervous and pensive during their play (as a Steeler fan this is almost always the case). I would further agree with the rankings with the exception of the Colts during the week that DVOA placed them at number 5 (and Aaron put them at number 1) and now placing them at number 1 after what was a decisive loss. Additionally, I have had a problem with the Bengals rankings all season, yes they have come a long way, yes they have an awesome QB, WR, and RB but they were essentially the best of the worst. I look forward to what will be another wonderful rivalry over the next few years.

My pointed comments towards Aaron were based on his commentary during the season. It seemed he favored certain teams and begrudingly placed others. Therefore, my point is DVOA is intriguing, interesting, and by all rights - accurate. I'll admit I need to have a better understanding of the system but I think the Steelers form of football needs to carry a bit more weight in the rankings. Furthermore, I'm a bit surprised that noone thought the Steelers could compete against either the Colts or the Bengals when they were obviously stepping it up a notch while the previously mentioned teams were slipping - shouldn't DVOA address this inaccuracy in the future?

So.... Now the Steelers go to Denver. DVOA (and just about every other source) put's Denver to win the Super Bowl. Denver is a great team, they played very well against a team that I fear and have great respect for. However, their offense didn't overly impress me and there was nothing special about their special teams. Their defense looked scary (for a Steeler fan) but I would contend that expecting to win games off of turnovers is a dangerous proposition. Ben has been amazing over the past few weeks. The Steelers defense was incredible in the Colts game and our special teams have given the offense good field position and buried the opponents deep in their end. Given all of these observations I believe we are in for one very exciting game - that could go either way.

Once again, thank you all for your comments and I promise I will learn this system so that I can participate in these discussions further. As a note, I really enjoy this website because it doesn't seem to have the ridiculous "my team is better than your team" crap. After reading the comments posted here I am very impressed with the football knowledge that is expressed in these discussions. Thank you.

by pawnking (not verified) :: Wed, 01/18/2006 - 4:29pm

Continuing with my Kubiak Rant:
Running Back
Terrible predictions:
Domanick Davis - Listed as a top 3 or 4 RB, Davis ended up on the scrapheap. He played in 11 games, and was not effective in them. In my PPR league, he was a HUGE disappointment, never catching more than 8 balls in a game.
Although maybe I'm being too hard. Davis was an adequate 2nd or 3rd RB, but Kubiak had him so much higher than he turned out being...
Kevin Jones - Oh, the agony of all the smug KUBIAK fantasy owners who snagged these two in the Draft. Oh, the Humanity!!

Nailed these -
Tiki Barber - not highly regarded on most draft boards, usually listed as between #10-15 in the RB column. What a great second RB for KUBIAK players!
Priest Holmes - Correctly predicted his decline and injury
Ahman Green - Ditto

QB -
Poor Predictions -
Daunte Culpepper - along with everyone else, missed his train wreck of a season
Carson Palmer - along with everyone else, missed his great season.
Good Predictions -
Hasslebeck did turn out to be a top 5 fantasy QB.

WR -
Steve Smith - I'm guessing Kubiak has a hard time predicting a player coming off an injury. Smith was predicted to be a solid 2nd WR, not the stud he ended up being.
Larry Fitzgerald - Ditto
Randy Moss - Correctly predicted Randy's value would decline in '05
Santana Moss - Didn't forsee how huge his year would be, but had him significantly higher on the draft boards than other sites.

All in all, a very mixed bag for KUBIAK. On the other hand, they did donate some of their proceeds to Katrina sufferers, so it wasn't a total loss.

by MRH (not verified) :: Wed, 01/18/2006 - 5:03pm

Re #71 “The Steelers ranked as poor defending the pass to the RB but that wasn’t something Indy exploited.�

and #73 "Interesting you wrote this. I could not believe the Colts did not dump the ball off to edge, esp. with so much pressure coming from the OLB."

The Colts throw the ball to RBs only 13.6% of the time (attempts, not receptions). This is the lowest rate in the league (TEN 13.7% and SEA 14.0%). The league average is 20%. They were the lowest in the league in % passes thrown to RBs in 2004 as well.

Should the Colts have gone against their tendency and take advantage of the Steelers' weakness? Possibly. Is it surprising that they didn't? No.

by DGL (not verified) :: Wed, 01/18/2006 - 5:21pm

#82: To pile on the Throw Peyton Under the Bus bandwagon, how much of the "Failure to play against tendancy" do people think is attributable to the leeway Manning gets in his playcalling? Wouldn't it seem easier for conscious decisions to play against tendancy, taking advantage of a particular matchup or opposition weakness, to come from an OC sitting in a nice safe booth instead of from a QB in the heat of battle?

by dryheat (not verified) :: Wed, 01/18/2006 - 6:27pm

#57 et al re: Fair catch

Just watched the replay. The ball ricochets of Brown's hands to his left. He tries to grab the ball before it hits the ground when he is hit by Devoe. By the letter this is a fair catch violation (could have been a halo rule too, but that's a rare call). Although I would argue that at that point Brown had lost the exclusive right to catch the ball, it falls under the same rule as the Sproles play.

by Chris M (not verified) :: Wed, 01/18/2006 - 7:11pm

Re: 80

You weren't impressed by Denver's offense last week (and I doubt many people were), but this is one of the advantages of DVOA. It can chart how well Denver has played all year, and weighted DVOA can give a sense of a team's recent success. Both metrics have said that the Broncos are an excellent team. (I'm pretty confident in DVOA here because I can speak subjectively too - I've seen Denver play a bunch of times this year, and their offense looked quite good to me.)

Is it possible that defenses have figured out the Broncos and their earlier success will be irrelevant? Sure, it's possible. Computers can never understand the evolution in game planning that occurs over the course of a year.

by Catfish (not verified) :: Wed, 01/18/2006 - 7:23pm

Re: 81

I believe Aaron stated somewhere that the problem with the Steve Smith prediction was that KUBIAK treated all injuries as equal. It didn't split out broken bones (like Smith's leg) - which usually heal cleanly and without any ill effects- from injuries like 'the terrible triad' which have long term effects.

For all its failings though, I'd still take KUBIAK projections over the other fantasy projections out there, which use a glorified status quo method: "Everyone good last year will be good this year, old players will get slightly worse, and young players will get slightly better."

by Crushinator (not verified) :: Wed, 01/18/2006 - 7:56pm

The biggest thing that's really impressed me at DVOA is that it's been able to accurately determine a teams strengths and weaknesses - this doesn't mean that good teams always beat worse ones - The bad team might just know how to approach the good one, while the good team might just not take advantage of the bad ones weaknesses.

A lot of it I think is also just getting back to actual football. People aren't getting their football information very well. They're getting it from commentators who read a stat sheet and go "Oh, 300 yards passing. He had a terrific game. Make a point to mention this later". They're not looking at how - which often times leads into this belief that bad teams are bad all around and good teams are good all around and just because a team has a lot more Ws than Ls, they must be a good team.

What makes teams that are competitive year in year out isn't overwhelming talent and big names - It's film study, discovering weaknesses and exploiting, playing to your strengths, along with cap efficiency.

What makes the Patriots strong as a franchise is that they adapt well - and many teams just don't. The fringe playoff teams and non-playoff teams look at their stats, the tape, and they say "We have Ricky Williams, lets just run him every play for a few years". If a team has a strong run D and a weak pass D, they say "Well, we have a strong running game!" rather than "Well, lets rework our approach and take advantage of that"

As much as Joey Porter hates it, football is a game of strategy as much as it is a game of skill and the team that outsmarts the other usually wins. The salary cap usually does a job of keeping any one team of having disproportionately high or low talent.

by Björn (not verified) :: Wed, 01/18/2006 - 8:27pm

RE: #80

"However, their offense didn’t overly impress me and there was nothing special about their special teams."

Nothing special about their special teams? How about Todd Sauerbrun's fantastic punting accuracy? I haven't seen such a good punting performance in a long time. If Denver hadn't won the field position battle in the first half thanks to Sauerbrun, that game is tons closer.

RE: Steve Smith

I could go dig it up, but at the start of the year I predicted Steve Smith's season. Really, when a good receiver has a career year when he is the only option on a team, and is subsequently replaced by a better player, the writing is on the wall.

by chris clark (not verified) :: Wed, 01/18/2006 - 9:08pm

RE 79

There is some real complexity under-the-hood in DVOA. It's simplicity is just at the final result. Why 45% on 1st down, 60% on 2nd, and 100% on 3rd? Because that represents likely successful cutoffs as observed. However, those cutoffs are somewhat arbitrary--although well reasoned and I think they have been statistically validated. Okay, Aaron &co, have you validated the cutoff points for success on each down versus actual results? Is it possible that the correct cutoffs are 48% (or 42%) on 1st down and 75% (or 54%) on 2nd.

It doesn't well account for all special situations, e.g. running out the clock or trying to catch up, where the team may use the 4th down differently than in the normal course of the game. Similarly, resting starters potentially caused abberations. However, they have used good judgement in figuring out how to adjust so that those problems are minimized.

I had no problem the week where he listed Ind in slot 1, while showing their ranking put them lower. He still gave the objective stats and presented his reasoning for thinking they were missing something and thus incorrect.

I think strength of opponent is under-estimated, which has inflated Ind in the evaluation. However, showing that the current scheme minimizes wild swings in ratings by its opponent adjustments is an objective fact and supports the current model.

That said, I don't mind their being complexity under-the-hood in DVOA, especially not when it is statistically validated complexity that improves the systems predictive power. Make the DVOA internals as complicated as necessary to achieve the best predictions (as long as they continue to be predictions). We don't need the weather simulation that predicts tomorrows weather with 100% accuracy after 24 hours, we already have that. But I really do want as good a prediction as I can get before Sunday's games of their outcomes (and the reasons why).

And, many thanks to FO for giving us just that. They aren't just caretakers of the DVOA stats, they've done the hard work of checking to make certain that their stats are meaningful, even if not perfect yet.

by CaffeineMan (not verified) :: Wed, 01/18/2006 - 9:41pm

The biggest thing that’s really impressed me at DVOA is that it’s been able to accurately determine a teams strengths and weaknesses - this doesn’t mean that good teams always beat worse ones - The bad team might just know how to approach the good one, while the good team might just not take advantage of the bad ones weaknesses.

Crushinator, I agree with this. I love seeing the DVOA numbers after games because it gives me insight into what went well and what went badly, compared to the tendencies shown in an overall season. For instance, in spite of the final score, DVOA shows the Pats defense played pretty well, especially compared to their regular season performance. They were expected to be the weak link going in, instead, Denvers defense and special teams outplayed the Pats offense and special teams.

Bjorn, although I thought Sauerbrun punted well, I thought Denver's defense and offense did more to win that field position battle early. NE started drives at their 25,35,4,3,11,7,11 in the first half. Punts accounted for three of those, at the 35, 4, and 7. The 4 and 7 cases were kicked from NE's 45 and Denver's 46 respectively, resulting in kicks of 41 and 48 yards. So he was kicking from midfield. The punts were nice, but to me the big thing was that the Denver D shut down the Pats O and then the Denver O either scored or gained enough yards on their drives to gain at least good field position, even if the drive stalled. The only exception was the very first drive that was a 3-and-out for Denver.

by David Brude (not verified) :: Thu, 01/19/2006 - 12:44am

I would like to see a breakout for passing that is similar to the ajusted line yards for rushing that would have a few columns to show how dependant a team is on downfield passing or on Wr's making plays after the catch. For example i would envision categories like

average pass length
average yards after catch
deep ball success

by Pat (not verified) :: Thu, 01/19/2006 - 12:50am

The NFL's standard play by play doesn't record point of catch and yards after the catch. It'd be wonderful to have them, but there are only 2 stadiums in the league which record them in the play by play.

Maybe FO can start a petition for the NFL to expand this to the rest of the league? It'd be great.

by Fnor (not verified) :: Thu, 01/19/2006 - 1:37am

We're charting that, so hopefully they'll be able to use data from what we saw to get stats for more of the passing game.

And yes, Bill, I'm working on the backlog... please don't hurt me :(.

by RowdyRoddyPiper (not verified) :: Thu, 01/19/2006 - 2:36am

JC, glad to see you are open to what's going on here...it's really difficult to get a handle on this without reading the model documentation first. It's like walking in during the middle of a movie...maybe you can figure out what's going on and catch up, but maybe it's best to check out the first half and dive back in.

With that being said, allow me to introduce you to the FOMBC (Football Outsiders Message Board Curse)...[trots out still shaken Atlanta fan]...basically teams where fans complained en masse about the DVOA rating of their team were struck down and relegated to 8-8 seasons.

I think once you get comfortable with how the system works...you'll get more comfortable with how the rankings are coming out. DVOA has noticed our fair Steelers uptick in play; DVOA has increased week to week over the past month. Again as it says up top, DVOA is just a guide; apply your judgement accordingly.

I do agree that it appears counter-intuitive that the Colts end up with a higher DVOA than the Steelers for the Sunday game. But look back on the commentary and see, without the drive after the overturned INT the Steelers end up with a higher DVOA for the game. I think you'd agree that the overturned INT did tip the balance of the game in Indy's favor and put the outcome in doubt.

PS. What is with all of you guys wanting DVOA to predict things perfectly...don't you realize that once these guys get this thing perfected, DVOA is proprietary, the site is gone and they're on Ipanema Beach oiling down Brazilian supermodels and counting their gambling winnings!! Viva la imperfection!!!!

by John Arteaga (not verified) :: Thu, 01/19/2006 - 5:55am

It was said earlier that the Colts' DVOA would have been lower than the Steelers' had the INT by Polamalu and the subsequent Colts drive not been counted. Could you give us the adjusted-for DVOA for the Colts, with Offense, Defense, and ST percentages? Thanks. I can't wait to see this weekend's games.

by Tom Kelso (not verified) :: Thu, 01/19/2006 - 12:08pm

Something for the FOMBC file:

As bad as Atlanta was, the first team whose fans trolled the board to protest their ranking was Denver -- "going Denver" was actually the reference for about the first day of the Atlanta s***storm.

So far, though, the Broncos have avoided the curse. Unless...............

by J. Kevin King (not verified) :: Thu, 01/19/2006 - 12:24pm

I think DVOA does have some predictive power, but most people have a problem understanding statistics, so they sour on it if a particular prediction is wrong on a partticular game.

Homo Sapiens, in general, have a hard time understanding probabilities and randomness; this is why quantum mechanics seems so strange to us. Probably for evolutionary reasons, everything for us has to have a reason. For instance, one is flipping a coin and it comes up heads 10 times in a row, our instinct is to either search for some reason or assume that the next flip will have a higher chance of being tails, even though the chance is still 50/50. This goes against most persons' instincts.

In the same way, injuries, etc. aside, if a team has a higher DVOA, and the teams played 100 times, the team with the higher DVOA will most likely win more than it will lose, probably as a function of how much higher a DVOA the team has. Teams in the NFL are on so even a playing field that there generally is no sure thing.

So if one accepts the "Many Worlds" concept of theoretical physics, and one looks at 100 universes where the Steelers play the Colts, maybe the Colts win 55 (just to pick a random number) and the Steelers win 45. Because, in our reality, one of the 45 Steeler wins comes up, does not invalidate the information that DVOA gives us.

Stating that DVOA is somehow invalid or fatally flawed because a team with a lower DVOA won is like someone saying polls are invalid because they meet an African-American who voted for George Bush (or a conservative evangelical who voted for John Kerry). 85-90% of African-Americans have been voting Democratic in presidential elections. Still, that means that 10-15% of African Americans voted for Bush, and you might meet one. That does not make the poll invalid. (My numbers might be slightly off, but the general idea still stands.)

by B (not verified) :: Thu, 01/19/2006 - 12:38pm

96: Denver got bit by the curse when they lost to the Giants. Since then, the fans have behaved and the team has been playing better.

by Sid (not verified) :: Thu, 01/19/2006 - 1:28pm

As pointed out elsewhere, yes, DVOA believes that the Colts outplayed the Steelers, although it wouldn’t have the Colts higher if you included the Polamalu interception and the rest of that Indianapolis drive.

What do you mean here? Do you mean if you excluded the rest of that drive?

by Sid (not verified) :: Thu, 01/19/2006 - 1:33pm

RE: 8

I think Kelly Holcomb has been a good example of that (along with possibly Billy Volek).

RE: 10

Yes, it says so at the top.

RE: 23

I think part of the rise in ST DVOA for the Eagles is the fact that it is weighted. So I think Weeks 1-5 don’t count at all. That means quite a bit of crappy football is off the books.


by Robert Visser (not verified) :: Thu, 01/19/2006 - 2:01pm

Good Morning Aaron,

I just wanted to write and compliment you and the rest of the Football Outsiders team on what in my mind is the most outstanding work on the internet. It is rare to find the type of in-depth analysis that goes so far beyond the knee jerk rantings of the talking heads week in and week out.

I do have one question, I understand the reversal on the Broncos pass defense from a DVOA approach but am a little confused as to how the run defense is also reversed. While I concur that the NFL statistical rankings based on yardage are infantile I still can't get past the rushing defense as "middling" being accurate for the Broncos. If weighted adjustment allows for the longer runs being carried out against "prevent type" defense settings where the Broncos have held comfortable leads as opposed to say the desired goal of acheiving a first down does this impact the analysis?

Also, with very few good talking heads out there - Collinsworth and Costas spring to mind - do any of the regular net or TV crews spend in depth time with you and your team to enhance and augment their video analysis?

All the best and keep up the excellent work!

Rob Visser - Port of Spain, Trinidad

by Sid (not verified) :: Thu, 01/19/2006 - 3:56pm

At this point, the biggest worry has to be the legal status of right tackle Sean Locklear, arrested for domestic violence this weekend. Julius Peppers will be eager to chomp if he sees a pork chop across the line on Sunday.

Peppers is the questionable one here. If he does start, he won't be facing Pork Chop. He'll be facing Locklear.

RE: The Nolan clown

6th or 8th? WTF? How so?

by Sid (not verified) :: Thu, 01/19/2006 - 4:03pm

RE: 31

Agreed. I'm not a Patriots fan, but I don't understand how this is the "end of the Patriots dynasty." The Patriots are not the Chiefs, who have a bunch of over-the-hill players and will experience a steep drop-off next season. The New England core is mostly young players, and they will be getting Light back on the OL. Not sure how effective Harrison will be, but if he can make an effective comeback, the Patriots defense will undoubtedly regain its "swagger" :P
I would expect next season's Patriots to be better (and have a better DVOA) than the 2005 version.

by Catfish (not verified) :: Thu, 01/19/2006 - 4:50pm

Actually the first people to complain were Skins fans (including the since repentant james) after week 3, when the Skins were 3-0 but ranked 23rd or so. FO even ran an article that the skins weren't as good as their record. People complained, the Skins lost their next 2 and the complaints stopped. Then, oddly enough, the Skins were 5-6 and ranked 10th and people started complaining again. We all know how that worked out.

by JC (not verified) :: Thu, 01/19/2006 - 5:24pm

RE: 94
After spending the past few days reviewing this website I completely agree with you. I'll admit, at first I was put off by Aaron because I felt there was a bias involved in some of the rankings. I also viewed him as "just another bobblehead" as someone so succinctly put it, but after reviewing the site, the comments, and the articles I believe most of the people (professionals and readers) are well informed and educated in what is a very complex game. I have followed Aaron's rankings through the FSN site all year. I will say, I had agreed with every aspect of DVOA (including the Steelers) with the exception of the Bengals. If I was to make any comment about DVOA it would be that there should be an heavier emphasis on strength of schedule and using the play clock to win games - maybe there is and I don't understand it - yet.

I will refrain from any other pointed or scathing comments until I do get a grasp of the system. Thank you again for not taking an attitude towards a "newbie" - I look forward to next year when I can really get involved.

by Björn (not verified) :: Thu, 01/19/2006 - 5:56pm

Caffiene Man:
You have a good point. Perhaps it is just my homerism showing through, but Todd Sauerbrun really impressed me. Maybe it is because we haven't had a real punter in so freaking long.

I think alot of it is the general idiocy of the people who can't interpret the FO rankings. As a (somewhat) competent football viewer, I was able to recognize that Denver wasn't playing up to the level of their record, and that they deserved their non-elite ranking on the FO list. There were others, however, who were unable to see Denver's inadequacies, so they got angry. The more undeserved the ranking seemed, the angrier they got, but the more the likelihood increased of a Denver loss.

by Sid (not verified) :: Thu, 01/19/2006 - 6:45pm

RE: 50

Denver, by contrast, hasn’t looked bad since Week 1.

That's just inaccurate. What about Weeks 12-15?

by Björn (not verified) :: Thu, 01/19/2006 - 8:17pm

"Denver, by contrast, hasn’t looked bad since Week 1."

Denver looked worst in week one, but they have looked bad since.

Week 2 - ~20% DVOA
Week 5 - Below 10% DVOA
Week 6 - Bad 2nd half
Week 8 - Bad 2nd half
Week 12- ~20% again
Week 13- DVOA below 0%
Week 15- DVOA below 0%

20% isn't that bad, but their average for the season was 36.5%, so I am holding them to that standard.

I was surprised to see that in week 14 against the Ravens, Denver's DVOA is ~40%. I really didn't think they performed that well that day.

So Sid is correct. Saying that Denver hasn't looked bad since week 1 is inaccurate.

by CaffeineMan (not verified) :: Thu, 01/19/2006 - 8:45pm

Bjorn, I don't think it's homerism. I'm not saying you may not suffer from it. :D I'm joking, of course, we all do. :D I wrote that long post partly in order to reinforce a point that someone on this site has made in the past: what one remembers in a game is not necessarily the most important aspect of why the game is won or lost.

I researched your comment because I remember viewing that first half differently than you. I remember the Pats being trapped in their own end. But I remember the Denver D not letting them out while the Denver O kept chipping away at the field position, drive after drive. What you remember is Sauerbrun's punts putting them there. I think it's mostly just because of the differences in the perceptions of any two humans.

I think one of the great things about DVOA and other work like it is that it provides a consistent framework for checking out our perceptions. It allows a game to be measured against success criteria that aren't swayed by our impressions. That's not to say it provides The One Truth or that the success criteria are perfect. But it provides a much better common ground for understanding and discussion than: "I remember thing1! Oh yeah, well I remember thing2!"

by Björn (not verified) :: Thu, 01/19/2006 - 9:52pm

Too true, too true.

I will say though that at the very least, Sauerbrus did well to not give the Pats free field position by shanking one out of bounds or booming it through the endzone. I didn't mean to imply that Denver's D and O didn't contribute to the field position troubles for New England, but I still maintain that without Sauerbrun's solid play, the field position battle would have been much closer. Also, I don't think it is fair to count out Sauerbrun's second half punts. Sauerbrun kicked a 58 yarder at 13:32 left in the third (returned for 13 yards) and a 56 yarder at 4:51 that was returned for 14 yards. So those two punts netted 43.5 yards, which is a pretty significant number. If I am not mistaken, those punts were also well placed, leaving the returners no choice but to run up the sidelines, where the damage they could do was limited.

by CaffeineMan (not verified) :: Thu, 01/19/2006 - 10:49pm

Ahh, Bjorn, now you're changing the subject on me. :D Your original post was something like "Sauerbrun won the field position battle for Denver in the first half." I addressed that statement only.

You're focused on Sauerbrun, but I think the question behind your statement is about how well Denver's special teams played, as your statement was triggered by a comment about the Denver special teams not looking special. At 10%+ DVOA, they scored well. To me the real question was, what were the biggest components of that 10%. The long FG? Punting? Kickoffs? Coverage of punting or kickoffs? Hobbs fumble on one of the kickoffs? I think that's a good question for the FO mailbag. I'd like to see the flip side of that as well. What was the biggest component of the Patriots' special teams DVOA of about -10%? For one thing, I think the ST DVOA removes the offense or defenses contribution to field position, so it's pure in that respect.

by dryheat (not verified) :: Fri, 01/20/2006 - 12:10pm

Wouldn't the failed fourth and ones on consecutive drives (on the Denver 36, and then the New England 3) have more to do with New England's playing on the short field than Sauerbrun's skill? The Denver offense got one long drive, and their defense didn't give up one. The exchange on punts struck me as even. I suppose you could credit Sauerbrun for not punting into the end zone in that situation, but that's his job. Sauerbrun's big contribution to the game was hitting Ellis Hobbs waist-high, not diving at his shoes the way most kickers and punters do.

by Björn (not verified) :: Fri, 01/20/2006 - 2:49pm

"Nothing special about their special teams? How about Todd Sauerbrun’s fantastic punting accuracy? I haven’t seen such a good punting performance in a long time. If Denver hadn’t won the field position battle in the first half thanks to Sauerbrun, that game is tons closer."

Hmm... you are right. I was wrong to say "thanks to Sauerbrun." But I do say that his punting was special. Good distance, fantastic accuracy.

#112: The drive after the 4th and 1 fade route try ended in a pick that Samuel caught at the NE 11. Is this what you meant?

by dryheat (not verified) :: Fri, 01/20/2006 - 2:57pm

No, I meant consecutive in the sense that the first was New England's drive where the fourth and one play was Brady throwing low and behind Fauria, and on Denver's ensuing drive was the failed fade route.

However, upon re-checking the drive chart, those two drives weren't consecutive.

Besides the sequencing inaccuracy, my point stands: Had the Patriots punted instead of going for the first on fourth and 1 (which IMO would have been the wrong call), it would have been Denver who likely would have been playing on the short field the first half.

by Rich Conley (not verified) :: Fri, 01/20/2006 - 5:56pm

Aaron, is anything done to compensate for the situation where a team has 3 drives, where they go 3 and out, and then one drive where they go 12 plays for a TD?

It seems to me that you'd have 3 sets where they failed, and then had 4 sets where they did well. If you look at this, 3 bad set, 4 good sets, it looks like theyre playing pretty decent football, and it would average out to a decent number,

whereas if you look at it from a drive perspective, theyre getting crushed.

by Pat (not verified) :: Fri, 01/20/2006 - 6:38pm

How does your methodology account for the relative value of "clock time." The advantage provided by running the ball repeatedly to control the tempo of a game and the number of offensive opportunities provided to the other time - vice the disadvantage of a passing attack that can result in drives that may be sucessful, but consume little or no clock time.

by Clod (not verified) :: Fri, 01/20/2006 - 8:05pm

Re: 19

Why thank you!!

by garrett (not verified) :: Sat, 01/21/2006 - 3:49am

pats won 2 bowls with antowain.
enough said

by TRhere (not verified) :: Sat, 01/21/2006 - 3:58pm

If the Steelers win the Superbowl ranked #5 , does it mean the DVOA system isn't worth the digital type it is published in?

The Steelers dominated the game until they got a substancial lead at which point they went into "safe" mode, ran the ball to eat the clock, and played prevent on defense.

I wonder if good play calling is at all a factor in the DVOA system. If not than the system can be said to be a good analysis of tactics and execution, but a poor analysis of strategy.

by Jim (not verified) :: Thu, 01/26/2006 - 5:31pm

We don't need a DVOA to tell us anything. The Steelers are the best team in the AFC because they won all the games that counted, and the Seahawks are the best team in the NFC for the same reason. And soon we will know who the best team in the NFL is. If the Steelers win the DVOA will have been wrong in every game they played. Any system that gets the entire road to the Super Bowl wrong is obviously broken. I have a pet bird that has only gotten one playoff game wrong so far. But I don't sing its praises and I certainly don't publish its results.

by mike (not verified) :: Tue, 02/07/2006 - 6:26pm

Please update the Power Rankings already, we've been waiting since the week after the Colts lost!! I want to see Pitt on top!!