Writers of Pro Football Prospectus 2008

Most Recent FO Features


» 2017 Adjusted Games Lost

Two NFC teams were hit hardest by injuries last year. One already set the AGL record in 2016, while the other has a coach with the worst AGL since 2002. Also: the Rams' incredible bill of health in L.A., and Tampa Bay's questionable injury reporting.

18 Jan 2010

Week 19 DVOA Ratings

by Aaron Schatz

The Saints jump from 12th to second in our playoff-adjusted weighted ratings, which is a pretty good sign of just how closely packed the top teams in the NFL were this season, as well as just how much the Saints dominated the Cardinals in this week's playoff game. In addition, since weighted DVOA is meant to lower the strength of older games, these ratings do not include Weeks 1-5, and Weeks 6-11 are somewhat discounted. Because of our formula, a couple of the Saints' closer wins from midseason dropped in weight this week, particularly the Week 7 win where Miami dominated the Saints in the first half before the Saints came back to win 46-34. With that game dropping in weight from 67 percent strength to 20 percent strength, the Saints' defensive DVOA really improves. The Saints are now the highest remaining playoff team in weighted DVOA, followed by the Colts, Vikings, and Jets. The playoff odds report is also updated through this weekend's games.

All 32 teams are listed in the ratings below. As we've noted in years past, these ratings include every game from the past 14 weeks, and so teams that are now out of the playoffs are going to end up higher than teams that are still alive. The best teams over the last two weeks are not necessarily the best teams over the past three months. If you want to see the actual regular season final ratings, click here.

Finally, in response to some of the comments left in discussion threads in recent weeks: There have been some complaints that our formulas overestimate the importance of turnovers. We've fiddled with the value of turnovers a few times over the years. Based on previous years, at least, lowering the penalty for turnovers would make the ratings less predictive, not more. As for the effect where the NFC East teams were ranked highly despite poor records against strong teams in other divisions: I'm perfectly willing to consider a concrete suggestion for how the formula could be improved so that our ratings for the NFC East teams better matched the results of wins and losses. However, as I often say, such a suggestion would have to improve the formula for all teams in general in all seasons. Often, suggested solutions to a problem with certain teams in one year will make the formula less accurate in most other years and with most other teams. Constructive criticism is always appreciated over un-constructive criticism, and we're not going to consider your ideas if your comment reads like the angry troll hatred Mad Lib below, but I think we have a good record of considering reader suggestions. Also, we're much more likely to try your suggestion if you send in an e-mail (to mailbag-at-footballoutsiders.com) rather than leave it in a discussion comment. It's easy to miss a concrete, useful idea in the middle of a discussion thread, and even easier to "star" an good e-mail idea and come back to it in February when the season is over.

* * * * *

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>

If you are new to our website, you can read the explanation of how DVOA is figured here. As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE. These will be the final DVOA ratings given for all 32 teams; next week, we'll just do the single-game ratings for the conference championships in Audibles at the Line, and then spend the next two weeks focusing on specific matchups of the final two teams.

1 BAL 34.7% 1 10-8 12.3% 12 -17.6% 2 4.7% 4 34.7% 1
2 NO 33.3% 12 14-3 33.1% 1 0.4% 14 0.7% 18 27.7% 3
3 GB 30.1% 4 11-7 31.8% 3 -6.2% 7 -7.9% 32 29.0% 2
4 IND 25.4% 6 15-2 22.8% 4 -3.5% 9 -0.8% 22 14.2% 12
5 PHI 24.0% 11 11-6 11.3% 14 -8.5% 4 4.1% 5 24.0% 4
6 SD 23.6% 3 13-4 32.3% 2 7.6% 23 -1.1% 24 20.6% 7
7 DAL 20.2% 2 12-6 19.6% 6 -1.2% 12 -0.6% 21 20.2% 8
8 MIN 19.8% 14 13-4 16.2% 10 -2.4% 11 1.2% 15 19.8% 9
9 CAR 19.8% 5 8-8 11.6% 13 -7.9% 6 0.3% 19 23.8% 5
10 NE 18.3% 10 10-7 18.7% 7 2.1% 15 1.7% 14 18.3% 10
11 PIT 17.6% 13 9-7 16.8% 8 -2.9% 10 -2.1% 26 17.6% 11
12 NYJ 17.2% 7 11-7 -4.3% 23 -18.9% 1 2.6% 9 20.8% 6
13 ARI 15.2% 8 11-7 20.8% 5 4.7% 18 -0.9% 23 13.5% 13
14 HOU 12.5% 9 9-7 16.5% 9 5.2% 20 1.2% 16 12.5% 14
15 DEN 9.6% 18 8-8 -1.3% 20 -7.9% 5 2.9% 7 9.6% 15
16 SF 6.0% 15 8-8 -6.2% 24 -12.3% 3 -0.1% 20 6.0% 16
17 TEN 5.8% 16 8-8 16.2% 11 12.3% 27 2.0% 12 5.8% 17
18 NYG 3.8% 22 8-8 10.2% 15 4.8% 19 -1.6% 25 3.8% 18
19 MIA 1.7% 17 7-9 1.6% 18 3.1% 17 3.2% 6 1.7% 19
20 WAS -1.5% 19 4-12 1.2% 19 0.4% 13 -2.2% 27 -0.1% 20
21 ATL -2.2% 21 9-7 5.4% 16 5.4% 21 -2.3% 28 -2.2% 21
22 CIN -3.6% 20 10-7 -2.6% 22 2.8% 16 1.8% 13 -6.1% 23
23 BUF -12.4% 23 6-10 -19.3% 29 -4.3% 8 2.6% 8 -2.8% 22
24 CLE -12.8% 24 5-11 -1.8% 21 18.8% 30 7.8% 1 -12.8% 24
25 JAC -13.1% 25 7-9 4.7% 17 14.1% 28 -3.7% 29 -13.1% 25
26 CHI -19.4% 27 7-9 -15.1% 27 6.7% 22 2.4% 10 -19.4% 26
27 OAK -19.5% 28 5-11 -11.3% 26 10.3% 24 2.1% 11 -19.5% 27
28 TB -20.7% 26 3-13 -16.1% 28 11.1% 25 6.5% 2 -20.7% 28
29 KC -27.2% 29 4-12 -8.8% 25 11.9% 26 -6.5% 31 -27.2% 29
30 SEA -37.3% 30 5-11 -24.0% 31 14.3% 29 1.0% 17 -37.3% 30
31 STL -46.0% 31 1-15 -31.7% 32 19.2% 31 5.0% 3 -46.0% 31
32 DET -48.2% 32 2-14 -23.6% 30 20.6% 32 -4.0% 30 -48.2% 32

Here are the one-game DVOA ratings for the second round of the playoffs. Although a number of people have compared the Chargers' loss to their loss to the Patriots in 2006, our DVOA ratings tell a very different story. DVOA says the Chargers dominated the Patriots in that 2006 game, and lost because of horrendous luck. But the Jets end up with the better DVOA in this week's game, even when we remove opponent adjustments, although their performance certainly can't hold a candle to the dominating victories by New Orleans, Indianapolis, and Minnesota.

DVOA (with opponent adjustments)
NO 85% 53% -6% 26%
ARI -77% -7% 43% -27%
IND 59% 16% -43% 0%
BAL -5% -31% -19% 8%
MIN 75% 8% -57% 10%
DAL -67% -44% 7% -16%
NYJ 12% -17% -18% 11%
SD 4% 25% 4% -17%
VOAf (no opponent adjustments)
NO 75% 49% 0% 26%
ARI -103% -13% 63% -27%
IND 36% 2% -34% 0%
BAL -24% -35% -3% 8%
MIN 52% 6% -36% 10%
DAL -88% -49% 23% -16%
NYJ 1% -8% 3% 11%
SD -17% -8% -7% -17%

* * * * *

As explained in last week's commentary, the "playoff weighted" DVOA ratings above do not include offensive and defensive plays from the following portions of Week 16-17 games:

  • Week 16, Colts-Jets, after Donald Brown's touchdown with 10:20 left in the third quarter.
  • Week 17, Bengals-Jets, second half.
  • Week 17, Cardinals-Packers, final three quarters.
  • Week 17, Chargers-Redskins, after Philip Rivers throws a touchdown pass to Antonio Gates with 6:42 left in the first quarter.
  • Week 17, Colts-Bills, final three quarters.
  • Week 17, Panthers-Saints, the entire game.

These plays are included in the red "standard weighted" ratings.

Posted by: Aaron Schatz on 18 Jan 2010

159 comments, Last at 08 Sep 2010, 2:51pm by The Peepshow


by ammek :: Mon, 01/18/2010 - 2:11pm

DVOA says (after the event) that the Chargers and Jets are about as closely matched as you can imagine. The #2 offense and #1 defense. The #23 offense and defense. Weighted DVOA within one-fifth of a percentage point. It ought to have been a close game.

Oh, it was.

by Carl (not verified) :: Mon, 01/18/2010 - 2:12pm

What's up with the highlighting in the one-game chart? Maybe it's intended to ease comparison of the two teams that played each other, but it's confusing to have the same highlight colors as in the main chart but used for a different purpose.

by whodat :: Mon, 01/18/2010 - 2:36pm

As long as you're working on a playoff weighting, I wonder the impact of considering personnel on the field beyond just meaningless games.

I appreciate the Saints' defensive leap in the formula occurring due to the drop of the MIA game, yet the media accounts say Greer going out in the second half of the season forced Williams into different schemes viz Sharper.

Disclosure: all for piling Pelion on Ossa when it enables Saints fans to indulge in SB fantasies.

by NHPatsFan (not verified) :: Mon, 01/18/2010 - 2:36pm

Is there one "ideal" football strategy, for a given domain of rules and rule-interprestation, to which teams should converge? Or is the game itself strategically dynamic?

Round 1: Holy Crap I got no o-line, Joe, it's gonna be all three-step drops and slants and flats, got it?
Round 2: Hey, this "West Coast" offense works? Hey let's all run the West Coast Offense!
Round 3: We've gotta gear our defense to stop the West Coast offense
Round 4: You know, if everyone is doing x, y, and z to stop the West Coast Offense, that means we can do q, r, and s and their personnel are going to have a hell of a time stopping it!

Round 5: Hey this spread offense works?...

ad infinitum, ad nauseam...

I believe -- admittedly more as gut instinct than with data, that strategy is adaptive.

I bring this up because I think it matters with respect to the FO model -- some of your model parameter values may be not intrinsic to the game of football but inherent to the current mix of strategies in use. As the strategies shift, the parameter values perhaps need to as well.

What this really means is that the predictive value of a change in the model might be time-bound. Single-season tweaks are cheating, of course, but something that's true for 5 seasons might not be true for 20.

Put in another light, I think the DVOA model undervalues and underappreciates strategic divergence (nor do I think that's something that can be built in to it).
In more concrete terms, my hypothesis isn't that the Jets are playing football "the right way," but that they're playing significantly differently -- smashmouth has been out of style for a while -- than most everyone else.

Semi-disjointed post, I know but I thought it was interesting

by DaninPhilly (not verified) :: Mon, 01/18/2010 - 3:08pm

I agree the DVOA model is biased in favor of consistent play over big play, hense bias in favor of West Coast Offenses (hello, Eagles). In defense of DVOA, that generally will win a lot of games. However, one model trying to quantify 32 teams over a 16 game schedule, with a total of tens of thousands of plays imput into it is going to have some quirks in it.

Once you get to a one game situation, the factors of game planning, which DVOA can help understand but cannot totally predict, will come to throw more and more monkey wrenches into any statistical system. Which is why narratives can still be useful, even if DVOA were perfect.

To me, what DVOA can do is give us fans who don't have the time to pour through hours and hours of tape each week a shorthand way of analyzing the matchups for upcoming games. We can see that the Jets are #1 against the pass, but only #8 against the run. We also see that SD is #1 throwing the ball, but #32 running the ball. Therefore we can predict that the Chargers were going to have a hard time scoring any points, giving the Jets a fighting chance to have a few fluke plays to win it. Which makes the result on Sunday, although not predictable, at least conceivable.

By the way, since Indy is much better passing the ball than running it, we might see something similar this next week, although given that Indy's D is a bit better than SD's (#16 vs #23), 14 points may be enough.

by Dales :: Mon, 01/18/2010 - 9:56pm

Excepting for the fact that the Eagles were predicated on the big play this year.

by The Ninjalectual :: Mon, 01/18/2010 - 3:16pm

Wow, great post! I'll bet there is something to that.

However, no matter what style of offense or defense is being played, a teams goal is still achieving or preventing first downs; moving the chains consistently or breaking long plays leads to scoring, which leads to winning. The most basic idea behind DVOA is focused on that most basic foundation of football. But because the modern DVOA formula considers far more factors than that, with the idea that considering more factors leads to greater predictive accuracy, you may have a good point.

Perhaps the vaue of a turnover is greater/lesser due to the modern style of NFL football. Perhaps the value of a completed deep bomb is greater/lesser due to the current climate of relative parity and the emphasis on the passing game. This is the kind of thing that should become more clear as DVOA extends back past 1994. Once we have the data for the entire history of the NFL, I think we should see increased predictive accuracy. What an interesting question the ponder in the meantime...

"Just look at that pumpkin."
-John Madden, looking at the moon.

by AnonymousA (not verified) :: Mon, 01/18/2010 - 4:18pm

1) Football is stochastic. This makes measuring success difficult, because a proper model is needed, rather than a pure observation of a small set of outcomes. In the vernacular, "any given Sunday".

2) Football is hard. No one knows the "perfect gameplan", "perfect draft strategy", etc. That doesn't mean it doesn't exist -- chess is hard, too, but we know that a perfect strategy MUST exist.

3) There are two kinds of statistics: descriptive (the "what"), and predictive (the "why"). DVOA is an attempt to be predictive -- not to say "who won", but to say "who played better, and is more likely to win next week". Predictive statistics, however, can be subdivided into empirical and theoretical statistics. A great example of this comes up in Brian Burke's win probability model, which is empirical. When a crucial situation arises (for example a 4th-and-1 from midfield down 3 with 3 minutes left and all your time outs), the model is based on what has happened in the past -- teams that have this situation have won X% of the time. However, coaches are frequently overly-conservative in these sorts of situations. The probability his model produces, then, accurately describes what will happen, but not what "should" happen if all coaches were as aggressive as they should be to optimize their chance to win.

What you're suggesting, then, is a highly empirical model. This has both advantages and disadvantages. Two of the key advantages are that it tends to be easier to build while also being more accurate. Unfortunately, it also tends to break badly and require changes to the model when something new happens, and have trouble making valuations (is the wildcat a "good strategy" or has the league been caught by surprise? Empirical models are biased towards the former).

by JetfanMike (not verified) :: Mon, 01/18/2010 - 4:58pm

I'm a statistics idiot, but why must a perfect chess strategy exist?

by Eddo :: Mon, 01/18/2010 - 5:10pm

I'm a chess idiot, but I can't imagine that is true that a "perfect" strategy exists. Isn't Game Theory the idea that there can never be a "perfect" strategy when you're facing a rational opponent? That randomness/variation in strategy is actually the best way to win?

by Arkaein :: Mon, 01/18/2010 - 5:23pm

Chess is a completely deterministic game, i.e., making the same moves in the same situations will always give the same result.

A perfect chess strategy could in theory look at all possible board states, and look at all moves from these board states, and evaluate all possible outcomes, all the way to all possible end conditions, and make decisions that will always win, or at least will never lose (two perfect chess players would draw every game).

Perfect doesn't necessarily mean it will win every time, rather it will always create the best possible outcome given the moves made by an opponent.

by Eddo :: Mon, 01/18/2010 - 5:38pm

I ask this in true curiosity: doesn't that idea assume the opponent will make certain counter-moves? Doesn't game theory then dictate that he should behave differently, thus altering your perfect strategy? That is, the perfect strategy only works when you know what moves your opponent will make?

by DGL :: Mon, 01/18/2010 - 5:47pm

Simplify to tic-tac-toe. Whether your opponent makes "good" or "bad" moves, it's possible to have a perfect tic-tac-toe strategy because you can analyze every possible move, every possible counter, every possible counter to the counter, and so on to the end of the game, because the game is deterministic. If you start out with an X in the center square, and your opponent knows that your perfect strategy assumes he'll put an O in a corner square, he can behave differently and put an O in a side square. You'll then make a different move than you would have made - but you are still playing the "perfect" strategy.

The same goes for chess; it's just computationally harder...

by Eddo :: Mon, 01/18/2010 - 5:50pm

This makes a lot of sense to me. Thanks guys.

by Nathan :: Mon, 01/18/2010 - 9:10pm

Actually, if you put an O in a non corner square on move 2 in Tic Tac Toe (assuming a center move to start) you lose every single time. Which I think further proves your point.

by Key19 :: Mon, 01/18/2010 - 10:51pm

Not if your opponent is incompetent.

by billsfan :: Tue, 01/19/2010 - 10:44am

Fine, simplify things further with Global Thermonuclear War instead of Tic Tac Toe.

(I also like the Eagles)

by Theo :: Wed, 01/20/2010 - 8:14am

The only winning move is not to play!

by RickD :: Tue, 01/19/2010 - 1:31am

"Perfect doesn't necessarily mean it will win every time".

That's only because chess is likely a draw game. But it's not really proven that "two perfect chess players would draw every game". It may be the case that the advantage of white means that a perfect player would always win as white.

by cfn_ms :: Wed, 01/20/2010 - 12:59am

And it's at least theoretically possible that a "perfect player" would actually always win as black, though that would be a REALLY surprising result. Ultimately, the decision trees are sufficiently complicated that it's basically an unanswerable question. My suspicion is that two perfect players would draw, but really, who the hell knows.

by NHPatsFan (not verified) :: Mon, 01/18/2010 - 6:06pm

After further reflection, what I'm suggesting is to ask how we can - or even if we should - ask some questions about the about the DVOA modeling parameters / assumptions.

First, assuming football strategy is adaptive, what effect, if any, does change in the current mix of strategies in use have on the assumptions / parameters built into the DVOA model (e.g. "Fumbles don't matter much")?

Second, if there is an effect, how can it be measured and bounded in time without the process degenerating into special pleading?

Third, (and this one is not one that I think is a new thought, I assume this is a continuing process for Aaron and the others) what are the implicit assumptions made in the model that need to be identified and made explicit, and tested?

by Will Allen (not verified) :: Mon, 01/18/2010 - 2:37pm

The Vikings can beat the Saints, obviously, but for that to happen the Vikings offensive line is going to have to make the sort of leap in performance that the the Vikings defensive line did yesterday. The Vikings really didn't block all that well yesterday, albeit against a defensive front which is superior to the Saints'. If the Vikings can't make the Saints safeties provide run support with some frequency, it'll be really hard for them to outscore the Saints. I just can't see the Vikings having anything close to the success they had in rushing Romo, when it comes to pressuring Brees.

If Ray Edwards is hurt (I don't want to ruin my enjoyment of yesterday's win by checking the injury report), that will make the Vikings' task significantly more difficult.

by Rick A. (not verified) :: Mon, 01/18/2010 - 2:53pm

Check the post I left in answer to you on the Audibles article...

The Vikes rush game should get quite well vs the Saints D Line. AP will likely rip them a new ***hole. Now, on the other side of the ball, the N.O. O Line, contrary to popular opinion, is superior to the Cowboys, that's a fact--but NO ONE can control the Vikes D Line when they are focused. You call it a "leap in performance", I call it the predictable result of giving them a week of rest. The public has zero conception of what a draining thing an entire NFL season is, not even a team's fans will cut some slack to their guys. And these owners want to make it a 17 or 18 week season now. But that's another subject. Anyway, Mr Allen, there is another Mr Allen in your town and N.O. will have their hands full with him, to say the least...

by Will Allen (not verified) :: Mon, 01/18/2010 - 3:23pm

Well, the Cowboys have been a crappy pass blocking team for the better part of a decade now, which got covered up a little bit this year with scheme. In a noisy road environment, however, against good pass rushers, they reverted to form. Again, to give myself a break, while acknowledging that my pessimism was obviously unwarranted, I did also say last week that having Pat Williams regain some youth was a key to a Vikings victory, and as dramatic as Edwards' and Allen's play was, it sure did help to have Pat collapsing pockets again. Eight more quarters, Pat, eight more quarters.

I agree the Vikings offensive line could gash the Saints, and force their safeties to do what they are normally lousy at. Not having the home field will make rushing Brees a bigger task, even before factoring that the Saints are much better pass blockers than the Cowboys. Having an extra point blocked against the Bears could end up being a very big play in the grand scheme of things.

by Kevin from Philly :: Mon, 01/18/2010 - 4:18pm

Losing their starting LT with no viable backup probably didn't help the Cowboys either.

by Will Allen (not verified) :: Mon, 01/18/2010 - 4:25pm

Sure, it hurt, but Flozell Adams was getting whipped as well, and if it weren't for the zebras deciding to look the other way (really, have you ever seen an offensive line get that outclassed without drawing a holding penalty?), he would have been flagged two or three times in the first quarter, and the Cowboys would have been even more futile looking on offense. What was really weird was that it was crew that typically calls things close.

by Aero Gopher (not verified) :: Mon, 01/18/2010 - 5:26pm

Will, appreciate your comments in FO. You're a Vikings fan and you try to make good comments without wearing purple shades. 1) I am concerned about the Vikings pass defense against the Saints and Sharper's past successes against Favre. Hopefully, Mr Frazier can give some team more reasons to hire him and come up with a design to slow down Brees & Co. 2) The Vikings have had much success against the Saints however and I'm hoping that the Vikings running game gets into gear. 3) Much of the Vikings running game is based on long drawn out plays. I hope they try some quick hitters to try to pop AD into the secondary.

by jmaron :: Mon, 01/18/2010 - 5:46pm

Coach Brad Childress typically doesn’t divulge injury information until submitting his Wednesday injury report. Speaking Monday, Childress said: “Ray’s a tough guy. If he can possibly go, he’ll go.”

That's Childress speak for "there's not a chance in hell he's playing"

by Rover (not verified) :: Mon, 01/18/2010 - 2:40pm

Any thoughts on Wall Street Journal's Saturday column where they use super secret game day stats and a first down based scheme to predict games? They had the Jets winning by 10, but otherwise were pretty close to the betting lines. Same with last week, except for predicting a narrow GB win (which I count as a good prediction). Small sample size, etc, but interesting approach.

by RickD :: Tue, 01/19/2010 - 1:33am

"They had the Jets winning by 10".

Sounds like a bunch of drunken Wall Street traders.

by are-tee :: Mon, 01/18/2010 - 2:48pm

Most inaccurate FO writer's observation of the year:

"The Jets don't have an elite defense. You can throw downfield on them." -

Biil Barnwell, commenting on Matt Schaub's performance in Week 1 Quick Reads.

by Rick A. (not verified) :: Mon, 01/18/2010 - 2:56pm

Along with many, many, many, many, many other proclamations...

by C (not verified) :: Mon, 01/18/2010 - 8:59pm

Reminds you of one of those old NFL get your story straight commercials...

The Jets don't have an elite defense... you could throw on deep on them...

by Treima (not verified) :: Tue, 01/19/2010 - 4:20am

I don't get it, Rick.

If you dislike what everyone says around here, you dislike DVOA, and you're only around to glorify your own predictions, why the heck are you even here?

by bravehoptoad :: Tue, 01/19/2010 - 3:05pm

Rick never makes predictions. He only comments on things after they've happened, berating everyone for not seeing why they were obvious.

If Rick ever made a prediction, he'd run the risk of being wrong, and that's something he'll never do.

by Rick A. (not verified) :: Wed, 01/20/2010 - 2:36am

Ah, the hotdog returns. Where has your big mouth been ? I guess you never read my predictions as far back as week 6 regarding Minny and Indy. Or my recent, pre-playoff predictions that the Jets would make it to the Championship game. You continually prove that you cannot read, as you are doing this week, as it is plainly obvious that I do not berate everyone at all. There are many clear thinkers on this site, you are not one of them. Just this week I complimented a few people. I never came to this site with any sort of rancor, Mr Cowardly Hotdog, just polite discourse, but was attacked very quickly by you and some others of your ilk. On the prediction front, I saw many weeks ago that you couldn't pick your nose. But it is on the record, in the archives, those forecasts I made in plain English--some of which were to answer your lame ass, and some of which would've been there anyway in the course of intelligent communication with some of the sentient beings around here...

by Rick A. (not verified) :: Wed, 01/20/2010 - 3:10am

Where do you get the massive assumption that I dislike what EVERYONE around here says ? You have come late to this game I think.

As to DVOA--I come from the sports world, specifically the world of pro football. Had you been here earlier this year, or in year's past, you would've read my words about how I came to learn a few years ago of this invented "system" which attempts to quantify the business of the sport of American Football. It's a hoot watching it, watching it's interaction with it's followers. Alot of it is pure nonsense. Most of it is about making money. Some of it is very good, as far as the whole "power rating" concept goes, which is not really very far in the real world. This year it was pretty far removed from reality, obviously. It's also a hoot to watch you guys trying to figure out how it was so wrong etc etc.

As to some effort to "glorify" my predictions--this Hoptoad toad underneath my post came with alot of adolescent challenges earlier in the year, voicing what he and other similar minded turds, I mean toads, were thinking. So I just started being a little more generous with my analysis, giving a bit of it to those out there who, unlike the Hotdog, are open minded, intelligent, polite, and looking to learn. I'm sure a certain percentage of the people here are betting and looking for DVOA to give them an edge. Yeah right. It'll give them a ticket to the poorhouse. And others are playing in Fantasy leagues and all that. Perhaps my steadfast and knowledgeable insistence that the Vikings were the most talented team in the NFC, and the Colts the most talented in the AFC, may have been of assistance to them. Most of them, though, were all caught up in their cult devotion to DVOA ratings and their own ignorance of football. They are stat geeks, living in the numbers. Once the Jets made it in I warned them all that the Jets were going to create some havoc and that I certainly saw them with the possibility of getting through 2 games, at least. All of this is in the archives. But they wanted to ridicule the Jets and profess this nonsense regarding how allegedly DVOA has revealed this and that "mythology" of past paradigm's of analysis. Yeah right...

Now, of course, Minny may get beat, and Indy may get beat--but at least I did go on the record, regardless of the ravings of Mr Hopturd, who, by the way, I don't recall ever putting anything forward in the way of evidence regarding any supposed knowledge that his arrogant demeanor implies he has. And my analysis throughout this season has accurately called two of the final four teams. These calls of mine were made in week 6, BEFORE Minny's second beat down of Green Bay, and when everyone was going on about "how does Indy do it ? these close calls week after week?" Also, when DVOA was, each and every week, rating New Orleans, New England, Philly, and etc, as "the most efficient" teams. Yeah right. New England couldn't win a road game until Buffalo. Gimme a break. So does that satisfy you ? It's the wee hours, unlike alot of people on this thread I have a life. I catch up with all this when nothing else is cooking. Every now and then there is actually something of value on it for me besides the "hoots". But that's my business as to what it is. Now why do you come to these threads pal ?...

by Rick A. (not verified) :: Wed, 01/20/2010 - 3:12am

By the way, this last post was in answer to someone named "Treima"...

by C (not verified) :: Wed, 01/20/2010 - 10:48am

I respect ya. Do you think the OVER hits in the Vikings/Saints game?

I'm thinking Colts/Vikings super bowl.

by Kevin from Philly :: Wed, 01/20/2010 - 3:03pm

Back when I first got into losing money...I mean betting, I remember a game between the GSOT Rams and Raiders. The O/U was about 60, and everyone I talked to said that'd go over by half time. Of course, the total ended up around 55. This game feels the same to me.

by Led :: Mon, 01/18/2010 - 3:13pm

To be fair, that was a true statement of the Jets defense last year. Barnwell could've said it's UNLIKELY the Jets have an elite defense given their performance last year and marginal off season improvements, and that would've been perfectly reasonable. But careful, reasonable comments are less interesting and harder to fit into a chart.

by Nilblog (not verified) :: Mon, 01/18/2010 - 2:52pm


It's a bruise, he'll be OK.

by Bobman :: Mon, 01/18/2010 - 3:09pm

rub some of those field turf little black rubber pebbles on it and get back out there.

by The Ninjalectual :: Mon, 01/18/2010 - 3:20pm

never underestimate the healing power of those black rubber pebbles

by Bobman :: Mon, 01/18/2010 - 8:59pm

chemically inert and probably antiseptic, and just slightly static-clingy. Never come out of socks or leg hair. They are truly a miracle substance. Annoying as all hell, but if they heal, well, who am I to argue with them?

by billsfan :: Tue, 01/19/2010 - 10:48am

They're just recycled tires.

(I also like the Eagles)

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 3:03pm

How in Hell are the Cowboys ahead of the Vikings?

Sweet system, fellas.

by Aaron Schatz :: Mon, 01/18/2010 - 3:17pm

No, no. This won't do. Doesn't follow proper angry troll Mad Lib form. Try again.

by The Ninjalectual :: Mon, 01/18/2010 - 3:32pm

Dallas is clearly ranked too high because they cried like little babies on Sunday. Troy Aikman's memories from 20 years ago are way better than this. As Troy would say, "I was unaware that there were 32 teams in the NFL. That's 31 more than necessary, if you ask me. GOO BOI-ZZZZZE!"

"Just look at that pumpkin."
-John Madden, looking at the moon.

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 3:26pm

No. No I won't.

How. In. The. Hell. Does. YOUR. System. Say. The. Boys. Are. Better. Than. The. Vikes.

It's your system, man.


by The Ninjalectual :: Mon, 01/18/2010 - 3:36pm

Have you considered that perhaps the Cowboys are better than the Vikings?

"Just look at that pumpkin."
-John Madden, looking at the moon.

by Three seconds' worth of effort (not verified) :: Mon, 01/18/2010 - 3:41pm


It's right there at the top, you know.

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 3:46pm

Read it. Doesn't explain how team A can lose 34-3 to team B, and be rated higher. Almost invalidates the whole thing....

by coboney :: Mon, 01/18/2010 - 3:53pm

In very simple words.

Over the second half of the year, the system says that Dallas played better on a per play basis. This is adjusted for opposition.

In a single game, Dallas completely tanked. They played their worst game by far. Minnesota played one of their best 3 games. They are very similar tiered teams, separated by a small amount.

It was one game. Which, in mathematical terms, is a small sample size. A lot of things can happen in that small sample to toss it off. A handful of plays change a games tenor completely.

In the future, please use proper troll formating though.

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 3:58pm

So in simple terms, the system's flawed.


by Will Allen (not verified) :: Mon, 01/18/2010 - 4:01pm

Earth to TheChadHenneMeme: Everything is flawed. I know, I know, shocking, isn't it?

by The Ninjalectual :: Mon, 01/18/2010 - 4:05pm

Perhaps Aaron owes you an apology because his system is merely "very good" and not perfect. We now understand that TheChadHenneMeme demands impossible-to-achieve perfection!

"Just look at that pumpkin."
-John Madden, looking at the moon.

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 4:09pm

The Saints jump from 12th to second in our playoff-adjusted weighted ratings, which is a pretty good sign of just how closely packed the top teams in the NFL were this season, as well as just how much the Saints dominated the Cardinals in this week's playoff game.

Yet, the Vikes can't hop over the Cowboys?

I ain't sayin' the Emperor's naked; I'm just sayin' his wardrobe is out of fashion.

by C (not verified) :: Mon, 01/18/2010 - 9:05pm

Mr Chad Henne.

Don't say anything is wrong with the system... You have some teams jumping up ( to save face) and some last minute wheeling and dealing with the DVOA. Just live with it. They will give you the standard " use the boiler plate complaint" response to slight your comment.

If you are a Vikings fans be happy that the NY Giants weren't even in the top 1/2 of DVOA rankings and were the "most likely team to get blown out" every single round of the playoffs the year they won it all, and the Patriots were the "bestest team ever".... The following year the Patriots missed the playoffs and the Giants were the #1 seed.

There are two possibilities... the entire NFC East looks overrated, and your eyes just confirmed what your brain said was right. Remember the team that wins in the real world gets the trophy, there is no trophy for having the #1 dvoa.

by Dales :: Tue, 01/19/2010 - 8:10am

Or, perhaps, on average, teams that stomp the living hell out of bad teams tend to be teams that will win against other good teams.

And, either that average is randomly distributed, with it a fluke that this year the four NFC East teams which stomped bad teams happened to just lose more than would have been predicted, or

The average is not randomly distributed, and there are characteristics that have yet to be incorporated into the model that, if they had been, would have caught them as not the kind of high-stomp teams that would have enduring success.

by C (not verified) :: Tue, 01/19/2010 - 10:10am

There are some teams that have talent and can stomp inferior teams but that will lose over and over and over again against other talented teams. Different teams have different characteristics.

Some teams can't play on the road
Some teams can't play on grass
Some teams play down to their competition
Some teams just overtalent other teams

Some teams are the opposite of "playing down to their compeition", they destroy weak teams, but can't beat good teams. There is a name for that sort of team and it's after a similar fish I can't put the name on.

by Paul R :: Tue, 01/19/2010 - 11:13am

The New England Patriots.

by TheChadHenneMeme (not verified) :: Tue, 01/19/2010 - 10:50am

and there are characteristics that have yet to be incorporated into the model

That's what I'm saying.

Sorry to crash the circle-jerk, but I'm just saying if Philly's number 5 and ahead of Dallas, who are ahead of the Vikings (along with Green Bay), there's something wrong here. Green Bay and Philly's D was exposed; why isn't that reflected? How come Minnesota's offense didn't get a bigger jump after what they did to Dallas's D, but the Saints go from 12 to 2 after scoring 45 against Arizona? Any weight given to Arizona abandoning the run that was working pretty, pretty, pretttttaaaaayyyy well?

I don't know what could be done, I'm not the staistician. Maybe more weight should be given to playoff wins is the simplest idea I have.

Completely fine with Baltimore being one, though. Get Marshall or Boldin, already.

by RickD :: Tue, 01/19/2010 - 1:37am

No, dummy. The system measures the level of the team over the entire year. If you only care about how they played on Sunday, read the box score and leave us alone.

by The Ninjalectual :: Mon, 01/18/2010 - 4:01pm

Both the Chiefs and the Raiders beat the Broncos this season. Are they both better teams? So for every upset, we were just wrong, and the better team always wins?

"Just look at that pumpkin."
-John Madden, looking at the moon.

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 4:05pm

That would imply the Vikings upset the Cowboys.

by andrew :: Mon, 01/18/2010 - 4:19pm

On the BS report, Aaron said that while DVOA... (and his gut instinct) favored the cowboys, when they ran the numbers on individual breakdowns, they favored the vikings.,

by C (not verified) :: Mon, 01/18/2010 - 9:06pm

Nice, play both sides of the fence. That way you can't be wrong. I didn't even see actual picks... Just a break down.

by cfn_ms :: Mon, 01/18/2010 - 3:48pm

It's a nice, detailed explanation of the methodology, but it doesn't specifically answer why the model likes Dallas better. Some level of "here are the things DVOA especially likes about Dallas" or "here are the things DVOA especially dislikes about the Vikings" would be useful IMO.

by cfn_ms :: Mon, 01/18/2010 - 3:46pm

A partial answer is that the system had Dallas WAY ahead of Minnesota before that game, and that just one game wasn't enough to overcome the gap. However, I didn't really get why Dallas was that far ahead of Minny in the first place. I'm curious about that one too (though minus the hostile language).

P.S. I think that, in general, it would be a good idea to come up with a few things that "look weird" (i.e. Dallas over Minnesota) in your numbers and take the time to explain them in your comments, either agreeing with them or saying something to the effect of "this looks off, and here's why".

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 3:51pm

Agreed with this guy - though minus the kind language.

It'd be better if they did that, instead of stick out a bunch of stats and act like's it gospel. I get why Green Bay and Philly are rated high, I didn't say anything about that - though, you know, I Could Have.

But Dallas ahead of Minnesota? Yeah, some explanation would've helped.

You can put it in a Mad Libs for me.

by The Ninjalectual :: Mon, 01/18/2010 - 3:56pm

Who's acting like it's gospel?

by Will Allen (not verified) :: Mon, 01/18/2010 - 3:58pm

It is only a figment of your over-fertile imagination that anyone has said or implied that the DVOA model is anything resembling gospel.

by Rick A. (not verified) :: Mon, 01/18/2010 - 4:15pm

Thank you. Boy these guys that invented this sure have scored some devoted cult members...

by Aero Gopher (not verified) :: Mon, 01/18/2010 - 6:14pm

My prayers are that the Vikes win their last game. I don't care if they are the top DVOA team. Maybe they get a fluke call or two along the way. I believe that the ball has to bounce right during the season a time or two or three to win the Super Bowl as well as be a talented team.

by displaced_saints_fan :: Mon, 01/18/2010 - 4:13pm

DVOA is a rating based on many game outcomes; it is a weighted average, but surely you already knew that.

There's this curious but apparently broadly held idea that how "good" a team is can be precisely quantified, and that the final score of a single game is its inviolable measure.

That the Cowboys lost does not prove that the Vikings are absolutely better, only that they were better yesterday (and how). If the Vikings and Cowboys played each other 16 times in a row, few would expect the results to be 16 Viking blowouts. Can you even be certain that the Vikings would win more than eight? Any given Sunday...

Seriously, if you're this worked up about the Cowboys being one spot higher than Minnesota, this is probably the wrong forum for you.

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 4:19pm

DVOA is a rating based on many game outcomes; it is a weighted average, but surely you already knew that.

There's this curious but apparently broadly held idea that how "good" a team is can be precisely quantified, and that the final score of a single game is its inviolable measure.

That the Cardinals lost does not prove that the Saints are absolutely better, only that they were better yesterday (and how). If the Saints and Cardinals played each other 16 times in a row, few would expect the results to be 16 Saints blowouts. Can you even be certain that the Saints would win more than eight? Any given Sunday...
Think about that; but the Saints jump from 12 to 2.

Seriously, if you're this worked up about the Cowboys being one spot higher than Minnesota, this is probably the wrong forum for you.
I ain't worked up at all. I asked a question. I received sarcasm and derision and people on a mathematical high horse (can someone explain sample size again, please). I thought it was a pretty valid question (and how).

Head-to-head should probably mean more. Something to think about for next year, Schatz.

by The Powers That Be :: Mon, 01/18/2010 - 4:34pm

"I ain't worked up at all. I asked a question. I received sarcasm and derision..."

I love when people write stuff like this, claiming that their first post, with phrases like "how the hell..." and "sweet system" was simply asking a perfectly innocent question.

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 4:49pm

It Was a perfectally innocent question. I apologize that tone can't be heard over the internet, but saying "how in the hell" and "sweet system" doesn't necessarily mean I was being facetious. That's how I talk, and write. It's up to you to put a face and emotions and inflections to what I said. Not me. You, and others, took it the way you Chose to take it, and threw aside my valid question.

Now...if someone can come up with a reason for the Saints jumping from 12 to 2 after beating the mighty Cardinals by 31 at home, and the Vikings can still be behind the "probable NFC champion" Cowboys after beating them by 31 at home (and only giving up 3 points), that'd be (gotta watch what I say, feelings get hurt easily here I see (that? That wasn't perfectally innocent)).......anyway, it'd be Just Swell if I could get an explanation.

by Will Allen (not verified) :: Mon, 01/18/2010 - 5:03pm

Actually, it is largely a writer's responsibility to ensure that his intended meaning and tone are conveyed to the reader.

Seriously, do you commonly juxtapose a phrase like "how in the hell" with a term like "sweet system" in that way, without "sweet system" being used sarcastically?

by The Ninjalectual :: Mon, 01/18/2010 - 5:50pm

ChadHenne, if you don't want to put the effort in to write well, perhaps use a smiley face for the benefit of your readers? Those are great as crutches to convey sarcasm, or that no ill will was intended, despite the text, you stupid jerk ;)

by RickD :: Tue, 01/19/2010 - 1:44am

You also said
"Read it. Doesn't explain how team A can lose 34-3 to team B, and be rated higher. Almost invalidates the whole thing...."

You're not "just asking a question".

If you want to understand why the Saints jumped from 12 to 2, you might want to, oh, read the text above the article, where Aaron explains that the weighting of prior games saw a large decrease in the amount of weight for the Miami-New Orleans game. Since this was one of the worst games of the season for the Saints, esp. defensively, having its importance dramatically decrease would naturally see the rating of the Saints increase.

And then there's the part about "how closely the teams were packed together". The ordinal value of the teams doesn't really matter all that much. It can happen that the difference between the 2nd best team and the 3rd best team is much larger than the difference between the 3rd best team and the 14th best team. You are implying that a large ordinal rank implies a huge change in the DVOA value, when Aaron has made it clear that the opposite is true.

by tgt2 (not verified) :: Mon, 01/18/2010 - 4:47pm

Think about that; but the Saints jump from 12 to 2.Ignore ranking. Look at DVOA. The Saints jumped 16% and the Vikings jumped 7%. The model said the Saints were 10% better this week, so we're left with about an 8% discrepancy to explain. If you read the explanations at the top, you'd know much of that difference was probably the Miami game being weighted 40% less this week. Maybe the Vikes and a good game or two that dropped in weighting.

Also, note that though we see DVOA totals for games, total DVOA is not the average of game DVOAs, it's looking at each play of the season equal (weighted by when it occurred). If one team dominates in a 100 play game and gets dominated in a 50 play game, they're likely going to have a pretty good DVOA.

Head-to-head should probably mean more. Something to think about for next year, Schatz.Wins are teh awesome! If you win, you must be better!

The whole methodology is based on correlating how well a team has played with their future liklihood of winning. The actual past W/L record doesn't matter. There's very little information contained in Ws and Ls. All of that information is already included in DVOA.

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 4:54pm

Wins are teh awesome! If you win, you must be better!

Ha....I didn't think it'd be that easy to get this response. Awesome. The internet sure is fun.

Ignore ranking. Look at DVOA. The Saints jumped 16% and the Vikings jumped 7%. The model said the Saints were 10% better this week, so we're left with about an 8% discrepancy to explain. If you read the explanations at the top, you'd know much of that difference was probably the Miami game being weighted 40% less this week. Maybe the Vikes and a good game or two that dropped in weighting.

Any explanation for how the Saints jump 16% and the Vikes only 7%? Based on opponents and weighting, that doesn't make much sense either.

I'm guessing that the Saints running the ball late is the answer; their run game ends up looking better than it actually is because of padded stats, while the Vikings run game looks worse than it is because the opposite - despite all Favre, Rice, Harvin, and Shiancoe have done - refuse to have less than 8 in the box against AP.

by cfn_ms :: Mon, 01/18/2010 - 4:56pm

Head-to-head should probably mean more. Something to think about for next year, Schatz.


I actually disagree with this one. I haven't seen any evidence that H2H has any special predictive value (though admittedly I haven't really investigated it either). If anyone has looked into this I'd love to hear it. People frequently put H2H on a pedestal, and IMO it's severely over-valued (especially in college, but that's another story for another day).

I just thought it was weird that Minnesota had a nicely better record than Dallas and was substantially behind, and that after the Vikings had a huge win and Dallas a huge loss (never mind that they were against one another), Dallas was still ahead.

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 4:59pm

I was makin' a joke with the H2H; I knew that would irritate some stat guys. For a process/formula like this, I don't belive H2H should, or even could, help.

I just thought it was weird that Minnesota had a nicely better record than Dallas and was substantially behind, and that after the Vikings had a huge win and Dallas a huge loss (never mind that they were against one another), Dallas was still ahead.

Pretty much what I was saying, just kinder.

by Sophandros :: Mon, 01/18/2010 - 5:53pm

Minnesota faltered greatly down the stretch and Dallas played extremely well down the stretch. Dallas then went out and dismantled Philly, and that extra game likely gave them enough to maintain whatever lead they developed over Minnesota.

I think that if you compare the two teams' weekly performances and their trends (see here, in the Divisional Preview column: http://www.footballoutsiders.com/game-previews/2010/nfc-divisional-round... ) that should go a long way in illustrating why Dallas, despite getting destroyed by Minnesota, still has a higher rating, as DVOA is based on a body of work larger than just one game.

The late season surge by the Cowboys, combined with the late season swoon by the Vikings is the reason.

Sports talk radio and sports message boards are the killing fields of intellectual discourse.

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 6:16pm

Dallas then went out and dismantled Philly, and that extra game likely gave them enough to maintain whatever lead they developed over Minnesota.

That makes sense: The extra game theory.

You'd think that extra game would've hurt Philly more than it did, though.

by cfn_ms :: Wed, 01/20/2010 - 1:03am

That makes sense. I'm not sure whether I agree or not (need to think about it), but at least I can see the reason.

by Rick A. (not verified) :: Mon, 01/18/2010 - 4:22pm

Needless to say (except to some of these devoted followers) football cannot be quantified, even though it is the most machine-like sport. However, the Vikings are an "absolutely better" team than Dallas and would dominate them in Minnesota until they got bored doing it...

by Goldenfoot (not verified) :: Mon, 01/18/2010 - 5:20pm

If you're talking about Dallas being rated ahead of the Vikings I think you also have to talk about Green Bay being ahead of the Vikings. I know the Vikings tailed off, but sweaping a team does mean a lot. And I don't think Vikings have ever been ranked ahead of Green Bay. Even right after the 2nd win.

by Aero Gopher (not verified) :: Mon, 01/18/2010 - 5:31pm

Actually, the Packers have climbed in the standings. The Vikings were ahead of them earlier in the season. The Packers beat the Cowboys too. Maybe the Vikes learned something from that game.

by billsfan :: Mon, 01/18/2010 - 4:48pm

If you get why Philly is rated highly, then you should understand why the team that blew them out in consecutive weeks is also rated highly. I'm not sure if DVOA accounts for those wins being over a team that had just lost its starting center, and, in the second game, both safeties, but it's pretty clear that it wasn't the same Eagles team we had seen for the first 16 weeks. And then Flozell Adams gets hurt, and can't hold Jared Allen any more.

It's easier with baseball, where you have 150+ games for things to sort themselves out in a statistically reasonable way, but even then, with the playoffs, you see that over, say, only 7 games, it doesn't always work. And that's for a sport that easily breaks down into 1-on-1 matchups. In football, there's much more going on, and it's not all easy to quantify. Teams that do certain things more (and/or) better than average tend to win more games. But 2-3 random events can be enough to lose a game for a better team. Injuries, especially on the offensive line. Fumble luck. 3 misses by a pro-bowl kicker. A catch on a guy's helmet. etc.

Any given Sunday...

(I also like the Eagles)

by TheChadHenneMeme (not verified) :: Mon, 01/18/2010 - 4:56pm

If you get why Philly is rated highly, then you should understand why the team that blew them out in consecutive weeks is also rated highly.
Yeah, I knew. I just enjoy fuckin' with people. Panties, bunches, it's funny to me.

I'm not sure if DVOA accounts for those wins being over a team that had just lost its starting center, and, in the second game, both safeties, but it's pretty clear that it wasn't the same Eagles team we had seen for the first 16 weeks.
I'm guessing it doesn't account for that.

by TheChadHenneMemeIsADouche (not verified) :: Mon, 01/18/2010 - 7:06pm

It Was a perfectally innocent question. I apologize that tone can't be heard over the internet, but saying "how in the hell" and "sweet system" doesn't necessarily mean I was being facetious. That's how I talk, and write. It's up to you to put a face and emotions and inflections to what I said. Not me. You, and others, took it the way you Chose to take it, and threw aside my valid question.

Yeah, I knew. I just enjoy fuckin' with people. Panties, bunches, it's funny to me.

Tahnk you for putting the face on it yourself. You came here to stir people up (you are a fuck) and you were called out on it and denied it (a liar.) So, you are a liar and a fuck.

by Dales :: Tue, 01/19/2010 - 8:15am

Let's count the techniques that were used:


I count five. Maybe more.

by RickD :: Tue, 01/19/2010 - 1:46am

"Yeah, I knew. I just enjoy fuckin' with people. Panties, bunches, it's funny to me."

Come back when you graduate from 5th grade.

by mm (not verified) :: Mon, 01/18/2010 - 3:13pm

Ahh, Carolina fans must continue to wonder how they would have done if Delhomme had gone out earlier.

Can we get the weighted passing and running DVOA for the offense and defense of the final 4?

I think some ways the NFL has evolved recently has made it harder for DVOA to be totally accurate:
1) The ways teams are more willing to use multiple backs or receivers can make an overall numbers more difficult. If you use your best back 80% of the time against a good team, but only 40% of the time against a team that never really threatens you, the 'average' of this performance won't really tell you how the team will perform against a future team.
2) Because most teams now actually have pretty decent passers, you should see if starting to ignore 'blowout' factors increases accuracy for recent seasons. Most teams now have the ability to get good yardage once defenses start to simply guard against the big play, rather than play aggressively. Comparing these teams with teams that keep their defenses aggressive seems like it will provide a bigger bonus than it used to be.
Likewise, it seems like the teams that don't have decent passers are really, really bad in comparison. Penalizing the Raiders or Browns extra in blowouts because their QB can't pass even against prevent defenses will underestimate their ability to win a close game, when their passing game won't be used much. At least in a close games, they have a surprise factor when they call a passing play.
3) It seems like there are more injuries now; whether they keep a player out or merely keep him from playing at 100%. Of course, that could just be my impression

by wr (not verified) :: Mon, 01/18/2010 - 3:16pm

It just seems wierd that Indy's defensive DVOA would be better
than their VOA against the Ravens, even though the numbers from
the first DVOA chart indicate that's what should happen. Guess
the notion of the Nevermores' offensive ineptitude is still strong
in me...

by Led :: Mon, 01/18/2010 - 3:25pm

I think that with Flacco hobbled the Ravens offense was a good deal worse than it was earlier in the year. Given the quality of their receivers, it's a credit to Flacco that they were as successful as they were all year passing the ball.

by dk240t :: Mon, 01/18/2010 - 3:56pm

There are a couple of things that are ALWAYS going to keep something like DVOA from being. The first are random events like the bouncing of the ball leading to turnovers or miraculous catches, unpredictable injuries, people slipping at key times, bad penalty timing, etc. The second is the fact that football is a very strategic game. Sometimes, a coach of a seemingly weaker team will truly find some major tendencies that they can exploit against a truly better team. And sometimes the better team won't be able to figure out on gameday how to counteract this unexpected difference. No amount of looking back is going to predict when this effect will occur.

by cfn_ms :: Mon, 01/18/2010 - 4:21pm

4 teams left. Colts and Saints have elite passing offenses (does Minny also? I don't remember offhand), and the Jets have an elite passing defense. At the least, 3 of the 4 stand out with the ball in the air. Probably more than usual, but I don't think that this sort of trend is going to go away, though "normal" might just be 2 such teams in the final four rather than 3-4.

by huston720 :: Mon, 01/18/2010 - 4:32pm

That is a really big reduction in the value of one game for New Orleans. I'm not sure why the weighted DVOA formula is grouped into blocks of games rather than a per game decrease week to week. I can't remember if this was ever looked at or not. It doesn't seem like it would be too difficult to devalue each game gradually on some linear scale rather than use a stepped scale. Does anyone know if this was looked at in the past, or why a stepped scale is used?

by Aaron Schatz :: Mon, 01/18/2010 - 4:49pm

Let me answer, since I was checking the discussion and this is a serious and reasonable question... I originally used a linear scale, then I did some testing. The percentages I use now were the percentages that did the best job of predicting each team's performance for the remainder of the season. I even re-tested everything last summer with the newer version of DVOA and the changes would have been so small (changing a couple weeks by one or two percent) that I just kept things the same way.

Also, for the person who wondered about why we didn't delve into the reason why the ratings for certain teams were different from conventional wisdom -- actually, that's what I do every Tuesday during the season. I think I addressed the Bengals a couple times, for example. Now that we're down to four teams, I would rather just run numbers and devote the rest of my week to more game charting, watching games with the final four teams, and writing conference championship previews. Plus the "Stat of the Day" project.

by huston720 :: Mon, 01/18/2010 - 5:51pm

Thanks for the answer Aaron, interesting that there is such a drop-off after a certain number of weeks. I can't really think of a reason why a linear scale would work better than what is used, but I'll take your word for it. Anyone have any ideas explaining what might be going on here?

by zip.4chan.org/sp/ (not verified) :: Mon, 01/18/2010 - 6:18pm

The percentages I use now were the percentages that did the best job of predicting each team's performance for the remainder of the season.

Aaron, how do you deal with the potential from overfitting your model to the (any-given-sunday-fuzziness of the) data?

Given that you're aiming for predictive rather than descriptive statistics, overfitting to empirical data (i.e. better describing the past) may actually hurt your model's ability to predict next week's results. Is there also an a priori reason for the choices you make?

by Salvi's Headband (not verified) :: Mon, 01/18/2010 - 4:42pm

My mind is short circuiting at the idea of the Jets in the AFC title game. WTF?!

by Rick A. (not verified) :: Mon, 01/18/2010 - 5:17pm

Defense and strong rush will always win in December/January no matter what other theories and practices get concocted. You've read too many articles ridiculing Ryan and the Jets...

by Eddo :: Mon, 01/18/2010 - 5:35pm

Odd that you say that when we just saw the Colts (1.7 yards per carry), Vikings (3.3), and Jets (3.1 if you throw out the long TD) struggle to run the ball effectively, yet still win.

Now, you're right that the defenses of all three teams played extremely well; however, how do these results perpetuate the myth that running the ball is vital to win in the playoffs?

by Rick A. (not verified) :: Tue, 01/19/2010 - 1:40am

I didn't say it was vital. However, if you have it in December/January it is certainly not a myth that you stand a much better chance, even in indoor games. Duh...Football has not changed in this regard and never will. And if you have it versus one of these air specialist teams you stand an even better chance. Baltimore certainly would've done alot better the other night had they taken better care of the ball, wouldn't they ? And I'm someone who was telling you guys Indy is the best team since week 6, while the quantification crowd was mystified over how were they doing it. (and still are) But I certainly knew that Indy had a serious threat on their hands with Baltimore coming in. Excuse me but it's really hard to answer this sort of thing. And the Jets pose a serious threat to Indy this weekend and I'm someone who told all of you a couple of weeks ago that they would create havoc in these playoffs and that I saw them making it to the Championship game. Do you, and did you, hear that Mr Hoptoad ? That was while this site was ridiculing the Jets and scratching their heads at how they got in. Which is still going on. What is your point Mr Eddo ? Are you an agent for the mythology known as DVOA and it's premises ?

by peterplaysbass (verified?) (not verified) :: Tue, 01/19/2010 - 12:30pm

I will never overestimate the value of a good running game coupled with a good rushing defense after watching last year's Vikings get embarassed in the playoffs.

by Rick A. (not verified) :: Tue, 01/19/2010 - 9:52pm

They had the worst QB in the NFL and they met another good defense...

by Eddo :: Tue, 01/19/2010 - 10:33pm

So maybe, just maybe, quarterback play trumps the running game?

by Rick A. (not verified) :: Wed, 01/20/2010 - 3:28am

I would never say, as a former player, that QB play TRUMPS the running game. It's the most TEAM sport there is. While QB is probably the single most important position, it's only one position. The entire pundit world in America has created some of the most commonly accepted perceptions among fans but which are so removed from reality as to be hard to even talk about. It is common to speak of QB's won-lost records and so on. Or the judging of people on whether or not they "ever won the big one" etc. Look Eddo, Dan Marino never won a Super Bowl, Trent Dilfer did. Enough said ? You want to know what everyone in the game always said about Dan ? It was a real simple "jesus, if they would ever get the guy a running game Miami would win every year"...

by cfn_ms :: Wed, 01/20/2010 - 1:40am

Didn't Peterson get hurt in that game?

by tuluse :: Tue, 01/19/2010 - 7:31am

They have the best defense in the league, and poor, but not terrible offense. Why does that short circuit your mind, but not San Diego potentially making it? You know San Diego of the excellent offense, and poor, but not terrible defense.

by Jeff Fogle :: Mon, 01/18/2010 - 5:03pm

Some ideas to put up for discussion. If any sort of consensus develops from posts, I can send an email as requested that can be marked for future reference. If they’re unanimously seen as dumb ideas, I won’t do that (lol). Good to have the power of the collective at work, particularly with this collective.

1…Credit DVOA that’s accumulated when teams are within one score of each other with a higher weight. Limit the impact of DVOA accumulated when teams are more than two scores apart (which lessens the errors that can develop when comparing a team that runs up the score to one who eats clock with a lead).

2…Break out of the rushing/passing dichotomy and consider passes of a certain short length to be “extended rushing plays” rather than passes. Meaning, there’s a focus on “plays that move the chains” versus “downfield plays” rather than run/pass. Defenses are judged at “defending the chains” and “defending against the big play” rather than some of the “stops the run, can’t stop the pass” illusions or vice versa. Look for a correlation between strengths in those areas and won-lost records in games that matter most. If you find one, recalibrate to showcase the stuff that matters most.

3…Consider that the goal of the season is to win a championship, rather than to win each and every individual game. Teams automatically get to the playoffs by winning their division, so divisional games are given heavier weight. Teams can also reach the playoffs via Wildcard, so conference games outside the division are given the next rung of importance (conference records count in Wildcard tie-breakers). Non-conference games get the least priority. Once a team has clinched a division, or maybe has a 3-game lead, place less weight on their games until a championship is once again in play (when the playoffs start up).

4…Consider lessening the statistical impact of games where one team has a bad body clock (early East Coast starts for West Coast teams), or where the visitor travelled more than xxx number of miles. Or, reward good efforts for travelling teams and penalize the hosts if they perform poorly…but don’t place much weight on the results if the traveler pulls a virtual no-show.

5…Consider lessening the impact of non-divisional games that occur after a string of at least two divisional games. Teams aren’t penalized for “out of gas” efforts in that string. Opponents aren’t rewarded for beating up a team that was out of gas.

6…Consider using a “percentage of full strength” multiplier based on the injury list, so that teams who abuse injury riddled opponents aren’t rewarded…and that injury riddled opponents are then considered in the proper light once they’re back at full strength.

7…When you go “weighted,” make that mean everyone at full strength rather than “what’s happened lately.” What’s happened lately can become greatly polluted if teams are coasting to divisional titles, or have a sequence of low priority games (non-conference, poor travel).

8…Express DVOA in a language that seems less like Klingon. Whether it’s points, yards, wins (estimated wins is helpful already), whatever. Football fans, readers, the media, already have instincts for those based on years of watching games. Recall any acronyms from the Bill James Abstracts? He used words…words that meant something (runs created, range factor, approximate value). DVOA doesn’t resonate. The words it stands for are awkward. Calling it “efficiency” is confusing because there are various indices calling themselves efficiency at this site…and other sports have other definitions for the same word. Define it in real language, then call it what it is. Acronyms are codes that hide meaning. You can keep the mechanics proprietary without having the output in code.

9…Assume reality is a moving target you’re never going to quite hit…and if you do, it’s just going to move again anyway. The “this is the way it is” tone that works its way into articles hurts credibility when that wasn’t the way it was. Do I need to start a list? Demoralizing as a fan of the site how often that's happened in just the past few weeks.

10…Monitor the legal betting markets. Multiple very advanced modelers have told me that the best estimate for a final score in an NFL game still comes from the widely available side and total lines. If you’re way out of step with the market, it’s time to change shoes.

by C (not verified) :: Mon, 01/18/2010 - 9:18pm

Jeff Fogle,

That sounds like a good list for a gambler to interpret the DVOA. It's hard to quantify all of those points though.

by Rick A. (not verified) :: Tue, 01/19/2010 - 1:59am

Emminently reasonable. The most sense I've encountered on these threads as to the way to create some sort of quantification model. You should make your own system and start your own website. Lots of dough has to be in this or they wouldn't be doing it...

by RickD :: Tue, 01/19/2010 - 1:59am

1. sounds reasonable
2. It might be reasonable to differentiate between long passes and short passes, but any short pass could become a long pass via YAC. That may be something to consider when comparing QBs, but I don't think it should reflect on the team as a whole.
3. No. There's no reason to give extra weight to divisional games. You're better off caring only about the quality level of the opposition. Unless you think a game between the Cardinals and the Rams is more important to those two teams than a game between the Patriots and the Colts.
4. No. The "East Coast" thing is a recent way for fans of West Coast teams to avoid the fact that, on the whole, their teams all suck. (Minus the Chargers, of course.) If their coaches cannot get their teams to figure out how to play at 1 p.m. EST, then they are not getting them ready to play NFL football, since the majority of games are at 1 p.m. EST.
5. So the Cardinals are out of gas after beating on the Rams and Seahawks?
6. I think that it would be interesting to try to make a quick adjustment on a team's DVOA based on injuries, but as a practical matter, it would be infeasible. As for the other hope, players just don't come back form injuries that quickly.
7. "Weighted" means "what's happened lately". That's just what it means.
8. Got any ideas for better terms?
9. I think this site does a good job of recognizing that reality is a moving picture. I've never seen anything that suggests, for example, that the Patriots of week 9 were the same team as the Patriots of week 17. Indeed, if you followed the rankings (or watched the game), it was clear that the opposite was the case.
10. Nah.

by C (not verified) :: Tue, 01/19/2010 - 10:24am

I can't believe some of the stuff the people post here. Even an average gambler could sift through the fluff.

The West Coast team playing at 1PM Eastern is one of the hardest things to do in sports. You do realize that when the kickoff happens it feels like it's 9AM local time for a West Coast team. That they players DO have to get to the stadium early... That they can get jeg lagged. I guess in the world of stats you don't think about these things... Think about what happens if you get to the stadium 4 hours early. Then you are waking up at 4AM SF, SEA, SD time to go to the stadium. Have you ever traveled to Europe or from East-West coast? How did you feel? Were you running sprints and doing intense physical work that required you to think on your feet? One team gets up at 4AM local time after jet lag, to enter a roaring hostile stadium 80,000 deep. Yeah, that's why West Coast teams traveling east for 1PM starts are horrible against the spread.

Dude, Divisional games matter. Think about why. These teams "know each other" very well. They were talking about how the Cowboys offense might read the Eagles defense, call an audible, the Eagles defense knows the audible, so then they call and audible after the Cowboys audible. Look at Antoinio Pierce playing a divisional team calling last second audibles and audibles after they call audibles... Look at the Packers drafting all DB's one year to combat the Vikings Randy Moss, Cris Carter, Jake Reed. They were trying to build to stop them. Look at the Houston Texans drafting big name defenders to stop the Colts... Yeah, the game might mean less to YOU, but the scouts are very FAMILIAR with their opponents and often build to win thier division. That's why you look at say an NFCEast team vs the AFC teams ( that they don't know as well) and can compare it to teams they do know well. If the Cowboys were very good vs the NFEC but did bad against out of division opponents it would suggest that their scouts did well against the teams that they know, but they did bad against teams they don't know as much about or vice versa. Not the most important thing in the world but there is a dynamic there. Something even an elementry gambler would know.

by Xao (not verified) :: Tue, 01/19/2010 - 4:20pm

I can't believe some of the stuff the people post here. Even an average fifth-grader could sift through the time-zone conversions.

The West Coast to East Coast time conversion is one of the easiest things to do in math. You do realize that when the kickoff happens it feels like it's 10AM biological time for a West Coast team. That they players DO have to get to the stadium early... That they can get jeg lagged. I guess in the world of self-proclaimed successful gamblers you don't think about these things... Think about what happens if you get to the stadium 4 hours early. Then you are arriving an hour too early!

Seriously dude, if you're going to write a screed about time zones, get a simple +3 conversion correct.

by C (not verified) :: Tue, 01/19/2010 - 6:05pm

So players don't have to wake up at 4-5AM local time to get ready for games? Is that what you are saying or were you just trying to be a wise guy? It's not like NFL players get to the stadium 45 minutes before the game. If the game starts at 10AM pacific, and you have to be there 3-4 hours early at 6AM or 7AM pacific, and you had to wake up at the hotel say 1 hour before that... You are waking up at 5AM give or take... probably before most of those guys normally get up and you could be tired/out of wack from the jet lag.

There is a dynamic there, but if all you do is look at games and stats you miss some of the real world aspects of the game. I've seen the self proclaimed genius people here miss all kinds of common sense aspects of the game.

I also think that some people hate on a guy trying to make a buck off of knowing more than the market with respect to gambling.

by Rick A. (not verified) :: Tue, 01/19/2010 - 9:41pm

Thank you C. Most of the fans could give a shit about the player's perspective. They deserve all the reaming the owners give to them...

by Jeff Fogle :: Tue, 01/19/2010 - 2:30pm

To Rick A: Afraid the dough is influenced by strengths at promotion, and is dispersed in an Extremistan format (Black Swan). FO earned their spot at the top of the heap with a mix of interesting informative content, a passion for what they believe in (even if reasonable minds can different about various issues), and an ability to "self-promote" in the best sense of that term (meaning they went out and earned attention).

To C: I think we agree on a lot of things (lol).

To Rick D, appreciate your point-by-point commentary. My responses:

2. Agree that YAC is an issue to deal with. I'm trying to find a way to reward "move the chains" teams that grind out success, and pull back on "makes extra big plays in games they were already dominating vs. lesser lights anyway." I think a lot of the teams who strike thread commenters as "better" than they're rated are grinders...and a lot of the teams who are worse make big plays in blowouts but don't know how to grind when they need to in big games.

3. It's definitely possibly to cherry-pick a few divisional rivalries that aren't exactly blood wars...and a few non-divisional rivalries that get the juices flowing. Over the scope of the full season, I think coaches are prioritizing divisional games at the expense of others...and that rating systems should reflect that. ARIZONA surely cares about St. Louis more than they care about Indy or NE, as their scores vs. those teams over the last two seasons suggest. Even if you account for talent differences, you get Arizona coasting past St. Louis four times 120-46 (3-1 vs. Vegas expectations), but losing to Indy and NE by a combined 78-17 score (missing expectations by a zillion both times). Arizona has turned out to be MUCH better vs. playoff caliber teams than those losses to NE and Indy would have suggested. Bottom line, if you count all the games the same, you miss the clear (to me at least) prioritization of divisional games. If a ratings system prioritizes divisional games, it will only miss the handful of non-divisional rivalries that get the juices flowing.

4. We'll have to disagree about that one. I think high travel games are definitely low on the list of priority for many coaches. They have to pick their spots for "asking for everything" during a 16-game season. Games on long trips would be logical nominees for backing off and living with a loss. Jacksonville came to mind as a team that didn't travel well this year. Went to look it up, and they're probably much more extreme than normal. In non-divisional road games:

Jax (+1) lost at Seattle 41-0
Jax (+6) won at NYJ 22-20
Jax (+3) lost at SF 20-3
Jax (+9) lost at New England 35-7
Jax (+1) lost at Cleveland 23-17

That's 1-4 straight up and ATS, with a combined 61-3 loss in the non-conference games on the other coast. To mediocre non-playoff teams. Don't want to suggest everybody does this. Jacksonville had decent divisional results in a tough division, but crappy results elsewhere.

5...Glad you picked the Cardinals. I think sometimes it's "out of gas," others it may just be "low priority" in terms of the goals of the team. Arizona has to win its division to make the playoffs.

This year they only had one sequence of a back-to-back within the division. After beating Seattle and St. Louis 52-33, they went to Tennessee and allowed 532 yards to the Titans(a whopping 8.1 yards-per-play). Vince Young passed for 387 yards. That was the game Leinart played, and almost won because a kickoff return TD kept them in the game. Not a tank job in the least, but a misleading read on the Arizona defense needless to say.

Last year we saw this sequence for Arizona, on the way to their surprising playoff run:

Won at St. Louis 34-13
Won vs. SF 29-24 (non cover ATS)
Won at Seattle 26-20

Lost vs. NYG 37-29
Lost at Philly 48-20 (short week)

Won vs. St. Louis 34-10 (the win that clinched the division)

Lost to Minnesota 35-14
Lost at New England 47-7

Won vs. Seattle 34-21

Arizona ranked mid-level in DVOA last year if I recall, and there seemed to be a consensus that they would be in trouble in the playoffs. Given what we know now, a case can be made that DVOA under-rates Arizona because of their tendency to blow off some games vs. quality, particularly if they have to travel. The Cards are 4-2 straight up in the playoffs (underdog 5 of the 6, and a 1-point favorite the other time). Their results amidst the sequence of divisional games would suggest that's very unlikely. Yet, we all saw that Arizona was a much different team in the postseason.

6. Agree that injury adjustments wouldn't be easy. Think an attempt is better than not attempting and just hoping everything kind of cancels out by default. I think some Philly fans have mentioned the injury differentials between the Eagles and Cowboys the past few weeks. That looks to have helped create a false read on Dallas.

7. "Weighted" can mean whatever they/we want it to mean.

8. No, I'm still learning Klingon. I think they can express what they're trying say more effectively, and without acronyms. Even something crazy simple like "Football Outsider Team Rankings" is better than "DVOA Efficiency" in terms of people knowing what the words mean. Paraphrasing an old Bill James book, DVOA is a bone you don't want to eat after roasting a pig. Serve people meat (with a little sizzle still left) rather than the bones.

9. Agree if you're talking about week to week within a season. FO seems very open to the concept of teams moving around, getting bettor or worse, etc.. I think there's a good chance though that pro football is evolving away from what they believe they captured with their methodology (they tend to reference years well in the past, particularly 2005, in their work). They seem less open to that. They called this year "kooky" when it wouldn't strike many as kooky at all. Comparing FO to other ranking systems this year suggested the guys were on Neptune. You could say, "Well, these seem out of alignment with what I'm seeing and what other places are seeing...but if their top teams do well in the playoffs...maybe they were right all along." The DVOA sweehearts have all been eliminated, largely crushed by teams DVOA was saying weren't as great as people thought. The "demonstrably false" things Rex Ryan was supposedly saying don't seem so demonstrably false any more. The "new rules" of playoff success that Bill Simmons outlined in a linked article (which readers suggested were in line with FO principals)somehow dropped stopping the run down to 9th most important when putting a boulder in the middle of the line is still top priority for playoff teams. Neptune.

10. Hope you'll reconsider "Nah". Or, I hope they will. The market is a nice summation of football analysis...and it speaks in a language everyone understands.

by C (not verified) :: Tue, 01/19/2010 - 6:10pm

Jacksonville was a bad road team and Garrard didn't play well on the road. The QB is a pretty important position don't you think? Could it have been the loud noise that effected him? Poor in house scouting on those non divisional AFC opponents?

Not only was Rex Ryan "wrong" he was a fool we should laugh at... Because he's just (an old school baffon)?

I'd like to talk to you more, what's your email?

by Jeff Fogle :: Tue, 01/19/2010 - 6:34pm

Jacksonville didn't set the world on fire much of anywhere (2-6 ATS at home, compared to 3-5 ATS on the road)...but losing margins were pretty ugly in the listing mentioned. Average home game was a 1-point win after regulation (favored in five of eight games and never more than +3.5 as a home dog), average road result about -13 (a dog of more than 7 only once). To me, their 2009 results line up more with "division-top priority, non-conference or long trip out of division low-priority). A lot of ways to interpret something like that though. Went to check...4-2 ATS within the division (with a half point loss to Indy), 1-9 ATS outside the division. Doesn't reflect well on non-divisional scouting or effort outside the division I'd say. Email address is that name at gmail. Trying not to trigger spam gremlins by typing it out in a line. No space between the first and last name...jam it together with 3 f's in a row...

by Rick A. (not verified) :: Tue, 01/19/2010 - 9:46pm

Jeff, they got chummy with some of the League's marketing department, aka ESPN...

by jmaron :: Mon, 01/18/2010 - 5:37pm

I think one of the big problems in working to perfect the DVOA model is teams don't all play one another. If every team played each other once than it would be much easier to come up with some idea of which are the better teams.

But the way the schedule works where the one division plays 37.5% of it's games in it's own division and 50% against two of 7 other divisions than you can get a situation where the level of play is simply lower in those games. There aren't enough games (2 of 16) against the other 5 divisions to even out the ratings.

I think it's obvious that this happened this year with the NFC East and the AFC West. I think the scores in the games those teams played outside of games involving two teams from those divisions suggests strongly that was the case.

by huston720 :: Mon, 01/18/2010 - 5:58pm

True that every team doesn't play each other, however every team is connected by linking schedules, since we can evaluate how the opponents on a teams schedule did in games against non-common opponents and so on. So even though every team doesn't play every other team, the relative schedule strengths compensate for this. However it is important to remember that DVOA predicts how a team will play versus an average team, but it doesn't take into account how teams matchup, which is why DVOA is a tool, and not absolute. For example the Jets were a really bad matchup for the Chargers (and potentially the Colts) which is why the game was so close. If the Jets had the same weighted DVOA but were stronger on D against the run than the pass, and stronger passing than running on offense (essentially their polar opposite) we might have expected a drastically different outcome in their playoff game.

by Rich Conley (not verified) :: Mon, 01/18/2010 - 5:54pm

"1…Credit DVOA that’s accumulated when teams are within one score of each other with a higher weight. Limit the impact of DVOA accumulated when teams are more than two scores apart (which lessens the errors that can develop when comparing a team that runs up the score to one who eats clock with a lead)"

Why would you do that? Running up the score makes you more likely to win than running out the clock. You're asking to reward teams for playing Marty-Ball despite the fact that playing Marty-Ball makes you less likely to win.

by huston720 :: Mon, 01/18/2010 - 6:49pm

I don't know that running up the score with a two score lead is really that much more likely to win the game than killing the clock when up two scores. Both have their pros and cons, but neither seems like an obviously better strategy when there is a two score lead. As for Marty-Ball i think that was more a tendency to try and kill the clock too early, i.e. when it is still a one score game, which can definitely hurt you if it takes away from what got you that one score lead.

Also the description you quoted is a simplified version of the actual model as I understand it, since DVOA also takes into account time remaining on the clock. Thus a two score lead with 14 minutes to go is treated differently than a two score lead with 2 minutes to go. Hope that answers your question.

by asm335@gmail.com :: Mon, 01/18/2010 - 6:16pm

I don't see how there is much to complain about with DVOA this week considering that the system spit out all 4 winners in the premium content section.

by cfn_ms :: Mon, 01/18/2010 - 7:05pm

That's surprising, given that Dallas was #2 DVOA and Minnesota #14. With Chargers-Jets I could see the specific matchups swinging it to an upset, but I'm surprised that they'd pick the Vikings with that much of a gap. Interesting.

by Jeff Fogle :: Mon, 01/18/2010 - 7:42pm

May be different methodologies for rankings and picks. I would think there'd be some concern about:

#1 in weighted losing 20-3 to #6 in weighted
#2 in weighted losing 34-3 to #14 in weighted
#3 in weighted losing 17-14 at home to #7 in weighted
#4-5 in weighted not reaching the final eight
#8 in weighted losing 45-14 to #12 in weighted

That's 1-2-3 losing by a combined 61-20, and the four "weighted" superior teams losing by a combined 105-34 score with an 0-4 record.

If Premium went 4-0 ATS, that's good news that the selection indicators know what to discard from weighted.

If you're going unweighted from the final regular season numbers
#1 lost 20-3 to #8
#2-3-4 didn't advance to the second round, losing by a combined 128-73
#5 lost 34-3 to #7
#6 beat #13 45-14 (hurray!)
#10 beat #11 17-14 (hurray!)

So, did the selection methodology weigh full season over weighted? Thought full season was polluted because of the preseason issues factored into the selection process. Congrats if Premium went 4-0. Why the emphasis on "weighted" in the articles if they don't have that emphasis within the process? Isn't standard home field worth 2.5 to 3 points?

PS: why this comment in the Arizona/NO preview:
"Their (New Orleans) home-field advantage is no bigger than the usual, and that advantage is really the only reason they should be favored in an otherwise even matchup."

by God (not verified) :: Mon, 01/18/2010 - 6:17pm

Jets neber got crtedit for very fast, punishing defense. able to blitz creatively, yet provide solid coverage in thre secondary. Jets get dismissed even though they have the best offensive line - zone blocking in league. Defense wears down the opposition, offense slowly grinds out yardage and kills clock. Jets very dangerous, and so what if DVOA had them rated below a conference champion.
I love this site ~

by Jackson Jackson (not verified) :: Mon, 01/18/2010 - 6:21pm

Does God drink Sierra Nevada?

by Bobman :: Mon, 01/18/2010 - 9:01pm

I wouldn't be too surprised. Pretty good shit. And he's a man of wealth and taste.

oh, wait a minute. oh damn.....

by Rick A. (not verified) :: Tue, 01/19/2010 - 2:02am

Good God you are correct. They, indeed, have the best O Line. And that is "football reality" cybergeeks...

by dk240t :: Mon, 01/18/2010 - 6:36pm

The New York Jets are clearly ranked too low because the football gods are preparing a karmic revenge whereby the Jets get in the playoffs only because the Colts inexplicably decide not to go for a perfect season, then go on to knock the Colts from the playoffs. My crystal ball is way better than this. Cowboys sux0rs.

by AL DAVIS (not verified) :: Mon, 01/18/2010 - 9:15pm


by t.d. :: Mon, 01/18/2010 - 9:37pm

Didn't the Saints rest four or five defensive starters for most of the second half of the season? Shouldn't their perceived weaknesses be seen through that light? I thought the Saints-Vikings matchup would be the game of the year in week ten, and I still think so now (two explosive offenses, a great turnover defense verses the best front four in the league)

by mm (not verified) :: Mon, 01/18/2010 - 10:39pm

There were many players on defense who missed games due to injury in the second half, but they didn't rest any un-injured starters until the last game.

by Marnie (not verified) :: Mon, 01/18/2010 - 9:56pm

I'm genuinely curious (and also new to this site, so please take my DVOA ignorance into account!) Does anything about the stats/numbers explain why Baltimore didn't (and rarely does) match up well with my beloved Colts despite boasting a higher DVOA than Indy does? Are there specific categories I should be examining and comparing? Also, how does one account for the fact that the Colts were favored---even by you fine folks, I believe---to win against Baltimore despite having the inferior DVOA?

Again, I apologize in advance for the stupidity of these questions! The not-so-hidden geek in me is very intrigued by these stats, and would love to know how to apply some of the information you provide to help 'predict' the Jets-Colts game as well. Thanks!

by Paul R :: Mon, 01/18/2010 - 11:15pm

It's okay, I've been visiting this site for a long time and I'm still trying to figure out how to pronounce it.

by Nilblog (not verified) :: Tue, 01/19/2010 - 12:11am

DEE-VOAH - Are We Not Defense-Adjusted?

by Anonymous@ (not verified) :: Tue, 01/19/2010 - 3:01am


by RickD :: Tue, 01/19/2010 - 2:01am


by Lou :: Tue, 01/19/2010 - 4:41am

thats how i've always pronounced it.

by Dales :: Tue, 01/19/2010 - 8:21am


Generally pronounced with the same vocal effect (and horn) as in the cough drop commercial.


I think that's Barnwell on the horn.

by Packerpalooza (not verified) :: Tue, 01/19/2010 - 8:51am

It's rather foolish to look at one season's results to make an assessment of how a model works. And I would urge the FO crew to show restraint in making any serious adjustments based on the same premise. Patience, however tedious, is inevitably rewarded.

Well, except for Minnesota fans. Then it's just more pain and misery.

Or it is to be hoped..............

by cfn_ms :: Wed, 01/20/2010 - 1:52am

I sort of agree, sort of disagree. I think that when something looks weird in a season, it is a solid reason to go back and re-evaluate in the context of prior seasons. If the one season simply looks like an outlier, oh well, but if prior seasons are neutral to mildly supportive of some sort of shift, it probably makes sense to at least tweak the system a bit.

IMO a good thing for them to look at over the offseason is their penalty ratings, in the context of "why the hell was Baltimore #1". Maybe there's nothing to tweak, and it's simply an outlier, but I strongly suspect that Baltimore's extreme tendency to commit penalties (especially dumb ones) ought to drop their ranking somehow.

If I had to guess (and I freely admit I don't really understand their formulas), I'd say that the way their formula works is that if a team happens to have an unusual amount of penalties in a game or two, it's not predictive of future penalties, and therefore doesn't substantially affect the rating. However, when a team makes a never-ending habit of this sort of thing, I'd guess that it is predictive of future penalties. Whether there's a clear and easy way to tweak the system to make that happen, i don't know, but if I were them it would be at or near the top of the list of things to look at in the offseason.

by DeltaWhiskey :: Wed, 01/20/2010 - 7:04am

I've said it before, I'll say it again. Overall, DVOA worked the same this year as it has since 1994. The correlation of DVOA with wins for the 2009 season is a robust r = 0.860. The correlation from 1994 to 2008 is r = 0.854. Whatever DVOA measures, it is still measuring it the same.

by cfn_ms :: Wed, 01/20/2010 - 7:11pm

Is that predictive or retro-dictive? (i.e. based on DVOA at the time each week or on total year-long data). And how would that compare with simple metrics, such as the correlation between the team with the better record and who wins each game?

by DeltaWhiskey :: Thu, 01/21/2010 - 6:54am

Earlier in the year (WK 8) I looked at DVOA's utility as a predictor. I found that WK8 DVOA historically had a fairly robust correlation with end of season wins (r =0.75); however, the correlation was slightly less than the correlation of WK8 win percentage with end of season win percentage (r = 0.81). Using DVOA and win percentage to project teams' final records resulted in an accuracy of 55.4% for both to predict teams within +/- 1 game of there final season record. Expanding to within +/- 2 games yielded DVOA having a 96.9% accuracy rate, compared to a 93% accuracy rate for simple win percentage.

I am inclined to agree with earlier posts that have described DVOA as tool that allows those who do not have time to analyze every team in depth the opportunity to quickly assess how well teams are playing and in what broad areas. The non-premium DVOA data is of limited use to make confident predictions about individual games - the premium data base may be of more utility. In general, then, I believe (non-premium) DVOA is primarily a descriptor, marginally useful for longer term predictions and probably of very little utility for game to game predictions.

Unfortunately, DVOA is a awkward metric. Regression equations allow for conversion to easier to understand metrics. For example for the NYJ WK17 states,

DVOA of 16.3% suggests they are playing like a team that can be expected to win 10 games
OFF DVOA of -9.7% suggest their offense can be expected to produce on average about 19 points.
DEF DVOA of -23.4% suggest their defense can be expecte or yield on average 15 points.

To me, these descriptions make more sense.

by Jeff Fogle :: Fri, 01/22/2010 - 8:54pm

Agree with your notes at the end there DW (and, from memory, I think everything else you've ever said, lol). Talking about the Jets as about a 19-15 scoring team that would be expected to win 10 games is a lot more digestible and understandable. Hope the FO guys will consider that for next season.

Wanted to ask the field if anyone knows about the discrepancy between reports of FO Premium's record last week. Post 83 in this section by asm335@gmail.com (written back on January 18th) said:

"I don't see how there is much to complain about with DVOA this week considering that the system spit out all 4 winners in the premium content section."

Over in the Betting Discussion Forum, thePop is reporting:

"Divisional Playoff Round (1-3):

GREEN: NYJ +8 at SD - Won

YELLOW: ARI +7 at NO - Lost

RED: BAL +7 at IND - Lost
RED: DAL +3 at MIN - Lost"

So, one person is saying 4-0, another is saying 1-3. Could somebody from FO clarify that for us? And, if 1-3 is correct, explain why you let posters in this thread think it was 4-0 the last four days? You guys are now affiliated with ESPN, the NY Times, and other places as well. Important to keep the facts front and center. Either asm335 is in error, or thePoP is...unless you have separate authors who are in conflict or something. Thanks in advance for any clarification that can be offered.

by jhq (not verified) :: Fri, 01/22/2010 - 10:23pm

The core problem with DVOA is not mathematical: it is psychological.
As a result, you have gifted athletes on great teams, but not of necessity mentally strong. this conundrum cannot be solved by math alone, that is the inherent weakness of FO statistics.
The human mind cannot be wrapped in Pythagorean numerology, but is a fluid and ever changing dynamic,and, as much as Schatts and co. would like to pin the donkey, it will never occur, based on the nature of the human soul responding to varying stimuli.
Who can determine if a player plays to his capacity ? Or better, superhuman capacity, altered by love, hatred, compassion, or death ?
How can you statistically determine a man's response in these cases ?
That is the cause of the discrepency between DVOA and "boots on the ground" playoff teams, not the fault of stats per se, but rather the other elements outside the boundary of math.

by The Peepshow (not verified) :: Wed, 09/08/2010 - 2:50pm


is clearly ranked is way better than this.

Did I get the formula right?

The Peepshow

by The Peepshow (not verified) :: Wed, 09/08/2010 - 2:51pm


Seattle Seahawks is clearly ranked too low because I ate Special K for breakfast this morning while wearing my girlfriends underwear. Ranking the teams based on the comfort of their NFL approved women’s underwear is way better than this. Cleeraly you missunderestimate the impact the compfee underwears has on the impact of a team’s performance, and the Seattle Seahawks obviously have the most comfortable lady breefs on the market. 9ers SUX! PWND!

Did I get the formula right?

The Peepshow