Writers of Pro Football Prospectus 2008

Most Recent FO Features

HackenbergChr15.jpg

» SDA: From Death Valley to Happy Valley

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

15 Dec 2009

Week 14 DVOA Ratings

by Aaron Schatz

Can you name the best teams in the NFL this year? It doesn't seem that hard, considering that Indianapolis and New Orleans have each started 13-0. The Football Outsiders DVOA ratings don't disagree here -- after all the advanced play-by-play breakdown, New Orleans ends up currently number one in overall season-long DVOA, while Indianapolis currently holds the top spot in weighted DVOA that discounts the results of early-season games. While DVOA agrees on the identity of the best teams, however, it certainly doesn't agree with the official standings when it comes to the distance between those teams and the rest of the league.

Look at win-loss records, and it is pretty clear that the 2009 season has two dominant teams destined for a historic Super Bowl clash, although two other very strong teams (Minnesota and San Diego) have a chance to upset the apple cart. Most of the other teams are scuffling around, trying to avoid problems like pouty wide receivers, seven turnovers a game, or the little voice inside the head coach's brain that says "You can win with a 100-0 pass-run ratio! Try it!"

Look at DVOA ratings, however, and the Colts and Saints aren't even close to running away with this thing. The 2009 season looks like a mish-mash of very good teams, primarily separated by their consistency (or lack thereof), scrambled up in our ratings and headed for a postseason with no clear favorite.

There's no way around it. The contrast between the standings and the DVOA ratings has gotten really wacko. Last week's DVOA analysis looked at how Indianapolis and New Orleans are not as dominant as past teams that went deep into the season undefeated, but the other side of the story is that we have a lot of other pretty good teams that our ratings say are roughly equivalent even though they're sitting around with records that range from 7-6 to 11-2. (Of course, we were one Steven Hauschka field goal away from those specific teams actually being 8-5 and 10-3.)

  • Six teams currently have DVOA over 25%. In 2004, five teams had DVOA over 25% at this point; in no other season were there more than four teams over 25% after Week 14.
  • Eight teams currently have Weighted DVOA over 20%. There are only two other seasons where at least six teams had Weighted DVOA over 20% at this point: 2004 (seven) and 2002 (six). Furthermore, if we drew the line at 19% instead of 20%, we would have ten teams instead of eight, and that list of ten teams doesn't even include San Diego (we'll get to them in a bit).
  • Despite all those teams over 20%, not a single team has Weighted DVOA over 30%. The only other year where no team had Weighted DVOA over 30% after Week 14 was 2003 (the year Kansas City dominated the first half of the season, then collapsed).

You can see how strange this is compared to the past 15 seasons. When it comes to having lots of strong teams, the most similar season was clearly 2004 -- but in 2004, you didn't have a big close pack of good teams. Pittsburgh and New England finished that season with two of the five highest DVOA totals of all-time, and the Colts were up in the same range until they rested their starters over the final couple weeks.

Just in case things didn't look strange enough, there's a pretty obvious team missing from our pack of good DVOA teams: the 10-3 San Diego Chargers. The Chargers beat Dallas this week for their eighth straight win and actually dropped a spot in the DVOA rankings, down to 14th. Much of the NFL commentariat considers San Diego to be "the team nobody wants to play" -- for starters, we've seen in recent years that they are a terrible matchup for Indianapolis -- so San Diego's low DVOA may be even more surprising than Cincinnati's low DVOA. And, like Cincinnati's low rating, it's hard to find reasons in the usual places. San Diego's fumble recovery rate is 50 percent. The schedule has been easier than average, but not extreme. They get a little value from "hidden special teams," but not that much. It's hard to figure out why so many of their wins have ratings so close to zero. Part of the problem is that the entire team is being dragged along by Philip Rivers and his running buddies, Antonio Gates and Vincent Jackson. We see how great those guys are, and maybe we forget just how ordinary the rest of the team has been this year. The Chargers are below-average on defense and special teams, and their running game is horrid.. LaDainian Tomlinson's bust in Canton is already sculpted and sitting in a closet waiting for use, but at this point the guy is so toast that actual toast is considering a lawsuit for libel. As Vince Verhei noted in today's Any Given Sunday on ESPN.com, the Chargers are on pace to break the record for biggest difference between pass offense DVOA and run offense DVOA, currently held by the 2003 Tennessee Titans.

We also know that San Diego's rating is being dragged down by their 2-3 start. This table shows San Diego's DVOA ratings for the first five games, and then for their eight-game winning streak:

  OFF DVOA Rank DEF DVOA Rank ST DVOA Rank TOTAL DVOA Rank
Weeks 1-6 10.4% 16 21.5% 29 1.6% 11 -9.5% 20
Weeks 7-14 28.2% 2 3.5% 18 -1.6% 23 23.1% 9

Although Weighted DVOA drops the strength of games more than eight weeks old, most of those games still have some value in the formula, which is why San Diego is still just 11th in Weighted DVOA. Assuming the Chargers continue to play winning football, their Weighted DVOA rating will rise over the next couple weeks until it is a lot closer to 23.1% you see above. Still, even if we only consider the Chargers of the past eight games, we don't get a great team that should scare everyone come playoff time. We get another pretty good team in the year where there are lots of pretty good teams, about as good as Minnesota, Baltimore, or that Dallas team they just beat by a field goal.

Obviously, DVOA is not the end all and be all of how to judge teams. There's something to be said for home-field advantage, as well as the value of consistency. In many cases, the teams that are lower in DVOA than you might expect are also very consistent. San Diego is second in VARIANCE, and Indianapolis is fourth. New England, Philadelphia, and Arizona are all in the bottom ten. On the other hand, Cincinnati -- a team which is much lower in DVOA than its win-loss record would suggest -- is 26th in VARIANCE. New Orleans is actually middle-of-the-pack after a fairly inconsistent November. And get ready for a shock -- the most consistent team in the league this year by game-to-game DVOA is actually Dallas!

Compounding this weirdness is our recent history of playoff upsets, starting with Pittsburgh's wild card run to the title in 2005. Let me show you a table that will be running in an upcoming ESPN.com Insider column. This table shows you how often the team with the better regular-season record wins in the playoffs, splitting things up by four-year increments. (Games where teams had the same record are not listed.) Yes, the sample size is small, and this doesn't account for home-field advantage in the wild card round sometimes going to a team with fewer wins, but it's hard to not be stunned by the contrast between the '90s and the last few years:

Years Wild Card Divisional Conf Champ Super Bowl Total
2005-2008 7-5 6-8 4-3 1-3 18-19
2001-2004 6-6 13-3 3-3 3-1 25-13
1997-2000 11-2 13-2 4-2 0-1 28-7
1993-1996 10-3 11-4 7-1 3-0 31-8

Between the recent history of playoff upsets and the closely packed DVOA ratings, there's plenty of evidence to suggest that this year's postseason will be another surprising one. Right now, our playoff odds report still has the odds of an undefeated New Orleans vs. undefeated Indianapolis Super Bowl at less than two percent.

* * * * *

And now, an important housekeeping note... We're finally set with something we've been planning on since we introduced our new player pages a few months ago: Sponsorships! You can now show your support for both Football Outsiders and your favorite players -- as well as advertise your football-related product -- by purchasing a one-year sponsorship on any FO player page. Most players are five dollars each, even many good players. Better-known players cost more, with the top players going for $200 per year. If you want a specific player page, just search for the player with our search box on the top right of the site; player pages will always come up first on the list. We're hoping to have a lot of fun little lists and articles to draw attention to the player pages in the offseason, and we'll be adding more material to them as well. 2009 data will go up when the season ends, and we're close to having 1993 data for skill players and 1997 individual defense. We're also exploring the addition of some of Bill Connelly's advanced college player metrics, so you can get Chris Johnson's POE at East Carolina to go with his DVOA ratings in the NFL.

* * * * *

These are the Football Outsiders team efficiency ratings through 14 weeks of 2009, measured by our proprietary Defense-adjusted Value Over Average (DVOA) system that breaks down every single play and compares a team's performance to the league average based on situation in order to determine value over average. (Explained further here.)

OFFENSE and DEFENSE DVOA are adjusted to consider all fumbles, kept or lost, as equal value. SPECIAL TEAMS DVOA is adjusted for type of stadium (warm, cold, dome, Denver) and week of season. WEIGHTED DVOA is adjusted so that earlier games in the season become gradually less important. It better reflects how well the team is playing right now. 

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

To save people some time, please use the following format for all complaints: <team> is clearly ranked <too high/too low> because <reason unrelated to DVOA>. <subjective ranking system> is way better than this. <unrelated team-supporting or -denigrating comment, preferably with poor spelling and/or chat-acceptable spelling>


TEAM TOTAL
DVOA
LAST
WEEK
WEIGHTED
DVOA
RANK W-L OFFENSE
DVOA
OFF.
RANK
DEFENSE
DVOA
DEF.
RANK
S.T.
DVOA
S.T.
RANK
1 NO 31.3% 1 26.8% 5 13-0 31.1% 1 -1.8% 11 -1.6% 26
2 PHI 31.3% 2 28.6% 4 9-4 14.0% 12 -10.5% 5 6.7% 3
3 IND 29.2% 4 29.2% 1 13-0 24.4% 3 -4.7% 7 0.1% 19
4 NE 28.2% 5 28.7% 3 8-5 29.6% 2 3.5% 17 2.0% 12
5 BAL 27.7% 8 26.2% 6 7-6 17.8% 7 -7.9% 6 2.1% 10
6 GB 27.5% 3 29.0% 2 9-4 19.0% 6 -16.4% 2 -8.0% 32
7 MIN 22.8% 6 23.2% 7 11-2 15.8% 9 -0.2% 14 6.8% 2
8 ARI 18.8% 7 19.7% 9 8-5 13.7% 13 -1.4% 13 3.7% 5
9 DAL 17.5% 9 20.5% 8 8-5 22.9% 4 6.0% 21 0.6% 16
10 DEN 16.5% 10 14.2% 12 8-5 3.8% 19 -13.4% 3 -0.7% 24
11 HOU 13.9% 14 19.0% 10 6-7 16.7% 8 5.8% 20 3.0% 7
12 PIT 11.8% 11 12.6% 13 6-7 14.5% 10 -1.9% 10 -4.6% 30
13 NYG 11.3% 12 7.3% 16 7-6 14.3% 11 0.6% 16 -2.5% 28
14 SD 10.2% 13 16.6% 11 10-3 21.4% 5 10.8% 25 -0.3% 21
15 NYJ 8.7% 17 5.4% 17 7-6 -11.0% 23 -17.3% 1 2.3% 9
16 MIA 6.2% 16 7.7% 15 7-6 8.1% 17 4.9% 18 2.9% 8
TEAM TOTAL
DVOA
LAST
WEEK
WEIGHTED
DVOA
RANK W-L OFFENSE
DVOA
OFF.
RANK
DEFENSE
DVOA
DEF.
RANK
S.T.
DVOA
S.T.
RANK
17 CIN 5.3% 15 4.4% 18 9-4 6.4% 18 -0.1% 15 -1.3% 25
18 SF 4.9% 18 8.4% 14 6-7 -7.0% 22 -11.8% 4 0.1% 20
19 JAC 0.2% 19 -1.1% 22 7-6 9.0% 15 8.4% 22 -0.3% 22
20 WAS -0.5% 20 3.3% 20 4-9 -1.8% 20 -1.7% 12 -0.4% 23
21 TEN -2.1% 23 3.8% 19 6-7 10.0% 14 10.5% 24 -1.7% 27
22 ATL -3.5% 21 -8.5% 23 6-7 8.3% 16 12.0% 26 0.2% 18
23 CAR -7.0% 22 1.4% 21 5-8 -6.7% 21 -3.3% 8 -3.6% 29
24 BUF -13.1% 25 -14.3% 24 5-8 -18.4% 25 -2.2% 9 3.1% 6
25 CHI -20.8% 26 -23.7% 25 5-8 -19.3% 26 5.5% 19 4.0% 4
26 SEA -23.6% 24 -28.9% 26 5-8 -13.6% 24 12.0% 27 2.1% 11
27 CLE -31.3% 31 -31.6% 27 2-11 -19.7% 27 18.9% 30 7.3% 1
28 KC -32.8% 29 -36.1% 29 3-10 -25.0% 30 9.7% 23 1.9% 13
29 TB -33.3% 27 -34.7% 28 1-12 -20.1% 28 14.0% 29 0.8% 15
30 OAK -36.9% 30 -36.7% 30 4-9 -25.4% 31 12.5% 28 1.1% 14
31 STL -40.7% 28 -39.2% 31 1-12 -21.0% 29 19.9% 31 0.2% 17
32 DET -52.0% 32 -51.6% 32 2-11 -25.8% 32 21.2% 32 -5.1% 31

  • ESTIMATED WINS uses a statistic known as "Forest Index" that emphasizes consistency as well as DVOA in the most important specific situations: red zone defense, first quarter offense, and performance in the second half when the score is close. It then projects a number of wins adjusted to a league-average schedule and a league-average rate of recovering fumbles. Teams that have had their bye week are projected as if they had played one game per week.
  • PAST SCHEDULE lists average DVOA of opponents played this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
  • FUTURE SCHEDULE lists average DVOA of opponents still left to play this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
  • VARIANCE measures the statistical variance of the team's weekly DVOA performance. Teams are ranked from most consistent (#1, lowest variance) to least consistent (#32, highest variance).




TEAM TOTAL
DVOA
W-L NON-ADJ
TOT VOA
ESTIM.
WINS
RANK PAST
SCHED
RANK FUTURE
SCHED
RANK VAR. RANK
1 NO 31.3% 13-0 38.2% 10.1 2 -4.9% 29 -11.4% 26 13.1% 14
2 PHI 31.3% 9-4 36.5% 9.2 4 -3.8% 27 19.5% 6 16.6% 24
3 IND 29.2% 13-0 30.7% 10.2 1 4.4% 9 -2.1% 19 8.4% 4
4 NE 28.2% 8-5 23.5% 8.9 7 5.4% 5 0.5% 17 21.9% 30
5 BAL 27.7% 7-6 27.3% 9.1 5 0.7% 17 -22.9% 29 13.2% 16
6 GB 27.5% 9-4 37.2% 8.7 8 -10.7% 32 3.5% 13 14.7% 19
7 MIN 22.8% 11-2 29.8% 9.3 3 -6.9% 30 -8.3% 25 8.5% 5
8 ARI 18.8% 8-5 18.8% 8.9 6 -2.2% 25 -32.6% 31 22.5% 31
9 DAL 17.5% 8-5 19.5% 8.4 9 -2.1% 24 31.0% 2 5.8% 1
10 DEN 16.5% 8-5 15.2% 8.1 11 3.6% 12 -19.2% 28 19.0% 27
11 HOU 13.9% 6-7 14.0% 8.2 10 1.3% 14 -3.2% 20 8.8% 6
12 PIT 11.8% 6-7 21.3% 7.8 13 -8.5% 31 30.7% 3 12.4% 13
13 NYG 11.3% 7-6 3.9% 7.1 16 4.8% 8 7.6% 11 14.2% 17
14 SD 10.2% 10-3 14.3% 7.9 12 -2.3% 26 1.3% 16 6.9% 2
15 NYJ 8.7% 7-6 15.2% 6.9 17 0.6% 18 15.5% 8 15.8% 22
16 MIA 6.2% 7-6 5.0% 7.1 15 5.3% 6 11.8% 10 10.8% 10
TEAM TOTAL
DVOA
W-L NON-ADJ
TOT VOA
ESTIM.
WINS
RANK PAST
SCHED
RANK FUTURE
SCHED
RANK VAR. RANK
17 CIN 5.3% 9-4 8.6% 7.2 14 -0.9% 21 -6.9% 24 18.9% 26
18 SF 4.9% 6-7 5.5% 6.7 19 1.2% 15 -30.8% 30 10.8% 9
19 JAC 0.2% 7-6 -0.8% 6.7 18 -1.3% 22 13.0% 9 20.7% 29
20 WAS -0.5% 4-9 2.8% 6.6 20 -4.8% 28 19.5% 5 12.2% 12
21 TEN -2.1% 6-7 -12.4% 6.4 21 7.5% 2 -3.6% 21 38.3% 32
22 ATL -3.5% 6-7 -6.8% 6.4 22 6.7% 3 -18.8% 27 13.2% 15
23 CAR -7.0% 5-8 -12.6% 5.9 23 3.9% 11 32.7% 1 15.2% 21
24 BUF -13.1% 5-8 -11.5% 5.5 24 -0.2% 20 26.9% 4 14.2% 18
25 CHI -20.8% 5-8 -19.7% 4.6 25 -0.1% 19 -0.8% 18 10.1% 8
26 SEA -23.6% 5-8 -20.7% 3.8 26 -1.7% 23 -3.9% 22 20.7% 28
27 CLE -31.3% 2-11 -38.8% 3.4 27 5.8% 4 -34.7% 32 16.4% 23
28 KC -32.8% 3-10 -30.8% 3.4 28 3.5% 13 -4.8% 23 7.7% 3
29 TB -33.3% 1-12 -36.6% 3.1 30 9.3% 1 2.1% 14 15.2% 20
30 OAK -36.9% 4-9 -40.0% 3.3 29 5.0% 7 6.4% 12 17.9% 25
31 STL -40.7% 1-12 -40.7% 2.7 31 0.9% 16 18.8% 7 11.3% 11
32 DET -52.0% 2-11 -53.9% 2.0 32 4.3% 10 1.5% 15 9.4% 7

Posted by: Aaron Schatz on 15 Dec 2009

320 comments, Last at 20 Dec 2009, 11:31am by R O

Comments

262
by Rick A. (not verified) :: Thu, 12/17/2009 - 12:36am

Let me assure you that Dick Vermeil goes on about that to this day. He comes up with his own "handicap" primarily using this. This is what he based his expectations on for his teams during his coaching days in St Louis. For example, it was his expectation heading into the SB that the Rams should win vs Tennessee on the neutral field by 14. That proved to be overly optimistic, obviously, but that is how he approached everything heading into the game. By and large, his overall "handicapping", over time, based on that stat,proved mostly accurate...

275
by randplaty (not verified) :: Thu, 12/17/2009 - 2:20pm

wow that's interesting. If point differential is just as good or better at predicting who is going to win the Super Bowl and DVOA is better for assessing the strengths and weaknesses of team components, why does Aaron and FO rely on DVOA to do overall rankings? Is it just an excercise for fun? How can you conclude that teams like SD are "only" as good as Minnesota, Balt and Dallas. Isn't that conclusion an assessment of overall team strength which is not the strength of DVOA?

277
by jmaron :: Thu, 12/17/2009 - 3:02pm

according to FO - between 2000-2005 DVOA had a better correlation with year+1 winning than pt differential. I think that is significant enough to value the system.

If you look at it on a yearly basis - DVOA very closely mirrors pt differential. The rest of all the arguing over which team is better using statistics like DVOA is just pointless fun at this point. DVOA simply doesn't tell us much about who is going to win that things like pt differential don't already tell us.

That doesn't mean they won't improve DVOA but the only stats I see are from 2000-2005, which I suspect means that they haven't significantly improved their ability to predict future performance or they would be bragging about it.

I'm not slamming them here - there breakdowns of teams and how the individual units are contributing to wins and losses is wonderful. But trying to rank teams based on DVOA is really a very silly exercise.

287
by Arkaein :: Thu, 12/17/2009 - 5:38pm

I think it's pretty unlikely that DVOA is ever going to improve greatly in terms of predictive power or statistical significance, there is simply too many factors including luck to make really accurate predictions. Any system (or person) that could reliably predict NFL winners 70% of the time would be the greatest prognosticator ever, so progress in terms of DVOA or other analytical stats is likely to come in small increments.

On your other note, you say ranking teams by DVOA is silly. Care to elaborate? Is it because you think ranking teams at all is silly, or because you think other measures would be more appropriate? All stats are flawed in some way, I'd say that DVOA is superior to probably any others you could list (for example, pt differential and yards per play both suffer from lack of opponent adjustments, at least in raw form, which is why I find DVOA superior).

291
by jmaron :: Thu, 12/17/2009 - 7:55pm

by silly I'm talking about saying one team is better than another because they have a DVOA 1 percent or even 5 percent better. But people like to do such things so DVOA is probably as good or better than any other system...but it seems like a hell of pile of work for something that isn't even as predictive as simple pt differential in season.

311
by Arkaein :: Fri, 12/18/2009 - 3:35pm

That makes sense, that's why Aaron usually says to pay more attention to the numbers themselves than the rankings.

Also, I'm not sure if it's really less predictive within season. As I understand that chart, it's showing correlation for the scores calculated for the entire season after the fact. The chart doesn't say, for example, how well DVOA and PtDiff after week 8 correlate with games in week 9. I suspect that DVOA would have an edge here, but you'd have to ask Aaron to be sure.

288
by Marver :: Thu, 12/17/2009 - 5:58pm

I think my basic problem with something like DVOA is how it weights offense as 3/7, defense as 3/7, and special teams as 1/7. Offensive statistics routinely correlate better to future success than defensive statistics...weighting them both equally makes little sense. Similarly, rating special teams based on how it has performed in that season makes little sense to me, too; everyone knows the best way to evaluate a kicker is based on his career FG%, not on a sampling of 20 of his field goals. And since kick and punt return/coverage units BARELY correlate from one season to the next, weighting it as 1/7th makes little sense EVEN IF SPECIAL TEAMS DOES CONSTITUTE 1/7 OF THE RESULTS OF AN NFL GAME.

For example, if half of an NFL game happened to be decided by coin flip, and the other half by play on the field, you shouldn't evaluate the teams based on the overall game. Any scientist can vouch that any good statistic should be repeatable; while including the coin flip may tell you what HAS happened, it has no bearing on what WILL happen. The same holds for coverage/return units.

Telling me the Chargers special teams is lackluster, even though it has the most accurate kicker in NFL history and the Mike Scifres/Kassim Osgood combination, is going to draw a red flag every time. So the kick/punt return units haven't been good this season so far...that has almost no bearing on how it will do in the future. Evaluate teams with that in mind.

292
by jmaron :: Thu, 12/17/2009 - 8:01pm

You are articulating my qualms with DVOA far better than I and you don't insult people (I need to learn both skills).

Knowing that special teams don't correlate well season to season was one reason I was high on the Vikings this year because they were so abysmal on special teams last year (primarily due to ret tds and blocked punts).

300
by Mr Shush :: Fri, 12/18/2009 - 9:07am

Special teams league-wide may not correlate season to season, but there do seem to be some teams which have consistently good (Bears, Bills, Texans, Eagles-under-Harbaugh) or bad (Dungy's Colts) special teams. Some return men lack the speed to take kickoffs back for touchdowns, even if they get good blocking. That is repeatable (though a weak effect relative to randomness, I grant). The kickoff performance of kickers (as distinct from kick-off units) is strongly predictive. Guys like Mike Scifres and Shane Lechler are clearly consistently excellent punters, hence punting performance must be meaningfully predictive.

I don't think it's sensible to simply toss out past special teams performance in evaluating a team going forward. I do think that DVOA probably places more emphasis on high leverage, low frequency events like interceptions, blocked punts, return touchdowns and so on, than it should if it sought to be a predictive system. I would rather it was more predictive, but it's clear that FO are more interested in describing the value that a team's play has had than in predicting what they will do in the future.

304
by C (not verified) :: Fri, 12/18/2009 - 11:10am

Jmaron

His post really was fantastic and yours made me laugh

273
by bravehoptoad :: Thu, 12/17/2009 - 1:12pm

Offensive passing efficiency correlates higher to future success than any other statistic; the Chargers rank 2nd.

Not acccording to what I've seen: pass efficiency has a correlation of .84; points per drive is at .92 and touchdown % is .88.

jmaron got it right...the big advantage of DVOA is that it's so sliceable and diceable. If it's vaguely in the same neighborhood as point differential, that's good enough, because then you can use it look at any single aspect of the game, like blitz effectiveness in different formations or coverage vs. tight ends etc. You can't do that with these other stats.

286
by Marver :: Thu, 12/17/2009 - 5:37pm

Of course, a proper metric would use a linear combination of (nearly) independent variables, like pass efficiency, run efficiency, defensive pass efficiency, defensive run efficiency, turnover rates, etc.

The reason I don't particularly like point differential is because it ends up being swayed by how many turnovers a defense forces, when defensive interception rate barely correlates (.09) with future interception rate. Basically, the control over when an interception occurs is almost entirely on the quarterback throwing the pass; so you end up with defenses having much lower points allowed simply because of a high interception rate due to playing a much weaker slate of QBs. Obviously that wouldn't correlate quite as well.

If you want good rating systems that'll predict future outcomes, I recommend Massey's Ratings (his system is actually a component of the BCS) or advancednflstats.com

294
by jmaron :: Thu, 12/17/2009 - 8:25pm

I took a look at both of those sites - the Advanced NFL Site ratings seem to really love the NFC East and AFC West teams. Here's my problem with that - those teams have been horrendous when they have played other good teams (defined for these purposes as top 15 in DVOA - which looks not to different from top 15 in the Advanced NFL stats - just a much different order)

NFC East is 0-9 and negative 102 pts against top 15 teams (NO, Arz, GB, Den, SD). This of course excludes games they played among each other. Further SD and Den are 2-5 against those same top 15 teams other than playing each other and the NFC East (Balt, Pitt, NE, Cinn - might have missed a team in there).

Those teams must be compiling stats the Advanced Stats guys like while they play each other and other weak teams like Oak, KC, TB, Car, etc.

Personally I find hard to believe that if those 3rd, 4th, 5th, 6th, and 8th best teams in the league - they would have a 2-14 record and over -160 pt differential when playing other teams ranked 1-15.

307
by Rick A. (not verified) :: Fri, 12/18/2009 - 1:21pm

And not a one of the stat guys and "logicians" on this thread could've said that better. I suppose they will all try to insist the results you cite are not relevant, perhaps even attack you in spite of the spirit of civility that seems to have momentarily broken out here. No one other than you wants to look at the facts of the results you point out, which you have pointed out before. They will continue to say "DVOA is not predictive", but certainly they are hoping for it to be, that's why most of them are into it to begin with...

309
by jmaron :: Fri, 12/18/2009 - 2:18pm

I try to draw value from both camps. When I first started reading sabermetric stuff from Bill James I thought that they had all the answers, but traditionalists had some good knowledge as well.

I think getting your ego attached to your argument is the surest way of coming up with a less than ideal answer to any problem

310
by jmaron :: Fri, 12/18/2009 - 2:19pm

I try to draw value from both camps. When I first started reading sabermetric stuff from Bill James I thought that they had all the answers, but traditionalists had some good knowledge as well.

I think getting your ego attached to your argument is the surest way of coming up with a less than ideal answer to any problem

259
by jmaron :: Wed, 12/16/2009 - 11:39pm

Aaron - I notice in the FO FAQ you produce correlations for Year+1 correlations for

DVOA .33
VOA .31
Pt differential .26
Wins .24
and yard differential .21

but you don't list yards per play differential even though you list it for comparative purposes in correlation in the same year to wins . Do you have that correlation for y+1? Have you updated those correlations since 2005?

314
by Marver :: Fri, 12/18/2009 - 7:45pm

I think you'll appreciate this list of self-correlations:
http://www.advancednflstats.com/search/label/offense%20vs%20defense?max-...

As well as this one outlining how teams win:
http://www.advancednflstats.com/2007/07/what-makes-teams-win-3.html

You'll notice the four most important factors on determining what makes a team win: O Pass efficiency, D Pass efficiency, O FUMBLE%, O INT%. The Chargers rank near the top (2nd, 12th, 1st, 4th) in all four, and all four are fairly self-correlated: .58, .29, .48, .27.

319
by cfn_ms :: Sat, 12/19/2009 - 7:27pm

Does anyone have insight into why Philly is ranked as substantially better than Minnesota? I'd think two more wins and close to the same schedule would at least put Minny near the Eagles. What am I missing?

I'd also think the Colts were better than the Eagles as well, but that's a much more standard complaint, so I'm more curious about a case where intuitively it would seem that they'd be reasonably close.

320
by R O (not verified) :: Sun, 12/20/2009 - 11:31am

Go ahead and tell me again how much the Chargers stink after the team they beat down on the road, went on the road and beat down the Beloved Saints.