Week 6 DVOA Ratings

Week 6 DVOA Ratings
Week 6 DVOA Ratings
Photo: USA Today Sports Images

by Aaron Schatz

There's a bit of a shakeup on top of the Football Outsiders ratings this week, as the fifth-place and first-place teams switch places. Baltimore is now number one, narrowly edging past San Francisco. (I know I mentioned on this week's B.S. Report that San Francisco would be number one, but opponent adjustments changed after the Jets' victory on Monday Night Football, and that moved Baltimore slightly ahead.). Green Bay, the team that is unanimously considered the best in the league according to conventional wisdom, is third. (The Packers are also third in the pro-football-reference Simple Ratings System, so we're not entirely alone on this one.) Then come the three teams from the AFC East, but in a surprising order: the Jets first, then Buffalo and finally New England, the opposite of the order in the standings.

Perhaps the most notable thing about this week's DVOA ratings is how condensed they are. 2011 has been a season with a lot of competitive balance. I don't know if it is related to the lockout or not, but it is certainly evident in the record number of big early-season comebacks. In the 20-year history of DVOA, this is the first time that there has been no team over 30% after Week 6. Last year was like this too, which is a piece of evidence to suggest that this has nothing to do with the lockout. Last year, only Pittsburgh was above 30% after six weeks. But the year before, in 2009, six different teams were above 30% at this point.

The Ravens and 49ers are narrowly ahead of the Packers primarily due to schedule strength. You'll notice below that the Packers are number one in non-adjusted VOA, but slip behind two other teams once the opponent adjustments are applied. As great as the Packers offense has been, it hasn't been that much better than the rest of the league. In fact, it's not even on top in DVOA, New England is. However, the Ravens defense so far is lapping the field. The Ravens' defensive DVOA is nearly twice that of the second-place Jets.

As I noted on today's Numbers Never Lie, I do think the Packers are likely to be better than the Ravens going forward, and I would say they are likely to be better than the 49ers as well. I have more confidence in the Packers defense improving than I do in the Ravens or 49ers offenses improving. And of course, when it comes to making the playoffs I have a lot more confidence in the two NFC teams, because they have less competition. The playoff odds report gives San Francisco a 99 percent chance to make the postseason. We also get our first two teams to hit a 0.0 percent chance to make the playoffs -- at least, after rounding -- in Miami and St. Louis.

The Ravens moved ahead of the 49ers between Monday and Tuesday because of changes in their opponent adjustments for playing the Jets after the Jets had a strong win over the Dolphins. That gets us to the Jets' overall ranking of fourth, which is pretty unexpected. I assume nobody is going to fight about having the Jets ranked second on defense and third on special teams. We know those two units are strong, as they have been for years. On the other hand, you're probably wondering how the Jets' offense manages to rank 21st when they seem to go three-and-out roughly forty times a game and should maybe be ranked, I don't know, 34th, behind the rest of the NFL teams and Oklahoma State.

I have to admit to being a little confused as well. This is where I will come clean and talk about something I didn't want to talk about because I'm not sure I can explain it: The very strange DVOA result of last week's Jets-Patriots game. It certainly looked to all of us like the Patriots were in charge the whole game, right? Yet somehow, DVOA came out saying that the Jets had played better than the Patriots on a play-by-play basis. Using current opponent adjustments, the Jets get 33.9% DVOA for the game while the Patriots get only 3.3% DVOA. Yes, that seems as crazy to me as it does to you. The Jets got 29.7% DVOA on offense even though they went three-and-out seven different times and were penalized for playing a bad Patriots defense. On defense, the Jets let the Patriots into the red zone six times but only allowed three touchdowns, and of course Tom Brady threw the interception that bounced off Aaron Hernandez's hands. The offense didn't have time to get on the field and capitalize on that field position, but DVOA still sees that as a great play from the Jets defense.

Still, as I noted earlier, the big issue is not the Jets defense; it's the Jets offense. The weird results aren't just coming against the Patriots. Last night they went three-and-out five times, and four-and-out another two times, and were once again penalized for playing a bad defense, and once again got a positive offensive DVOA (7.2%). Why do we have them at a reasonable 21st in offense, which is enough to lift the entire team's rating up to fourth overall? The typical reasons why conventional stats underrate teams are not in play here. The Jets don't have lousy fumble luck on offense, and they haven't played a hard schedule of opposing defenses.

It looks like the biggest issue here is the red zone. The Jets have been really, really good in the red zone. On the first eighty yards of the field, the Jets have -7.6% DVOA, which ranks 28th in the NFL. But they have a 31.9% DVOA in the red zone, which ranks third in the NFL -- and remember, plays in the red zone are worth 25 percent more than other plays. The Jets may be going three-and-out a lot, but when they do have an extended drive, they tend to extend it into the end zone.

There's a good chance that the Jets' performance in the red zone will regress a bit towards the mean. You just don't see teams that are this much better in the red zone compared to their performance overall. This is one of the places where I'm stuck with my decision to make DVOA a stat that balances being predictive with being descriptive. Making red zone plays 25 percent more important than other plays really improves the correlation between DVOA and wins, but that expected red-zone regression makes it a little less predictive when you have a team like the Jets getting so much of its offensive value from red zone plays. This surprising Jets rating also has me wondering about the feasibility of creating an NFL rating that combines DVOA with drive-based statistics, much like we combine our play-by-play and drive stats for college into "F-plus." Unfortunately, that's a project for an offseason, so for now we'll have to live with the Jets in fourth place. Just know that for now, I'm not buying the idea that the Jets are really the fourth-best team in the NFL.

However, are the Jets the best team with a .500 or losing record? I'll buy that, and you should too.

* * * * *

These are the Football Outsiders team efficiency ratings through six weeks of 2011, 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.

Opponent adjustments are currently at 60 percent strength and will steadily grow stronger until Week 10. As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.

DAVE is a formula which combines our preseason projection with weighted DVOA to get a more accurate forecast of how a team will play the rest of the season. Right now, the preseason projection makes up 19 percent of DAVE for teams with six games and 27 percent of DAVE for teams with five games. Because DAVE uses weighted DVOA rather than total DVOA, Week 1-2 results are slightly discounted.

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>

1 BAL 27.8% 5 24.2% 1 4-1 0.1% 20 -29.5% 1 -1.8% 24
2 SF 27.5% 2 21.1% 5 5-1 2.9% 18 -11.2% 3 13.4% 1
3 GB 25.4% 3 22.6% 3 6-0 35.8% 2 9.0% 19 -1.4% 21
4 NYJ 25.2% 6 23.5% 2 3-3 -1.4% 21 -15.3% 2 11.3% 3
5 BUF 23.9% 1 18.5% 6 4-2 31.3% 3 8.1% 17 0.7% 13
6 NE 21.1% 4 22.3% 4 5-1 37.4% 1 17.2% 27 0.8% 12
7 NYG 18.9% 12 16.8% 8 4-2 18.8% 7 -1.6% 8 -1.6% 23
8 TEN 17.0% 8 6.7% 14 3-2 19.1% 6 -0.7% 11 -2.7% 26
9 OAK 13.6% 9 8.4% 11 4-2 21.6% 5 10.6% 21 2.6% 6
10 PIT 13.0% 10 16.9% 7 4-2 11.8% 9 -1.3% 9 -0.1% 17
11 CIN 12.3% 15 7.8% 12 4-2 3.5% 17 -7.1% 6 1.7% 8
12 ATL 12.0% 18 11.5% 9 3-3 10.7% 12 -0.8% 10 0.4% 15
13 NO 10.4% 7 10.1% 10 4-2 22.6% 4 13.6% 24 1.4% 9
14 DET 9.5% 13 6.8% 13 5-1 3.5% 16 -8.9% 4 -2.9% 27
15 DAL 8.0% 17 4.0% 17 2-3 3.9% 15 -7.1% 7 -3.0% 28
16 HOU 5.5% 14 5.5% 16 3-3 9.7% 13 6.4% 15 2.2% 7
17 PHI 1.8% 21 5.9% 15 2-4 11.6% 10 8.8% 18 -1.0% 19
18 WAS 1.6% 16 -1.8% 22 3-2 -6.2% 27 -7.2% 5 0.6% 14
19 TB 0.2% 20 -1.2% 20 4-2 11.5% 11 16.3% 25 5.0% 4
20 MIN -1.5% 11 -1.7% 21 1-5 2.5% 19 3.4% 14 -0.6% 18
21 CHI -2.4% 25 -0.4% 19 3-3 -3.7% 25 10.7% 22 12.0% 2
22 SD -7.5% 19 0.8% 18 4-1 6.5% 14 10.9% 23 -3.1% 29
23 DEN -8.9% 22 -10.4% 24 1-4 -3.2% 24 9.2% 20 3.6% 5
24 CLE -10.4% 23 -8.0% 23 2-3 -4.6% 26 7.1% 16 1.3% 10
25 SEA -15.2% 27 -16.6% 25 2-3 -8.7% 29 3.3% 13 -3.1% 30
26 CAR -18.7% 24 -17.7% 27 1-5 12.9% 8 22.8% 32 -8.8% 32
27 KC -22.8% 28 -19.1% 28 2-3 -6.5% 28 16.5% 26 0.3% 16
28 MIA -26.8% 26 -17.2% 26 0-5 -2.4% 23 22.3% 31 -2.1% 25
29 IND -26.9% 29 -24.0% 30 0-6 -1.5% 22 17.7% 28 -7.7% 31
30 JAC -27.6% 31 -22.0% 29 1-5 -26.8% 32 -0.4% 12 -1.1% 20
31 ARI -29.0% 30 -25.8% 31 1-4 -11.5% 30 18.6% 29 1.1% 11
32 STL -36.9% 32 -30.6% 32 0-5 -15.8% 31 19.7% 30 -1.5% 22
  • NON-ADJUSTED TOTAL DVOA does not include the adjustments for opponent strength or the adjustments for weather and altitude in special teams, and only penalizes offenses for lost fumbles rather than all fumbles.
  • 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).

1 BAL 27.8% 4-1 29.9% 4.4 2 4.8% 10 -3.0% 26 29.6% 32
2 SF 27.5% 5-1 25.6% 4.1 4 2.8% 13 -9.8% 32 10.2% 18
3 GB 25.4% 6-0 30.4% 5.0 1 -7.4% 29 1.8% 13 1.6% 3
4 NYJ 25.2% 2-3 21.6% 3.5 13 2.7% 14 2.5% 12 13.8% 24
5 BUF 23.9% 4-2 25.6% 4.0 5 4.3% 12 2.8% 8 9.0% 17
6 NE 21.1% 5-1 22.5% 4.1 3 1.9% 15 -0.1% 18 13.1% 22
7 NYG 18.9% 4-2 21.4% 3.9 8 -9.0% 32 12.9% 1 13.7% 23
8 TEN 17.0% 3-2 15.4% 3.9 7 -1.2% 21 -3.9% 29 20.6% 28
9 OAK 13.6% 4-2 16.9% 3.9 6 9.4% 6 -5.8% 30 4.1% 8
10 PIT 13.0% 4-2 11.4% 3.6 12 -6.1% 28 -0.9% 23 27.3% 31
11 CIN 12.3% 4-2 24.3% 3.8 9 0.8% 18 1.3% 15 2.5% 5
12 ATL 12.0% 3-3 7.0% 3.6 11 -1.5% 22 -2.2% 24 3.4% 7
13 NO 10.4% 4-2 10.7% 3.6 10 -2.9% 24 -2.6% 25 9.0% 16
14 DET 9.5% 5-1 20.0% 3.5 14 1.5% 17 3.6% 6 6.3% 12
15 DAL 8.0% 2-3 4.1% 3.4 15 15.1% 2 -3.7% 28 2.0% 4
16 HOU 5.5% 3-3 11.4% 3.3 16 1.9% 16 -7.0% 31 5.6% 10
17 PHI 1.8% 2-4 -3.5% 2.9 19 7.8% 7 0.1% 17 8.2% 14
18 WAS 1.6% 3-2 3.7% 3.3 17 -7.5% 30 3.3% 7 6.3% 13
19 TB 0.2% 4-2 1.8% 3.2 18 9.6% 5 1.1% 16 19.3% 27
20 MIN -1.5% 1-5 5.3% 2.9 20 -4.9% 27 6.8% 3 26.0% 30
21 CHI -2.4% 3-3 0.3% 2.6 23 6.2% 9 -0.5% 21 15.1% 25
22 SD -7.5% 4-1 3.5% 2.4 24 -7.8% 31 5.8% 4 8.3% 15
23 DEN -8.9% 1-4 -9.9% 2.8 21 12.2% 3 -0.4% 19 3.2% 6
24 CLE -10.4% 2-3 -2.8% 2.4 25 -2.2% 23 -0.9% 22 0.1% 1
25 SEA -15.2% 2-3 -20.2% 2.7 22 4.7% 11 -3.3% 27 15.5% 26
26 CAR -18.7% 1-5 -15.1% 2.3 26 -3.9% 26 2.8% 9 6.1% 11
27 KC -22.8% 2-3 -27.0% 2.1 28 -0.2% 19 5.2% 5 24.7% 29
28 MIA -26.8% 0-4 -25.1% 2.2 27 6.8% 8 9.7% 2 4.8% 9
29 IND -26.9% 0-6 -29.7% 1.8 30 -0.3% 20 2.6% 10 11.6% 20
30 JAC -27.6% 1-5 -29.7% 1.9 29 9.9% 4 -0.4% 20 12.6% 21
31 ARI -29.0% 1-4 -21.8% 1.7 31 -3.0% 25 1.7% 14 10.3% 19
32 STL -36.9% 0-5 -43.5% 1.6 32 15.1% 1 2.6% 11 1.0% 2


138 comments, Last at 22 Oct 2011, 4:37am

1 Re: Week 6 DVOA Ratings

49ers @ #2 two weeks in a row. I guess we will get as close as possible to finding how much coaching matters in football in terms of wins at the end of year.

25 Re: Week 6 DVOA Ratings

At least Alex Smith is back in the late teens in QB DVOA where he belongs. And Gore is not the worst RB in the league.

2 Re: Week 6 DVOA Ratings

I am getting tired having to scroll twice before getting to the Broncos. That is the only drunken statement I have to contribute.

82 Re: Week 6 DVOA Ratings

Aw come on, when all you have to do is scroll to the bottom of the list, it makes them easier to find! Although, there are an alarming number of teams between them and the bottom. This is not helping them in the Andrew Luck Derby.

3 Re: Week 6 DVOA Ratings

The Bengals are clearly ranked too high because they lost to Denver. DENVER! WTF?

4 Re: Week 6 DVOA Ratings

Just as strange as the Pats/Jets game was the Pats/Bills game. According to DVOA, Buffalo outplayed NE by something like 90%, which can only be attributed to the turnovers since virtually every other statistic came out even.

It does seem like DVOA has more of these "doesn't pass the eye test" examples than I recall in prior seasons.

5 Re: Week 6 DVOA Ratings

Harbaughs, 1 & 2.

It's a Harbaugh world; we're only living in it.

7 Re: Week 6 DVOA Ratings

... and you will notice the much-requested "Harbaugh Bowl" has been added to the "special Super Bowls" on the odds report.

BTW, if you are looking for the specific off/def/st tables, wait a few minutes, I'm just getting those online now.

10 Re: Week 6 DVOA Ratings

Oh wow, and along those lines, the odds for the "FFFFFFFFFUUUUUUUUUU Montana x 2" (John Candy Bowl) rematch would actually make the table at 1.7%.

44 Re: Week 6 DVOA Ratings

Where do the Skins rank per a D standpoint? They have only allowed 8 TDs which is a number stemmed from opponents such as Dal, NYG and Phi. I know people hate on Beck as much as Grossman, but if anyone wants to see how to shuffle around a pocket with the ball high and ready for release, watch Becks tape. No, he is not the next Brady, but he is 2x the QB as Rex and will light CAR up this weekend.

66 Re: Week 6 DVOA Ratings

Given that Carolina's D is absolutely horrible, Beck would need to light them up just to look like an average QB.

6 Re: Week 6 DVOA Ratings


Do you ever plot DVOA vs. time for head-scratchers like NE vs. NYJ? Would that work? It might help you locate plays/drives where DVOA diverges from your subjective impressions

8 Re: Week 6 DVOA Ratings

Obviously this is too big a question to actually be answered here, but what IS the point of trying to make DVOA both descriptive and predictive at once? Doesn't that undermine it as both a predictive and descriptive tool? And, perhaps I've just never read the descriptions that closely, but to be honest, I always thought it was entirely descriptive.

18 Re: Week 6 DVOA Ratings

I agree. My suggestion: why not just have two separate values? pDVOA could be a separate column that attempts to be predictive and takes into account things like regression to the mean for important variables.

Also, I'm curious how the San Diego offense breaks down in DVOA for the red zone vs. the rest of the field. I think they are the anti-Jets: awful in the red zone but moving the ball at will everywhere else. I wouldn't be surprised to see their rating go way up once they solve their red zone difficulties (assuming that they do ...).

56 Re: Week 6 DVOA Ratings

At this point in the season, we want DVOA to tell us who's played well (descriptive) and who's going to play well (predictive). Of course, those correlate well, so the combination isn't ridiculous.

81 Re: Week 6 DVOA Ratings

I think the red herring here is the continual tweaking to the DVOA formula to improve its correlation with wins.

Wins have a perfect correlation with wins; the fundamental idea behind DVOA (or the whole project of statistical analysis, for that matter) is that wins are a flawed, noisy proxy for "true team strength". But chasing that correlation with wins, you end up doing things, like giving a bonus for red zone performance, that aren't justified by an underlying model, but rather because doing so makes the correlations look better.

Now, in this particular situation, it is (tautologically) true that better red zone performance will result in more wins, and nearly as clearly true that strong teams will tend to have strong red zone performances. But by giving a bonus for some plays that happen to correlate with victory, you are, in effect, baking small-sample size variance into your computation of DVOA. (And, on a more technical level, 25% seems like a suspiciously round number to use as a red zone bonus.)

Red zone performance actually seems very similar to fumble recoveries, in fact. You have a small facet of the game that highly correlates with winning, in a way that a small number of plays have a disproportionate effect on the outcome of games. Now, teams almost certainly have more than random control over their red-zone performance than they do over fumble recoveries, but I see no reason to suspect the baseline hypothesis (in the long run teams tend to perform just as well in the red zone as they do at midfield) is far from the truth.

To me, it seems much more valuable to be able to say that a team has over/under-performed relative to their record, (and further, to be able to point out why that is--fumble luck, field-goal luck, unsustainable red-zone performance, etc.), and let "record" function as a descriptive statistic--something that it actually does quite well.

In short, I'd much rather a more predictive DVOA (that better measures "true" team strength) that occasionally ranks a 9-6-1 Eagles team at #1, than one that sacrifices the ability to get at "true" strength in order to better track the final standings.

83 Re: Week 6 DVOA Ratings

Wins have a perfect correlation with wins; the fundamental idea behind DVOA (or the whole project of statistical analysis, for that matter) is that wins are a flawed, noisy proxy for "true team strength".


The number of wins in week 4 do not have a perfect correlation with the number of wins in week 16. DVOA is tuned to correlate to total season wins.

Wins aren't the problem (and VOA is tuned to wins, not DVOA - the wording in the article is a bit unclear, but look at the methods page).

Descriptive portion: tuning to correlate to another variable using the same set of data
Predictive portion: tuning to correlate to another variable using an independent set of data (e.g. tuning variables to get the data from games 1-8 to correlate with wins from games 9-16)

Now the problem with the predictive portion is that wins are a poor metric, as you said. So what you do is you tune first to wins in the same set (so VOA becomes a less noisy proxy for wins) and then again tune DVOA in the first set to correlate to DVOA in the second set, while not having the win correlation of VOA drop too much. Note the switch from VOA to DVOA - that's because wins aren't opponent adjusted, so you correlate VOA to wins, but you want the correlation to stay opponent-adjusted.

That's the compromise, and it's really just because we don't get a chance to play the season multiple times and the number of games is sparse.

In a game like baseball, you could tune to pure predictivity because you can slice and dice the datasets up and they're still reasonable-sized.

9 Re: Week 6 DVOA Ratings

I'd love to see a DVOA histogram distribution by year. In addition to no real dominant teams (and a lot of good, but not great teams) is seems like there are a lot bad-but-not-awful-teams this year. I'm sort of guessing that this year that the distribution looks a little bit bi-modal in comparison to past years.

13 Re: Week 6 DVOA Ratings

So many good teams are over the hump and have much easier schedules coming up. Good teams playing easy schedules tends to inflate DVOA, so that could send some teams higher.

Baltimore: 10th-ranked schedule to 26th
49ers: 13th to 32nd
Oakland: 6th to 30th
Dallas: 2nd to 28th

23 Re: Week 6 DVOA Ratings

On the opposite end are the Giants, who go from the easiest schedule in the NFL over the first 6 weeks to the hardest schedule over the remainder of the season.

Let's see how Eli bears up.

11 Re: Week 6 DVOA Ratings

Thanks for posting the Harbaugh Bowl odds.

You may also want to consider posting the Jim Schwartz Homecoming Bowl odds in the future. Schwartz grew up in suburban Baltimore (attended the same high school as Mark Texeira).

12 Re: Week 6 DVOA Ratings

"I have more confidence in the Packers defense improving than I do in the Ravens or 49ers offenses improving."

The niners have been in their current scheme for three months and have yet to put their two starting recievers on the field. But no room for improvement.

77 Re: Week 6 DVOA Ratings

So you're saying a team that has already improved significantly in only ~3 months under a new scheme is not likely to improve further? Sounds very implausible to me.

96 Re: Week 6 DVOA Ratings

No, that's not what I'm saying. I was making fun of the concept of "room to improve." The Patriots' defense has a lot of room to improve! That doesn't mean that they're going to do so!

And what I'm saying is not intended to be a slam on the 49ers. But this is what I took from the comment about the relative likelihoods for the different units to improve: Aaron thinks it's highly likely that the Packers' defense will improve. After all, this is the unit that rated #2 in DVOA in the NFL last year. With that as prior knowledge, it seems reasonable to think that they can return to that form. The 49ers offense does not have this kind of recent history. Moreover, they are hampered by relying upon a widely-ridiculed starting QB who is viewed to have a low ceiling to his ability.

I would not be surprised to see the Packers have a top 5 defense by the end of the year. I would be surprised to see the 49ers' offense make a similar leap.

14 Re: Week 6 DVOA Ratings

Raiders 9 and moving on up with addjtion Carson Palmer. Team licking chips getting reasy for KC. This going to br real spanling

89 Re: Week 6 DVOA Ratings

Accounting for typos, I think you meant to say "Team licking chops getting greasy for KFC". Are you talking about the Raiders or the Red Sox?

16 Re: Week 6 DVOA Ratings

I think the obvious explanation here is that DVOA has ended its love affair with the Eagles and moved on to the Jets.

17 Re: Week 6 DVOA Ratings

DVOA dumped Eaglrs after it see Andy Reid cheating on iy with DAVE

21 Re: Week 6 DVOA Ratings

What is the logic behind making red zone performance worth 25% more than regular performance?
1) Where did 25% come from?
and 2) Does past red zone performance seem to predict future red zone performance? In other words, is it a repeatable skill or simply a matter of luck?

45 Re: Week 6 DVOA Ratings

I don't know how arbitrary the 25% is but it does make some sense, since better performance in the redzone means more touchdowns, means better chance of winning. And since a touchdown is worth about twice that of a field goal 25% seems reasonable.

46 Re: Week 6 DVOA Ratings

Well, it's a substitute for keeping track of the yardage on each play. Doing that might lead to overfitting, but perhaps it would be possible to do something along those lines with 5-20 categories. (Separating red zone from non-red zone means two categories are being used.)

134 Re: Week 6 DVOA Ratings

There are more than 2 categories of plays - there are 5 total (deep, back, middle, front, and red - search for "zone" in the search bar). The other ones all have the same relative value, but plays are compared to plays within their own zone only.

Having a boost for red zone value (well, having red zone value be different than the others) makes sense to me mainly because the game is entirely different there since the field's compressed. "Bend But Don't Break" defenses have been fairly common, although I don't think they're by choice - I think they're due to having a few liabilities on defense which can be exploited, but then get hidden by the compressed field in the red zone (i.e. if your linebackers suck, but your defensive line and secondary are very good, deep routes + a pass-catching tight end can move you down the field well, but in the red zone, you can't isolate the tight end on a linebacker that easily anymore). But anyway, those teams are going to have better performance on a compressed field, and they're naturally going to allow fewer points (and win more games) than a team that is good in the open field and poor in the red zone for some reason (undersized corners, maybe).

26 Re: Week 6 DVOA Ratings

Gb had a proven defense while the 9ers have smith at qb. Gbs defense has looked better as of recently and is adapting to life without nick collins plus not having your best cornerback(top 5 cb) being able to play bump and run coverage since weeku one has really visibly affected their pass defense.

27 Re: Week 6 DVOA Ratings

I should say was proven last year and sorry if calling williams a top 5 cb is a stretch. I think it wass pff who had him ranked as the number 1 cb last year which partly had to do with high amount of targets thrown his way.. as a packer fan i will officialy start to worry if ponder puts up 250 yards against them next week

72 Re: Week 6 DVOA Ratings

Personally, I expect Ponder to throw for about 300 yards - and about 3 INTs - against the Packers this week. Shields will be out because of a concussion had received after stupidly running through the end zone after intercepting a ball last week, meaning Bush at CB opposite Williams unless they play a standard 3-4 with Woodson at CB. (I'd actually recommend this for any play Harvin is not on the field as I think they can generate a decent rush against the MN line.)

31 Re: Week 6 DVOA Ratings

I would really like to see how sibling rivarly (and the support the players on a team has for its coach - if only to make sure he's not shown up by his slightly crazy, football God brother) will play into this. As a Raven's fan, I'm thrilled for this season, but I will be on pins and needles on Thanksgiving night.

62 Re: Week 6 DVOA Ratings

Same here. That said, seeing that massive number in the Ravens' "Variance" column has me concerned on a week-to-week basis. I suppose it's great to know that when you're team's "on", it does so in a big way, but I'd be all the happier to see someone else take that ranking from them.

67 Re: Week 6 DVOA Ratings

I agree - I haven't figured out the variance number - it can't only be on Flacco/ol. Unless it means its statistically unusual to have such great numbers with week red zone success (which is on the whole team in my view). Really - 4 tries (including moving the ball closer twice and having penalty pauses)on that first drive against Texans? That is reverse efficiency.

86 Re: Week 6 DVOA Ratings

Maybe some of the variance is due to the inexplicable beating they took from the Titans, in comparison to the whipping they put on three of the AFC's best non-NE teams. They are outscoring opponents by a more than 2-to-1 margin, which is always something I look for in identifying a truly elite team. I don't think the Ravens are that, but with the weakness of the rest of their schedule, and a serious shot at going 13-3 (they'll be favored in every remaining game except maybe at PIT and at SD), I'll be interested to see if they make a run at a season-long 2-to-1 scoring margin.

94 Re: Week 6 DVOA Ratings

I'd think it's actually indicative of their overall offensive variance; if you look at game by game numbers, you end up with games 1-3-5 having good to dominant offensive performances, whereas games 2-4 obviously were not so much (particularly 4). When one critical aspect of your team is varying that widely (over an admittedly small sample, even for football stats), it's going to result in a big variance number.

I'm somewhat optimistic that they'll settle in, but they're not going to be catching up with the Packers this season. Historically, the Ravens' overall DVOA performance charts always look like a drunken lightning bolt (in my experience). The team simply never puts together back-to-back (let alone to another back) dominant performances. They still win more than they lose, but they'll scare the hell out of you doing so.

115 Re: Week 6 DVOA Ratings

I think it was Mike Florio who said, this week, it doesn't really matter what your view of Flacco is, he'll prove you right (or wrong) the next quarter - sums it up. I'm never happy until a) they have the ball; b) there is only 4 mins left; c) they have a three score lead.

Fun this year though - loving Smith and Suggs

32 Re: Week 6 DVOA Ratings

Any thoughts on what the difference is in each teams 'actual' DVOA vs 'calculated' (for lack of a better term)? +/- 10%? (i.e. the Packers could really be between 15% and 35%)

33 Re: Week 6 DVOA Ratings

I said it last week, but Dallas has played a brutal schedule and they should be clearly favored to pull away in the NFC East, especially since they are getting healthier.

47 Re: Week 6 DVOA Ratings

The Eagles should be favored because they clearly have the most talent.

The Giants should be favored because they're in first right now.

The Redskins should be favored because, well, they shouldn't really.

Tell you what - you take one and I'll take the other three. I picked the Giants in the preseason, but I wouldn't be surprised by any result other than the Cardinals winning the division.

58 Re: Week 6 DVOA Ratings

This year in the NFC East reminds me a lot of 2006. None of the teams were really great (although the 10-6 Eagles tied for the 2nd best record in the NFC - mostly because NO kind of rested late). Early on, the Giants ran off to a 6-2 record, then the schedule started. The Cowboys, with Romo taking over, overtook the Giants. The Eagles finally ended it with their season ending 5-1 run with Jeff Garcia.

Now, other than two teams doing it after being spurred by QB changes (one due to benching, the other due to injury), I think this year will be similar. The NFC East will go through numerous changes on the top, first as the Giants enter their tough part of their schedule, and then the Eagles red zone offenses regresses up to the mean.

What I also find interesting is that the top three teams are similar in the same ways: they just make stupid turnovers and stupid plays at the wrong time. They are all the NFC's version of the 2008-2011 Chargers in that they will find silly way to lose games. One team throws a deflected pick-six in the red zone to lose. One other misses two field goals and tries a ridiculous lateral on the goal line. The other wins a game by kicking six field goals, and then loses two games after leading by 10 and 24. Should be a fun year, but I think inevitably meaningless, since I don't think any of these teams are going to Super Bowl XLVI.

99 Re: Week 6 DVOA Ratings

The Giants' poor play down the stretch of 2006 had absolutely nothing to do with strength of their schedule. I would say it had a lot more to do with season ending injuries to Michael Strahan and Justin Tuck, as well as having a banged-up Osi Umenyiora... you know, the 3 guys who should've shared the Super Bowl MVP award the next season.

35 Re: Week 6 DVOA Ratings

Can't believe the Bengals are above the Saints. Saints obviously laid an egg this past week, but they've seemed like a top-5 team to me so far. I'm from Ohio, and usually have to switch off the Bengals games because they are unwatchable

36 Re: Week 6 DVOA Ratings

Are the playoff odds based on 20000 or 10000 simulations? The intro text says both.

37 Re: Week 6 DVOA Ratings

Is the same thing that's propping up the Jets offensive rank - red zone performance - responsible for the Bills offensive rank @ 3 as well? Because as much as the offense has improved this year, it does not seem like it's been 3rd-in-the-league good. Where we do excel is in the red zone, where in 21 trips we have 16 TD's and 5 FG's (I believe those are right - might be off by one) - so maybe that's inflating the Bills' apparent offensive prowess a bit.

Of course, we're probably closer to 3rd in the league in offense than we are to 17th in the league in defense, so...

38 Re: Week 6 DVOA Ratings

The competition in the Suck for Luck will be intense by week 14. I predict some controversy over thrown games.

40 Re: Week 6 DVOA Ratings

Re: Jets.

4 3-and outs = 12 plays (mosly unsuccessful ones)
2nd quarter TD drive = 11 plays. (mostly successful ones)

Some other team may have had 12 unsuccessful plays and 11 successful plays, resulting in 4 shorter drives with no TDs.

Does this explain the finding?

42 Re: Week 6 DVOA Ratings

Tampa's at 0.2%. Woohoo! Very, very, very, very slightly over average!

61 Re: Week 6 DVOA Ratings

Hate to break the bad news, but pretty sure you don't understand the statistic at all.
DVOA is Defense-adjusted Value Over Average. So by definition 0% is average.

65 Re: Week 6 DVOA Ratings

Except that offense has been improving recently. From the 29 Dec 2010 Week 16 DVOA Ratings: "the DVOA baselines are based on the period between 2002 and 2007. As strong as the offensive environment was during those seasons, it has been even stronger over the last three years." So while you have the right idea, John's also correct about the upward drift.

43 Re: Week 6 DVOA Ratings

On the other hand, you're probably wondering how the Jets' offense manages to rank 21st when they seem to go three-and-out roughly forty times a game and should maybe be ranked, I don't know, 34th, behind the rest of the NFL teams and Oklahoma State.

This is a great line, but being behind the rest of the NFL teams and Oklahoma State would leave them ranked 33rd.

51 Re: Week 6 DVOA Ratings

They'd be 34th, behind the 31 other NFL teams, Oklahoma State, and the platonic ideal of the Jets, not necessarily in that order.

49 Re: Week 6 DVOA Ratings

I think Scott B is right. Going 3 and Out while very bad in real life isn't that terrible by DVOA since it is just 3 unsuccessful plays and even multiple 3 and Outs can be offset by a few long drives.

Contrastingly, GB will probably never Pass NE in Offense since GB's plethora of 35+ yard TDs will never accrue enough value compared to NE's 8-14 playa Drives

53 Re: Week 6 DVOA Ratings

Actually it's the NE run game that is keeping them in front of GB.

Team - Overall -- Pass -- Run
NE --- 37.4% (1) 63.5% (2) 19.7% (3)
GB --- 35.8% (3) 78.4% (1) 02.7% (14)

The 35+ yard passing TD's seem to not be hurting the passing offense at all. It's first in the league by a wide margin (15.1%). It's just the New England has an excellent run game by DVOA (MIN is #1 at 24.9% so only 5.2% difference). Green Bay is middling with their 2.7%.

As a note, NE finished last year with a 72.5% Passing DVOA, and a 75.4% in 2007. I don't think any other team has finished over 70% (2004 Indy was 69.1%). That Green Bay passing game, really is spectacular. The only other team to break 60 was the 09 San Diego at 63.7%. There have been a number over 50, including Dallas in 95 and San Fran in 94. So yeah, if they keep this up it's a potential record breaking passing attack (and NE would still have a top 5 at their rate). #3 Buffalo is at 47% this year so yes the whole league is inflated as has been mentioned, but the overall numbers have been dropping and they drop as the year goes on as well. We'll see. GB has been very consistent with it too. Rodgers passing DYAR (just passing not total) has been 200, 114, 151, 189, 169, 143 in that order (average of 161, mean of 160). It's a pretty rough approximation for the entire passing game, but it conveys the point.

If they want the best overall offense they need to run the ball better though. It's not the passing game, or the long passes that are holding them back. :)

64 Re: Week 6 DVOA Ratings

I'm not sure that's really a refutation since the run game is presumably a part of those 9-14 play drives.

73 Re: Week 6 DVOA Ratings

DVOA is a rate stat, so it is irrelevent how many runs were taken away due to big passing plays.

Beyond that, GB is rushing for 3.8 ypc and NE is at a nice 4.6 clip.

74 Re: Week 6 DVOA Ratings

The original poster said the problem was that the 35+ yard plays (which have all been passes) were why Green Bay had a worse DVOA. Green Bay has a better pass DVOA. If they had a better run game, you can't assume the pass game would get worse (NE until that game vs Dallas was about 70% passing DVOA as well and the run game was about the same). The assumption is they would have less opportunity for the big plays, but I don't think that is correct. If they had a better running game, they would simply have more offensive plays and likely would score 7-10 more points a game. :) GB does have 9-14 play drives now. Most of their punts are the result of failed runs (from my memory, I haven't looked at the data), remove those, they then have the 35+ yard plays and they have more 9-14 play drives as well, which means more successful plays in total = higher overall DVOA, though most of it from the run game.

88 Re: Week 6 DVOA Ratings

QQ said that he thinks NE has a higher DVOA because NE uses more long drives and fewer long-play TDs than GB and DVOA value the longer drives more than the longer single plays. You said that the issue is that NE's run game is beter than GBs. But that's orthogonal. If DVOA values a bunch of short, but successful plays more than one long one then it doesn't matter whether those were runs or passes.

Consider, an 80 yard TD drive is an 80 yard TD drive. If NE gets one with 80 ten yard plays half of them runs, and GB gets one with 2 zero yard runs and an 80 yard pass then of course GB will have the lower run DVOA. That doesn't say anything about whether DVOA prefers a long drive to a quick drive.

And I'm not even saying DVOA is doing something wrong here. I can believe that the long drive is more predictive than the quick one and really should yield a higher DVOA. My only claim is that issue of short vs. long drive is not coupled to the issue of pass vs. run DVOA.

91 Re: Week 6 DVOA Ratings

The problem is not that GB's big plays are undervalued, as the passing DVOAs show. The problem is that they are bad at running and continue to do so, while the Patriots are not bad at anything.

95 Re: Week 6 DVOA Ratings

Just because GB has a higher Passing DVOA than NE does not mean that it's Long TD Passes are not still being undervalued. GB's Offense is much more vertical than NE's and as a result GB's scoring Drives typically have far fewer successful plays that can be racked up than NE's more methodical attack.

97 Re: Week 6 DVOA Ratings

DVOA thinks that GB's passing offense is one of the top 5 all time. I really doubt that its underrating them.

The problem lies entirely in their Run Offense not being good.

102 Re: Week 6 DVOA Ratings

It is pretty widely acknowledged that DVOA awards more points for a long 10 play Drive of 80 yards than a 1 or 2 Play 80 Yard TD Drive. As a comparison in the Offenses I looked at the lengths of all their TD Passes so Far this year:

GB TD Passes (Length)- 3,6, 7, 7, 7, 8, 10, 16, 17, 29, 32, 35, 49, 50, 70, 84,93

NE TD Passes (Length)- 1, 1, 2, 2, 4, 5, 6, 8, 10, 10, 14, 14, 15, 17, 26, 99

I think this comparison shows a pretty wide difference in Offensive Philosophy. GB has 8 TDS over 20 Yards, NE has 2. GB has 50 TDs of 49+ yards, NE has 2. Now obviously I have no idea how much it would affect their DVOA but it is pretty commonly accepted that their Offensive DVOA would be higher if they could replace those 50, 70, and 84 yard TD Passes with 10 Play Drives

105 Re: Week 6 DVOA Ratings

I wonder how much of this is due to the 25% Red Zone bonus, which the Packers aren't getting on half of their TD passes. Would be interesting to compare DVOA of GB versus NE without those bonues.

It would also be interesting to see some comparisons of DVOA for drives with similar yardage but different numbers of plays. I'm starting to suspect that DVOA may be making some bad assumptions about the distrobution/frequency of plays of different yardage, at least for certian teams.

106 Re: Week 6 DVOA Ratings

their Offensive DVOA would be higher if they could replace those 50, 70, and 84 yard TD Passes with 10 Play Drives

Only if all ten plays were successful. A typical ten play drive might involve seven successful plays and three unsuccessful (eg. one run for no gain and two incomplete passes). The successes rack up VOA, sure, but the failures also drop it down.

What it still seems is being missed is that it isn't the 50, 70, and 84 yard passes that are dropping Green Bay's VOA. It's the failed plays in between - runs for no gain being the most obvious of those. If every one of Green Bay's plays was an 80-yard touchdown pass, and every one of New England's plays gained exactly ten yards, wouldn't Green Bay have the higher DVOA? It's not the huge successes which lower their DVOA; it's the ratio of successes to failures.

In fact, could this be the opposite of the Jets situation? Where the Jets may have a better ratio of successful plays to unsuccessful due to going three-and-out so often (when they're bad, they're very bad), could it be that the Packers have a worse ratio because when they are successful they're -so- successful that they don't get to run more successful plays (when they're good, they're very good)?

107 Re: Week 6 DVOA Ratings

It is pretty widely acknowledged that DVOA awards more points for a long 10 play Drive of 80 yards than a 1 or 2 Play 80 Yard TD Drive.

I'm not sure that's the case.

It is true that DVOA likes a long, 10-play drive of 80 yards better than a 2-play drive of 80 yards and 8 unsuccessful plays.

109 Re: Week 6 DVOA Ratings

But should that be the case? A vertical scheme will have a lower per-play success rate than a ball-control scheme, but because it gains more yards for it's successes it doesn't need a high success rate to score the same number of points.

113 Re: Week 6 DVOA Ratings

A couple of things:

1). The reason why NE doesn't have so many long TD plays isn't because of a difference in philosophy. It's available talent. It's partly because Brady throws a lousy (relative to other elite QB's) deep ball. When he had a tall and talented Randy Moss to compensate was the only time the Pats had a deep passing attack. That one 99 yard pass this year actually only went about 18 yards through the air. The rest was Welker. And that's the other half. NE's best pass catchers are Welker (shifty, but not a speedster), Branch (a route running posession receiver), Danny Woodhead (a RB), and Gronknandez, who are fast for TE's, but still TE's. They don't have any receivers who are good at catching the ball (ruling out Slater) and also really fast enough to break long TD plays, even if Brady could accurately get them the ball.

2). I think it's exactly the case of the Jets, but in reverse. DVOA doesn't like short passes more than long passes...it certainly likes long passes better. While there is a damping of very long passes, there is a bonus for adding a TD, so those 80 yard TD passes are very very good. The problem is sampling rate. If you have 10 8 yard plays and get stuffed twice, you have a lot of good plays to minimize the effect of a couple of bad plays. If you get stuffed twice and then throw an 80 yard TD, you simply don't have enough good samples to counteract the bad samples. Remember, DVOA is per play.

Or, put another way, it's like a class that has 1 A+ student and 2 F students, versus a class that has 8 B students and 2 F students. The class average in the first case is going to be worse than the class average in the second case.

116 Re: Week 6 DVOA Ratings

The problem is that you can't average football plays together in the same manner, because an individual play is only important to the extent it impacts the current drive and/or scoreboard. I'm clear on what DVOA is doing, but I don't quite understand why.

Let's say there were two teams with two very different offensive schemes that both put up the same number of yards & points against the same opponents, one which had a normal distribution of yards/play and the other which had a bimodal distribution of yards/play, would DVOA value them the same? Should it?

121 Re: Week 6 DVOA Ratings

The numbers you want to balance aren't total points and total yards. They're points/drive and yards/drive. Teams alternate possessions.

Granted, if you had a big play offense that was so good that it had league-best points per drive and also left the opponent with bad field position when it didn't score, it would be good.

GB seems to fit that description pretty well, although Brady has more 20+ passes and more first downs than Rodgers. So the perception that GB is a big play offense while NE isn't may be just a perception based on a few highlight reel plays. Clearly, they are both quite successful.

The Packers are tied with NE in points per drive, and are just under 6 yards worse at improving field position. Field position is worth about a point every 15-20 yards, so you'd have to give a slight edge to the Patriots based on drive stats.

Lo and behold, DVOA gives NE a slight edge in offensive DVOA. It seems like DVOA is doing what it's meant to do.

122 Re: Week 6 DVOA Ratings

"GB seems to fit that description pretty well, although Brady has more 20+ passes and more first downs than Rodgers. So the perception that GB is a big play offense while NE isn't may be just a perception based on a few highlight reel plays."

I do not think that these 2 facts really indicate that NE is necessarily as big play of an offense.

First, a Big Play Offense should have fewer 1st Downs whereas a Small Play Offense would have more 1st Downs. NE having fewer 1st Downs than GB is what 1 would expect if their Offense is not as explosive.

In regards to the Second Point, with GB scoring from much further out distances than NE it isn't necessarily surprising that NE would have more in the 20 or so yard range

123 Re: Week 6 DVOA Ratings

Not the "20 or so yard range". The "20 or more yard range". GB has 3 more passes of 40 or more yards. But NE has 9 more plays of 20 or more yards. That's 12 more in the 20-39 yard range. Those are pretty big plays - with much more impact than a mere 3 plays of 40+ yards.

I don't think DVOA is shortchanging the Packers' offense at all. The objective facts are that they are slightly behind the Patriots in total drive effectiveness, as measured in points/drive (a tie) and yards/drive (advantage Pats). That's what DVOA indicates.

DVOA is designed to be derived from play-by-play results, but in a way that predicts future drive-by-drive success in terms of scoring and advancing field position. It works pretty well, and covers this case about as well as could be hoped.

125 Re: Week 6 DVOA Ratings

The problem is that you can't average football plays together in the same manner, because an individual play is only important to the extent it impacts the current drive and/or scoreboard.

DVOA is designed to be derived from play-by-play results, but in a way that predicts future drive-by-drive success in terms of scoring and advancing field position. It works pretty well, and covers this case about as well as could be hoped.

Yes, these two points are exactly what I've been driving at. Fundamentally, DVOA is a play-by-play metric that doesn't know about drives, and yet football is about drives...the team that wins isn't the team that gets the most yards, but rather the team that strings those yards together most effectively to turn them into points.

So DVOA has been engineered, by the definition of "success" taking down, distance, and field position into account, so that MOST of the time it rewards single play situations that tend to be effective at producing long, point-producing drives. But since it doesn't actually know about drives, it can make a mistake, if an offense or defense does something weird, especially with a small sample size.

Not the continue to belabor it, but look at the Jets-Patriots game last week. By VOA, the Jets gained 8-12 yards on about half of their offensive plays, and gained <0 - 4 yards on roughly the other half. VOA loved them for this, because in general, if you gain 8-12 yards 50% of the time, you will usually march down the field and score on almost every offensive possession (since in a given series you have three downs). However, the Jets actually only scored on two of nine possessions, because they weirdly stuck all their success into just two drives, and all of their failures into their other seven, which went 3-and-out.

Behavior like this will continue to be possible until drive context is included in DVOA, but given sufficiently large sample sizes, it should usually wash out.

127 Re: Week 6 DVOA Ratings

By VOA, the Jets gained 8-12 yards on about half of their offensive plays, and gained <0 - 4 yards on roughly the other half.

I think you're wrong about the 8-12 yard plays. In their Patriots game, the Jets ran 53 offensive plays. Greene had 4 runs of more than 7 yards (or a TD). Tomlinson had 1. Sanchez had 10 completions of over 7 yards. That's 15/53, which is not 50%. It's 28%.

What data are you using? I could be miscounting. But I don't think by that much.

130 Re: Week 6 DVOA Ratings

My bad. I was misremembering a quick analysis I did when this discussion point first came up.

The Jets were "successful" on about half of their plays, by the VOA success measure (40% of yardage needed on 1st, 60% on 2nd, 100% on 3rd or 4th). A fair number of those successes were more than what they needed and were in the 8+ish yard variety. In the intervening week, my brain had hyperbolized these numbers, since it's a more compelling argument the way I phrased it versus what actually happened. The actual argument remains unchanged, though. The Jets succeeded on more than half their plays, and got more than what they needed to succeed on many of those, meaning that VOA would expect them to have been able to make most of their possessions into long-ish drives. But they actually crowded almost all their successes into just two drives.

131 Re: Week 6 DVOA Ratings

That makes sense.

Oddly, by that same measure, the Patriots were successful 63% of the time, a much better success rate. True, they also had more bad plays. Again, they also had the 73 yard Welker catch, too, and a bunch more other longish plays.

That game's DVOA is still a mystery. The Patriots were vastly superior in terms of play success (63% vs. 50% means they were successful at a 26% higher rate than the Jets). They had longer plays, more yards, more first downs, and clearly more success on the field in terms of field position and scores. The good fortune of not having to pay for the one interception accounts for some mismatch between perception and DVOA. But it's really no where near enough. There must be something else wonky going on in the DVOA spreadsheet for this one.


are you making any progress on this? It's not okay to just say "It's a mystery". You should make this a case study. Either DVOA is right and the Jets really were doing what it takes to win at a higher frequency and/or greater degree than the Patriots, or DVOA has some very interesting limitations that your fans really want to know about. Either way this is great material for you to run with. Don't let us down.

133 Re: Week 6 DVOA Ratings

It's possible that this was one game where the Jets' good and bad plays were distributed oddly enough to lead to a strange one-game result. As MJK says in part above: "given sufficiently large sample sizes, it should usually wash out." That doesn't mean that Aaron shouldn't look at it to see if there's a systemic error and/or a possible improvement, but odd results in any one game may just be a fluke.

132 Re: Week 6 DVOA Ratings

Except that using points/drive and yards/drive creates the implicit assumption that drives/game is similar for the two different schemes. Granted, there is probably not a huge variance in the NFL, but the difficulty in analyzing Oregon over the past few years indicates that it's something which certianly needs to be considered.

119 Re: Week 6 DVOA Ratings

Football is a game based on drives.

If you have a vertical offense that fails 80% of the time, and gets a TD 20% of the time, you score a TD on roughly 1/3-1/2 your drives. If you get 10 yards on every play, you score every drive.

Consistency is extremely valuable when you're very good. Variance is valuable when you're not.

136 Re: Week 6 DVOA Ratings

In reply to by RichC (not verified)

This is a *huge* question. The game is based on drives, but the teams run plays. Do teams engineer drives? That is, does a drive succeed based on the play history of the drive? Or does a drive succeed because the plays in the drive succeeded?

It may seem simple, but it's not really a simple question. If a team has a bunch of drives where they have a string of successful plays and then they die out (even before the red zone, because failure in the red zone could be a symptom of the offense unable to perform on a short field), is this likely to continue, or is it just an aberration?

If you assume that teams engineer drives (that is, the "drive" is the fundamental unit of team performance) then it would be likely to continue. If you assume that drives are just successions of plays (so the "play" is the fundamental unit of team performance) then it would just be an aberration.

The equivalent in a sport like baseball would be "is a team's inning-by-inning performance the basic unit, or is it plate appearances?" We know the answer there - it's plate appearances. But in football it's a fair amount more open.

114 Re: Week 6 DVOA Ratings

New England's higher ODVOA has nothing to do with GB's long passes. GB has a higher passing DVOA than NE. It has everything to do with the fact that GB is bad at running the ball. GB has more negative plays dragging down their superlative passing attack than NE does.

135 Re: Week 6 DVOA Ratings

In reply to by QQ (not verified)

"It is pretty widely acknowledged that DVOA awards more points for a long 10 play Drive of 80 yards than a 1 or 2 Play 80 Yard TD Drive."

DVOA is a per-play statistic. A 2 play 80-yard drive will have waaaay more value per play than a 10-play drive.

It may be "pretty widely acknowledged" but it's also the most common misunderstanding about the way DVOA works. It doesn't give a crap if you score in 2 plays or 10 plays. It's a per play statistic. The point is that if your plays gain 1, 2, 1, 2, 1, 80+TD yards, you're not as good an offense as one that goes 3, 4, 3, 4, 3, 4, etc. down to a TD. Why?

Because in the first case you went 3 and out and put your defense in bad field position, and in the second case, you scored a TD and put the defense in neutral field position.

Offense != points scored. Offense both scores points and prevents points because field position is fluid.

(The other point to note is that the main difference between a 50 yard TD and an 80 yard TD is that the 80 yard pass probably came on the 20 and the 50 yard TD came on the 50, which is why you have diminishing returns for play value over 40 yards.)

"but it is pretty commonly accepted that their Offensive DVOA would be higher if they could replace those 50, 70, and 84 yard TD Passes with 10 Play Drives"

No, that's not it. Their offensive DVOA would be higher if they would have a lot more of those 50, 70, and 84 yard TD passes and fewer punts.

Go look at their drive stats. New England and Green Bay score exactly the same number of points per drive, and New England and Green Bay's VOA are almost exactly the same (37.3% vs. 37.2%). New England has a few more yards/drive, but Green Bay has fewer turnovers/drive, so the rest washes out.

The reason Green Bay's VOA isn't higher isn't because they've got those long scoring plays. It's because they don't have enough of them.

52 Re: Week 6 DVOA Ratings

Aaron, you have the success value of each play, the average success value of each situation, the opponent's adjustment, and the "Red Zone" weighting. A team can get a good DVOA by having good success (V), by playing in a lot of low expectations situations (OA), by playing against a good team (D), and by concentrating their success into highly weighted situations (mildly bogus from a predictive standpoint).

Current opponent's adjustments should be nearly a wash for the Patriots and Jets. So, did the Jets have unusually high success? (Well, no. They clearly did not.) Did they play an unusually high percentage of low expectation plays? Was their red zone (and other weighted situations, if there are any) play unusually successful?

It would be interesting to see how much variation there is in the total expectation for success (the A in DVOA) from game to game. Maybe some games set a very low bar, just because of the situations that come up. Other games might have a lot of very high expectation situations, making good results look below average.

55 Re: Week 6 DVOA Ratings

I don't think you guys appear to weight VOA variance enough in your rankings and analysis. Looking at your final estimated-wins table, that is the single-biggest thing that jumps out to me regarding GB over your other highly ranked performers. Look at where that ranking list is weighted. Most of the "consistent" teams are the worst teams overall. When you are terrible on a play-by-play basis, it's easy to be consistently terrible. However, to achieve what may not be the highest overall DVOA, but to achieve a very very good one consistently with little variance appears to me to be the rarest and most prized trait of all.

In short, Green Bay appears better to me by the numbers because they achieved their score through consistent performance rather than through some boom efforts followed by some bust efforts. Look at the variance ranks of everyone in your top-ten DVOA.

63 Re: Week 6 DVOA Ratings

This point is exactly what keeps jumping out to me (although as a Ravens fan, I notice it for the exact opposite reason as that of Green Bay fans).

57 Re: Week 6 DVOA Ratings

The Patriots-Jets thing was heavily debated in the DVOA thread last week. Vince and I were going back and forth on it, along with others.

A big part of the issue may be that DVOA is not drive based, combined with the fact that the Jets performance last week was extremely bimodal.

Out of nine drives (excluding the futile end of game one, and the one that had a short field due to a special teams play) the Jets had the ball with a lot of field in front of them, they went 3-and-out seven times. But on the other two drives, they strung a lot of 8-12 yard plays together in sequence, capping each with a TD, all of which are things that DVOA loves. When you add it up, you get that the seven failed drives accounted for about half their total offensive plays, and the two successful drives accounted for the other half.

So from DVOA's standpoint, the Jets had a great deal of success about half the time. Normally, if a team has a 50% chance or so of getting 8-12 yards on a given play, that team will have few 3-and-outs, string a number of plays together for TD drives, and drive into FG range a lot of the other time. In other words, will have a pretty good offensive day. The Jets, crazily, put all their success into just two drives, and were hopeless on the other seven. So our eyes (and per drive stats) tell us that the Jets failed utterly on 78% of their possessions, which suggests a bad offense (or a dominant defense). But to DVOA, which looks per play, it sees enough success that one would EXPECT success on a much larger percentage of possessions.

I think this accounts for most of the problem. The Jets found a very weird way to exploit the fact that DVOA doesn't know about drives to make their day look much better (to DVOA) than it did to conventional wisdom.

I'm not sure how predictive it is. The fact that about half the Jets offensive plays were successful should imply that they have a pretty good offense. The fact that a huge number of their possessions end in a 3-and-out should imply that they have a pretty bad offense. I would lean towards the idea that counting on being able to string all your successes together is probably not sustainable, and so the Jets will regress in DVOA as the season goes on, but then again, I hate the Jets, so I'm probably biased.

I think introducing some kind of drive metric to DVOA will help, but it's a hefty undertaking (I had some suggestions as to how in the previous DVOA thread, but I was told that I was essentially describing FEI).

59 Re: Week 6 DVOA Ratings

I think it's likely that the Jets offense looks completely different by the end of the year. Either it will have a same level of success but look much more normal or get a lot worse.

92 Re: Week 6 DVOA Ratings

This is probably correct. We're dealing with small sample sizes and the Jets will either smooth out their distribution of successful plays and stop getting so many 3 and outs, or they'll start having fewer successful plays. Either way, the problem will go away.

I think it may also be the case that 3 and outs *seem* a lot worse than a 6 or 7 play drive with one first down, even though it might be the difference of one successful play and a handful of yards. They're obviously marginally worse but in reality not as bad as they feel psychologically when you're watching the game.

60 Re: Week 6 DVOA Ratings

The Chicago defense vs types of receivers is interesting to look at. Awful against #1 receivers, bad against TEs, and excellent against everything else. I think that is mainly because of the safety play, letting receivers (primarily #1 receivers I guess) get behind them and not closing on TEs running up the seam. So, this could be fixable.

Also, I thought the playoffs were hopeless after losing to Detroit, but the Bears are still very much in the mix for the #6 seed. Tampa is the only team not leading their division or in the North with more wins than them.

68 Re: Week 6 DVOA Ratings

As a GB fan I would like to know what has happened to their Pass Defense against Opposing RBs. Last Year GB ranked #5 against them but this year are a Brutal 30th. The only major change in Pass coverage on Defense has been the loss of Collins but it is usually not Collins who was responsible for covering RBs. I have a few possible theories:

1- Teams in the Off Season determined that they did not attack Hawk/Bishop enough in Pass Coverage to RBs

2- The Decline of GBs Pass Rush means Opposing RBs do not have to stay in for protection as often

3-Hawk and Bishop are just worse this Season

4-Schedule Randomness. GB has played against some teams good at Passing to their RBs (Forte, Sproles, Stewart)

71 Re: Week 6 DVOA Ratings

I think a few things (and I'm sure there are manyothers I don't realize).

1) They have been playing more zone to protect Tramon's injured shoulder

2) The pass rush isn't as good (you pointed this out)

3) For whatever reason, the "all the action in one direction, then role the QB out the other way to the running back / tight end leaking into the flat" play has been KILLING them early in games. They seem to be really overaggressive early and then figure it out later.

Their pass defense against RBs (and pass defense in general) looks a LOT like the 2009 season, which is certainly a bit disconcerting. I do see signs of improvement though.

70 Re: Week 6 DVOA Ratings

My thoughts in looking at the rankings.

I don't think a team that gives up as many yds/play as they gain is strong team. SF gains 5.2 and gives up 5.2.

But given the fact that they have almost a guaranteed playoff spot due to their inept division - they probably have a pretty decent chance to do well in the playoffs because they have a good shot to do fairly well - particularly when it looks like there aren't any dominant teams this year.

75 Re: Week 6 DVOA Ratings

No offense to Aaron or Football Outsiders, I've always loved this site...but this year's rankings are really bad and show some glaring flaws. Does this mean the system needs tweaking next year?

I agree with the poster who said...
"I think this accounts for most of the problem. The Jets found a very weird way to exploit the fact that DVOA doesn't know about drives to make their day look much better (to DVOA) than it did to conventional wisdom."

I think this needs to be looked into for next year. Can anyone realistically sit here and say the Jets are the 4th best team in the league? They beat the Fins, ranked 28, and move ahead of Buffalo and NE in DVOA and the entire league except for Baltimore in DAVE? Pretty much indefensible and makes the system look pretty bad imo.

Not hating, just confused ... still love FO.

76 Re: Week 6 DVOA Ratings

In reply to by Bobby Wommack (not verified)

On top of that, it was pretty obvious to the naked eye that Miami would have been ahead 13-0 or 17-0 before NY even had a first down with a competent QB.

They didn't play well at all, but had the good fortune of running into a teal hellbent on playing worse.

84 Re: Week 6 DVOA Ratings

"They didn't play well at all, but had the good fortune of running into a teal hellbent on playing worse."

The Dolphins wear aqua, not teal.

79 Re: Week 6 DVOA Ratings

" they probably have a pretty decent chance to do well in the playoffs because they have a good shot to do fairly well"

Solid analysis. /JK

Nice analysis on yds/play though, funny a team can end up #2nd with that.

93 Re: Week 6 DVOA Ratings

Yes that sentence was not what I was trying to say...I was typing in a hurry and my brain and fingers didn't sink up very well.

What I meant to say was the mere fact they will get into the playoffs gives them a chance to do well even if they are an average team.

85 Re: Week 6 DVOA Ratings

Once again FB outsiders are on the wrong side. They have a team ranked 2nd best despite the fact that they have gotten outgained in yards per play this season. They continue to over value turnovers in their forumla despite telling people that turnovers are much to do about luck. No wonder the FB outsiders guys are always scratching their head with questions they can't answer. No wonder they never get it right. Kind of sad.

90 Re: Week 6 DVOA Ratings

You missed the " is way better than this" part of the complaint format.

Also consider that being outgained in yards per play means nothing if you don't score more points then your opponent. Additionally it seems clear that SF is ranked so high based on having the #1 special teams, #3 defense and an average offense. When you get your field position via special teams and defense you don't need much of an yards per play average.

98 Re: Week 6 DVOA Ratings

If you look at the breakdown of the SF DVOA score, you may see the culprit. It's in the far right-hand column, the special teams DVOA. Indeed, the possibly-overrated Jets similarly have a very high special teams DVOA. (This was also the case last week for Minnesota when they were rated #11 at 1-4, but their special teams apparently had a bad game against the Bears.)

I think DVOA is more useful for comparing the various units (offense, defense, and special teams) to each other than it is for comparing the overall sum of the three. I don't think FO has really found the right balance for the relative importance. Consider this: DVOA thinks that the 49er special teams are more valuable towards their team's performance than the Eagles' offense is. Or the Lions' defense.

100 Re: Week 6 DVOA Ratings

The way I understand it, FO gives offense and defense each three times the weight of special teams. This is kind of cool because (as someone pointed out, I forget who), it's like offense and defense are the touchdown, and special teams is the extra point.

If you multiply the special teams DVOA by 3, you get a range that's about the same as offense or defense. It doesn't seem that strange to me that SF's special teams are as good as GB's offense or the Raven's defense. I guess the point of contention is whether special teams are really worth 1/7th of a team's worth. I recall reading Aaron's justification for this somewhere. I'm pretty sure he experimented with different weights to give the best correlation with wins.

Actually, when DVOA was young, I seem to recall that special teams gave Aaron as much trouble as offense and defense put together. I'm glad he took the trouble...Brian Burke's model, for instance, is clear and predictive and transparent, and completely ignores special teams.

108 Re: Week 6 DVOA Ratings

Aye, and after watching every 49ers game this season the special teams does seem to be a major factor. They've consistently had a short field with Andy Lee punting and Ted Ginn returning, and then the defense either forces a quick possession or holds the other team to a field goal. Given that SF has had the best Los/Dr in the league so far it's not surprising they've been so effective with their average at best offense.

118 Re: Week 6 DVOA Ratings

Ironically, it's possible for SF to improve on offense and not improve that much in DVOA: a better offense would make Andy Lee less valuable to them since he would generally be punting with a shorter field.

104 Re: Week 6 DVOA Ratings

The Vikes' ST DVOA got Hestered. A kickoff return TD, a 27 yard punt return, and two bad punts OOB trying to avoid Hester.

87 Re: Week 6 DVOA Ratings

I'm very happy to see the Philly defense jumping from solidly last place (a indisputable #32 ranking) in Week 4 to #26 last week to #18 this week. It matches up with what my eyes are seeing: in the second half of the Buffalo game, they suddenly got their act together, dropped the 9 gap, had Asomugha play man-press 95% of the time, put Rolle & Coleman on the field, Jamar Chaney made a leap to "above average" at MLB (which is an improvement above Casey Matthews "possibly the worst starting MLB in the history of football" that can't be overstated), the DB's starting holding on to picks instead of dropping them, the defensive line stepped up at run d...

The questions are now: can they sustain this level of play? have they peaked? can they improve further. Jumping another 8 slots next week would make them a Top 10 defense. (And they'll get Cole back at some point, too!)

103 Re: Week 6 DVOA Ratings

I agree with your positive vibe but I guess it was bound to happen. With all the new pieces I thought it would take a while for the Eagles to get going. My personal target was 3-3 at the bye week and it could have been that with a bit of luck. All the hype and unrealistic expectations of a 'Dream Team' wouldn't have helped. Now with that bubble burst they can get on with some real improvements. I think there is a strong finish in store.

110 Re: Week 6 DVOA Ratings

Yes, Philly does seem to be improving. I find it amusing that NE has seen similar real life improvement defensively the past two games yet has actually worsened according to DVOA by 3%.

As I said before, this year does seem to have a number of "does not pass the eyeball test" results. More than I recall in prior years, anyway.

111 Re: Week 6 DVOA Ratings

Wait, but my entire ppint is that DVOA is matching up perfectly with what my eyeballs said was happening on the field! Anyhoo, I've said a couple times that I think the Eagles were hurt by the lock-out more than any team. Rolle would've won the starting LB spot if he just had a few more weeks to learn the playbook - instead they were forced to put the more "pro ready" Matthews in at MLB. Also, Castillo finding his footing and the DB's learning to work together would've been much better worked out - it was essentially 5 players playing together who had no experience playing alongside each other. The safety situation would've worked itself out: Coleman and Allen should clearly be the starters, Allen would've had time to get back to game speed and regain a little confidence. I'm not sure the 9 gap would've ever been anything other than a disaster, but with the other improvements, they probably could've held onto significant 4th quarter leads in the SF and Atlanta games.

That all acknowledged, I still think they'll struggle as they find their way and I'm not 100% convinced Castillo can cut the mustard. This will be a 9-7 team that embarrasses a few "contenders" late in the season...

101 Re: Week 6 DVOA Ratings

Jake Ballard is the 3rd ranked TE in the league in DYAR, trailing just Jimmy Graham and Rob Gronkowski and ahead of JerMichael Finley, Tony Gonzalez, and Jason Witten. He's number one in DVOA thanks to a 88% catch rate.

This does not compute... this does not compute.

120 Re: Week 6 DVOA Ratings

I don't understand how the jets defense is ranked at number 2 when in all three of their losses the opposing offense has scored 30 or more points. To make matters worse thier three wins were to sub 500 teams which, by the way they are a 500 team at 3-3,

Go figure FO

124 Re: Week 6 DVOA Ratings

The Jets D has defended 79 drives. The Patriots, as a comparison, have only faced 64. That's 23% more chances for their opponents to score. Bad offenses lead to fewer points scored AND more points scored against. Oh, and no team has given up more return (fumble, interception) TDs than the Jets. That's not the defense's fault.

Their defensive drive stats are really quite good. Only Baltimore is better in both points/drive and TOs/drive. This despite having mediocre field position to defend.

128 Re: Week 6 DVOA Ratings

still doesn't make sense. if their D has defend 79 drives that means the D has left the opposing offense on field getting first downs, maybe that explains the points in their losses, as compared to the other "Patriots" if they only defended 64 times that means the D did their job and got the opposing O off the field quicker.

129 Re: Week 6 DVOA Ratings

Actually, you have it backwards.

When a D forces 3 and outs it will usually face more drives, since less time will be taken off the clock for each opponent possession. It's bad defenses that face fewer drives, because they can't get the opposing offense off the field.

An offense that goes 3 and out will also make it's defense play more total drives. When you have a below average O and an above average D like the Jets, that will cause the D to face a maximum number of drives.

126 Re: Week 6 DVOA Ratings

Wow, just noticed how strong offense (particularly pass offense) is this year.

20 teams have a positive Offense DVOA. More strikingly, 26 teams have a positive Passing Offense DVOA. Is this the most since DVOA has been calculated? (as a reference, for the complete 2010, 23 teams had a positive Passing Offense DVOA) Perhaps it will be tough for this many teams to remain "better than average" through the rest of the season?

137 Re: Week 6 DVOA Ratings

In reply to by Ryan (not verified)

Passing offense usually declines over the season as the weather gets worse (and as people get injured and hope that 31-year old QBs with injury problems can come in and suddenly be good in the middle of the season, or other teams decide hey, it's time to start that rookie, oh my God he's awful).