Writers of Pro Football Prospectus 2008

Most Recent FO Features


» 2018 Free Agency Cost-Benefit Analysis

Is Kirk Cousins the best free-agent quarterback in recent memory? Should Trumaine Johnson or Malcolm Butler have gotten the larger contract? And what makes a free-agent contract good or bad, anyway?

11 Oct 2011

Week 5 DVOA Ratings

by Aaron Schatz

This was a week with some big movers in the Football Outsiders DVOA ratings, but not in the number one spot. Buffalo still rules the roost after a victory over Philadelphia. Below the Bills, however, we have teams shooting way up (San Francisco, Pittsburgh, Minnesota) and dropping way down (Tennessee, Tampa Bay, Houston).

The playoff odds report this week really shows the difference between the two conferences this year. The AFC has quality spread out among many teams. Nine teams have winning records, and one of the teams with a losing record (the Jets) still ranks sixth in DVOA. That's a lot of teams in contention for playoff spots. Eight different teams are listed with a greater than 50 percent chance of making the playoffs. Meanwhile, the four teams with the strongest chance to make the playoffs are now in the NFC: San Francisco, Green Bay, Detroit, and New Orleans.

I'm guessing the most controversial ratings here will be San Francisco (2), the Jets (6), Minnesota (11), and Detroit (13). So let's get into some discussion of those four teams.

San Francisco: The DVOA system loves big, dominating wins, and the 49ers certainly had one this week. That stomping of Tampa Bay lifts them all the way up from 11th to second. That being said, San Francisco had a pretty good rating even before this win. The 49ers are not going to stumble into the playoffs as an 8-8 champion in a bad division. Based on current opponent adjustments, the 49ers are one of only two teams with positive DVOA in all five games this season (Buffalo being the other; Green Bay's DVOA in Week 3 was -0.8%). They are one overtime loss away from being undefeated. The 49ers are winning big with defense and special teams, ranking second in both, and have been shockingly efficient in the passing game this year, ranking seventh in pass offense DVOA. No, I don't think the 49ers are truly better than the Packers, Patriots, or Steelers (and those teams are all higher in DAVE) but we need to be taking them seriously as more than just first-round playoff bait. On the other hand, their DVOA will probably go down as the opponent adjustments get stronger, as they rank 26th in schedule so far and dead last in schedule remaining.

New York Jets: The Jets had one big dominating win (Week 2 against Jacksonville) and actually don't score that badly the last three weeks. This week, for example, they were excellent in the red zone. They stopped the Patriots from scoring touchdowns on three of six opportunities, including an interception. Since red zone plays have a bonus in DVOA, that helps make up for giving up yardage on the rest of the field. The Jets also rank first in special teams so far. However, the specifics of when they've struggled mean they rank much lower in "estimated wins" than in straight DVOA, just 17th. The Jets are dead last in the league in offense in the second half of close games, and even their defense, otherwise excellent, is ranked 19th in the second half of close games.

Minnesota: The Vikings have one huge win and four close losses, all by a touchdown or less. They would actually be higher if not for schedule strength, where they rank 29th. One "hidden" element that's killing the Vikings is "hidden special teams," which have been worth -15.2 points according to our estimates. Vikings opponents have yet to miss a field goal, and have hit four of 50 yards or more. Punts against them have also gone extra long.

Detroit: Obviously the Lions blew the doors off the Chiefs back in Week 2, but otherwise they are leading a bit of a charmed life. Like the Vikings, they've had an easy schedule so far. Unlike the Vikings, they've had great fumble recovery luck, picking up all of their own fumbles as well as half the fumbles of their opponents. This week, their game with Chicago was the rare game where both teams ended up with negative DVOA. Chicago gets penalized for all those false starts, for example, but Detroit doesn't get a bonus because our research has never shown that false starts have anything to do with the specific defense on the other side of the field.

One thing to watch for in the future: Detroit right now is dead last in first-quarter offense, one of the elements in "estimated wins." Scoring in the first quarter helps you dictate the pace of the whole game, and if the Lions could fix that problem, they wouldn't be stuck in so many close games right now.

Now, I don't personally think Minnesota should be favored to beat Detroit if the two were to meet tomorrow on a neutral field. But DVOA is telling us that Detroit is not playing like an undefeated team right now, and the Vikings are better than their 1-4 record, and those broad differences between conventional wisdom and DVOA are much more important than the small difference between 11th place and 13th place.

* * * * *

These are the Football Outsiders team efficiency ratings through five 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 50 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 27 percent of DAVE for teams with five games and 40 percent of DAVE for teams with four games. Because DAVE uses weighted DVOA rather than total DVOA, Week 1 results are slightly discounted. (Weighted DVOA begins to discount results four weeks later.)

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 BUF 33.4% 1 23.3% 2 4-1 34.7% 2 2.4% 13 1.1% 13
2 SF 30.3% 11 20.3% 5 4-1 7.8% 14 -10.0% 2 12.5% 2
3 GB 27.1% 4 22.8% 3 5-0 34.6% 3 7.8% 18 0.3% 18
4 NE 27.0% 3 27.6% 1 4-1 44.1% 1 19.2% 29 2.1% 10
5 BAL 23.3% 5 19.1% 6 3-1 -2.5% 23 -30.6% 1 -4.8% 30
6 NYJ 19.4% 10 19.0% 7 2-3 -3.7% 25 -9.3% 3 13.8% 1
7 NO 18.3% 8 16.2% 8 4-1 26.8% 4 10.7% 22 2.2% 9
8 TEN 16.5% 2 6.7% 14 3-2 18.0% 7 -0.8% 11 -2.3% 24
9 OAK 15.9% 12 7.8% 13 3-2 24.9% 5 9.6% 20 0.5% 15
10 PIT 15.6% 19 20.7% 4 3-2 12.6% 10 -1.0% 10 2.1% 11
11 MIN 15.2% 18 10.6% 10 1-4 7.4% 15 -1.4% 9 6.4% 4
12 NYG 14.2% 6 12.5% 9 3-2 13.2% 9 -1.9% 8 -0.8% 21
13 DET 13.1% 9 8.9% 12 5-0 6.9% 16 -6.9% 6 -0.7% 20
14 HOU 11.6% 7 10.1% 11 3-2 13.2% 8 4.7% 16 3.1% 8
15 CIN 8.2% 14 2.9% 17 3-2 -0.7% 20 -8.6% 4 0.3% 17
16 WAS 7.1% 15 -0.1% 19 3-1 -0.8% 21 -8.2% 5 -0.3% 19
17 DAL 3.8% 17 -0.4% 20 2-2 2.1% 19 -5.5% 7 -3.8% 29
18 ATL 2.9% 16 4.5% 15 2-3 6.4% 17 3.9% 14 0.4% 16
19 SD -3.5% 20 3.3% 16 4-1 8.2% 12 9.1% 19 -2.6% 25
20 TB -7.5% 13 -7.5% 23 3-2 6.4% 18 19.7% 30 5.9% 5
21 PHI -7.6% 24 0.8% 18 1-4 11.0% 11 16.9% 26 -1.7% 22
22 DEN -8.5% 25 -10.1% 25 1-4 -2.3% 22 10.2% 21 4.0% 7
23 CLE -9.0% 21 -6.1% 21 2-2 -7.6% 27 7.2% 17 5.8% 6
24 CAR -11.3% 26 -11.9% 26 1-4 18.3% 6 19.8% 31 -9.8% 32
25 CHI -13.6% 23 -8.2% 24 2-3 -9.1% 28 14.2% 23 9.8% 3
26 MIA -16.9% 27 -6.5% 22 0-4 8.1% 13 22.3% 32 -2.6% 26
27 SEA -17.0% 28 -18.3% 28 2-3 -10.2% 29 4.1% 15 -2.8% 27
28 KC -20.0% 30 -17.5% 27 2-3 -4.5% 26 16.1% 25 0.6% 14
29 IND -24.2% 29 -20.7% 29 0-5 -2.8% 24 14.5% 24 -6.9% 31
30 ARI -26.3% 22 -23.7% 31 1-4 -10.5% 30 17.3% 27 1.5% 12
31 JAC -32.1% 31 -23.3% 30 1-4 -28.9% 32 0.1% 12 -3.1% 28
32 STL -40.8% 32 -29.9% 32 0-4 -20.1% 31 18.7% 28 -2.1% 23
  • 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 BUF 33.4% 4-1 35.7% 3.7 3 4.7% 12 4.3% 9 6.9% 15
2 SF 30.3% 4-1 34.7% 3.6 4 -4.0% 26 -8.6% 32 11.1% 20
3 GB 27.1% 5-0 31.4% 4.2 1 -2.4% 23 -1.2% 23 3.4% 7
4 NE 27.0% 4-1 28.7% 3.7 2 5.8% 11 1.5% 16 16.1% 26
5 BAL 23.3% 3-1 33.2% 2.8 15 2.7% 16 -1.7% 24 39.5% 32
6 NYJ 19.4% 2-3 16.3% 2.7 17 7.6% 9 3.8% 11 15.9% 25
7 NO 18.3% 4-1 17.5% 3.4 5 -3.6% 24 -2.7% 27 5.1% 12
8 TEN 16.5% 3-2 15.7% 3.2 7 -2.1% 22 -1.9% 25 19.2% 28
9 OAK 15.9% 3-2 15.1% 3.3 6 16.6% 1 -3.3% 28 4.5% 10
10 PIT 15.6% 3-2 8.8% 3.1 8 -1.3% 20 -3.7% 29 33.5% 31
11 MIN 15.2% 1-4 22.5% 3.1 9 -4.8% 29 5.9% 5 16.6% 27
12 NYG 14.2% 3-2 18.0% 3.0 12 -16.9% 32 14.8% 1 12.8% 21
13 DET 13.1% 5-0 25.6% 3.0 10 -4.4% 28 8.8% 3 7.9% 17
14 HOU 11.6% 3-2 15.4% 3.0 13 1.7% 17 -5.9% 30 6.7% 14
15 CIN 8.2% 3-2 17.9% 3.0 11 2.8% 15 1.2% 18 1.3% 4
16 WAS 7.1% 3-1 10.0% 2.9 14 -12.3% 31 4.4% 8 4.1% 9
17 DAL 3.8% 2-2 2.7% 2.8 16 14.2% 2 -2.3% 26 1.2% 3
18 ATL 2.9% 2-3 0.1% 2.5 18 -3.7% 25 0.6% 19 0.1% 1
19 SD -3.5% 4-1 4.1% 2.2 22 -0.6% 19 5.3% 7 9.2% 18
20 TB -7.5% 3-2 -6.2% 2.3 20 12.3% 4 2.8% 14 21.4% 29
21 PHI -7.6% 1-4 -10.1% 2.0 26 8.0% 8 0.5% 20 6.0% 13
22 DEN -8.5% 1-4 -9.5% 2.4 19 12.9% 3 3.1% 12 3.7% 8
23 CLE -9.0% 2-2 0.2% 2.1 23 -4.1% 27 -0.2% 21 0.2% 2
24 CAR -11.3% 1-4 -9.6% 2.1 24 -5.3% 30 4.2% 10 4.9% 11
25 CHI -13.6% 2-3 -10.7% 1.8 28 10.1% 6 2.1% 15 7.7% 16
26 MIA -16.9% 0-4 -19.3% 2.1 25 6.5% 10 11.5% 2 1.9% 6
27 SEA -17.0% 2-3 -19.8% 2.2 21 4.5% 13 -6.0% 31 14.1% 22
28 KC -20.0% 2-3 -26.7% 1.8 27 3.8% 14 6.4% 4 25.6% 30
29 IND -24.2% 0-5 -25.9% 1.6 29 -1.8% 21 2.8% 13 14.4% 23
30 ARI -26.3% 1-4 -21.5% 1.5 30 1.7% 18 -0.3% 22 10.8% 19
31 JAC -32.1% 1-4 -31.4% 1.3 31 10.2% 5 1.2% 17 14.8% 24
32 STL -40.8% 0-4 -47.4% 1.1 32 9.3% 7 5.4% 6 1.3% 5

Posted by: Aaron Schatz on 11 Oct 2011

229 comments, Last at 13 Oct 2011, 10:50am by bravehoptoad


by Keith(1) (not verified) :: Tue, 10/11/2011 - 3:39pm

These rankings are terrible. If you took the Carolina offense, the Dallas Defense, and the Chicago special teams, you would have the best team in football. Therefore, the Bills do not deserve to be first place. They could not beat the Football Conglomerate.

by Guido Merkens :: Tue, 10/11/2011 - 4:13pm


by Rupps (not verified) :: Tue, 10/11/2011 - 5:22pm


by MPavao (not verified) :: Tue, 10/11/2011 - 10:15pm

Why would the team with the 6th best offense, 20th best defense, and 3rd best special teams be the best team in the NFL?

If you want the best team then a combination of the Patriots offense, Ravens defense, and Jets special teams is the way to go

by Key19 :: Wed, 10/12/2011 - 1:51am

First off, Dallas is actually 7th in defense.

And the poster's point, from what I can tell, is identifying 3 average-or-worse teams that, if you combined the best parts together, would form a better team than the highest-ranked team, which he feels is ludicrous (LUDA!). I haven't done the math to confirm that such a combination would actually have a better Total DVOA than the highest ranked team, but I'll take the poster's word for it.

by spenczar :: Wed, 10/12/2011 - 3:22am

Did the math, and it's true: That team's total DVOA is 33.6%.

by thermhere :: Tue, 10/11/2011 - 3:41pm

Dallas is clearly ranked too low because America's Team. The AIKMAN ratings is way better than this. HOW BOUT THEM COWBOYS?

by Jim is rad (not verified) :: Tue, 10/11/2011 - 3:44pm

Denver is clearly ranked too low because obviously DVOA knows nothing about winning and winning is something you can't teach and Tebow has now decided to grace us with his winning because he is a winner that just wins. Knowing a winner is way better than this. LOLZ Tebow will circumcise the league with his winning ability to win because he is a winner.

Or so I have been told

by Philly Phan (not verified) :: Tue, 10/11/2011 - 4:42pm

"Circumcise the league" - i just spit up my Dr. Pepper.

by rageon :: Tue, 10/11/2011 - 4:54pm

Best Tebow-related comment since the Ryan Clady "footprints" comment.

by Karma Coma :: Tue, 10/11/2011 - 5:10pm

I can't drink the Tebow Kool-Aid because his college & pro uniforms remind me of Tang.

by Will Allen :: Tue, 10/11/2011 - 5:18pm

When the Broncos' trainers pass out the cups of Gatorade, do they chant, "Blood of Tebow, shed for you."?

by MilkmanDanimal :: Tue, 10/11/2011 - 5:23pm

I just assumed they passed water, and he transubstantiated it into Gatorade.

by commissionerleaf :: Tue, 10/11/2011 - 6:11pm

No, Tebow passed water, and lo, it WAS Gatorade.

The over/under on the Brady Quinn experience is...

by Bill Walsh's Holy Ghost (not verified) :: Wed, 10/12/2011 - 2:26am

How come I can't get no Tang around here?

by JPS (not verified) :: Thu, 10/13/2011 - 12:06am

There are no words to describe the interminable hysterical laughter that filled our apartment after I read these words. I must capture and immortalize them.

by matu_72 :: Tue, 10/11/2011 - 3:45pm

I just have one small complaint/request with the playoff odds report under the "Special Super Bowl Matchups". Why no odds for a Harbaugh Bowl (SF-BAL)?

by raorao (not verified) :: Tue, 10/11/2011 - 3:48pm

This. The Harbowl will not be denied.

by jimbohead :: Tue, 10/11/2011 - 3:52pm


by Aaron Schatz :: Tue, 10/11/2011 - 4:49pm

Will be added next week.

by Mavyrk (not verified) :: Tue, 10/11/2011 - 5:33pm

How about the odds on the Rust Belt Bowl (DET-BUF)? Would be an interesting thing to keep track of; a lot of people seem to be amused with the idea.

by Aaron Brooks Go... :: Wed, 10/12/2011 - 1:36pm

Wouldn't that be the Ralph Wilson Bowl?

Ford bought out Wilson to take ownership of the Lions.

by matu_72_reply (not verified) :: Tue, 10/11/2011 - 3:59pm

Maybe the Belated Chris Berman Bowl (SF-BUF) seeing both teams are currently ranked #1 and #2.

by sj_niner (not verified) :: Tue, 10/11/2011 - 4:33pm

Odds for SF-BAL Superbowl is 24.5% * 16.1%: 3.95% (odds of each team winning the conf.)

Odds for SF-BUF Superbowl is 24.5% * 16.4%: 4.01%

Go Niners!!!

by leviramsey (not verified) :: Tue, 10/11/2011 - 6:05pm

Not necessarily, sj_niner: Ravens winning AFC and 49ers winning the NFC are not totally independent of each other, especially since they play each other. 49ers beating Ravens would, ceteris paribus, tend to improve the 49ers seed (giving them a theoretically better chance of winning the NFC) and tend to hurt the Ravens seed (giving them a theoretically worse chance of winning the AFC), and vice versa if the Ravens beat the 49ers; there's almost surely a slight negative correlation between Ravens winning AFC and 49ers winning NFC as a consequence (so given scenarios where the Ravens win the AFC, the 49ers win the NFC less than 24.5% of the time and given scenarios where the 49ers win the NFC, the Ravens win the AFC less than 16.1% of the time). Does this negative correlation dramatically effects the probability (say, by at least a 0.1% difference from when they're independent), though? Probably not.

by RichC (not verified) :: Tue, 10/11/2011 - 3:48pm

Gotta love those opponent adjustments. The Patrots defense gets a big +5%(bad) after basically curb stomping the Jets offense.

What was their defensive DVOA for that one game? +60%?

by Stats are for losers (not verified) :: Tue, 10/11/2011 - 7:10pm

The Jets's offense is 25th in DVOA, and not exactly setting the world on fire by conventional stats. Just sayin'

by RichC (not verified) :: Tue, 10/11/2011 - 10:17pm

Still. Do you honestly believe that the game where they forced 7 3-and-outs on 10 drives is the worst game the patriots defense has played this season? Because thats what DVOA is saying.

The Patriots defense has been terrible, but they were actually good this Sunday. Thats a problem.

by CaffeineMan :: Tue, 10/11/2011 - 11:15pm

Opponent adjustments have increased in weight this week, so there's that. Off the top of my head, the Pats D has had the following performances:

- OK against the Jets (25th)
- Bad against the Bills (1st), Raiders (5th!), and Chargers (12th)
- Really bad against Miami (13th)

Alas, 29th seems right to me.

by Stats are for losers (not verified) :: Wed, 10/12/2011 - 9:05am

So, we can at least agree that their Defensive DVOA shouldn't *improve* for beating the Jets, right? The non-three-and-out scoring drives were of 11 and 13 plays, with only 3 3rd downs between them. One of them was "late and close."

I think that increased opponent adjustments, as others have said, can make up the difference. Opponent adjustments still aren't at full strength, so I'd imagine it'll take another two weeks for DVOA to fully appreciate how bad the Pat's D is--unless they get another signature 4th-quarter meltdown out of Romo, you can bet they'll be dropping again.

Subjectively, I think we can agree that defense ain't the strength of this year's Patriots--if DVOA somehow ranked them last in "passes to the TE," then you'd probably have a good argument.

by milo :: Wed, 10/12/2011 - 10:51am

Well, DVOA says the Patriots defense was equal to the Titans defense this past Sunday. I'd love to hear an explanation of that.

by bravehoptoad :: Tue, 10/11/2011 - 3:56pm

The last time I saw SF ranked that high was when FO ran the 1994 DVOA numbers. Yee cats.

Surprising that the Vegas line opened by making the Lions 6-point favorites. Not that I blame them...my head thinks things like, "Hmm, SF has similar advantages with their TEs over Detroit's LBs as they had against Tampa Bay's." The rest of my body is thinking, "About time we get blown out and crushed and come away at the end of the season thinking 8-8 was a triumph and maybe we'll get 9-7 next year."

by navin :: Tue, 10/11/2011 - 4:05pm

One question: just how terrible of a head coach was Mike Singletary? Harbaugh has basically taken the same team as last year, and with some slight upgrades (Carlos Rogers, Braylon Edwards) has totally turned them around.

I knew it was bad, but this level of turnaround is just shocking.

by jimbohead :: Tue, 10/11/2011 - 4:08pm

that's been on my mind as well. That and, can you imagine what he must be thinking now? if he has any sort of self-awareness at all, that is.....

by Bobby Wommack (not verified) :: Tue, 10/11/2011 - 4:13pm

Singletary was a position coach who got promoted because of his name. He was never ready to be a head coach. He was a "motivator," not an X&O's guy.

by Vincent Verhei :: Tue, 10/11/2011 - 4:36pm

Several times last season, Mike Singletary decided that Troy Smith was a better quarterback than Alex Smith. Troy Smith is currently a member of the UFL's Omaha Nighthawks.

by Keith(1) (not verified) :: Tue, 10/11/2011 - 4:45pm

I have always wondered something, and this seems an apt place to post the question: why do teams without "franchise quarterbacks" not run multiple quarterbacks in different personnel packages more often than they do? Even the teams that employ multiple quarterback strategies do so very infrequently. Every time some backup quarterback enters the game, there is some rhetoric about the defense not being prepared; however, if the defense has a week to prepare, the backup's performance will decline.

Therefore, I wonder, why not prepare both as starting quarterbacks, and adjust to whatever the defense is doing by throwing out different packages? Alex Smith can run the normal rhythm offense against teams with weak linebackers and defensive backs, taking advantage of the inability to cover. Troy Smith can run the option offense, with all the misdirection and power formations. There could be some overlap between the two strategies that allow each of the quarterbacks to bleed, so the defense is guessing at all times.

So, if you constantly practiced this, would it be viable? Football teams run so many formations anyway. Would adding another layer, essentially making two "half-layers", be so difficult or even possible?

by RichC (not verified) :: Tue, 10/11/2011 - 5:16pm

Teams think this sort of thing will prevent their guy from becoming the "franchise," when in reality, their guy usually has no chance of ever becoming that.

Other than that, I'm not sure.

by Will Allen :: Tue, 10/11/2011 - 5:22pm

I also think the finite amount of time for reps in practice, with the rest of the first team offense, makes coaches hesitant to use multiple qbs. However, I remember Landy alternating series, between Staubach and Morton, for most of a season, I think.

by CaffeineMan :: Tue, 10/11/2011 - 11:18pm

How about the year when the Dolphins quarterback was referred to as "Woodstrockula"? Can't remember when that was.

by RickD :: Tue, 10/11/2011 - 5:22pm

Players hate this idea. QBs in particular hate this idea. The players want there to be an QB who clearly has the keys to the car, who is the #1 QB and is allowed to make mistakes along the way without a fear of getting benched. Also, the QB needs a good number of reps to build up a rhythm and rapport with his receivers.
And the QB's skill level increases the more he works with the receivers. If you had two QBs each getting half the snaps, you'd be more likely to have mediocre QBing all the time.
Unless you're the 1987-1991 49ers, it's unlikely that your backup QB is anything other than a huge dropoff from your starting QB (except in those cases where both options are mediocre).

by jimbohead :: Tue, 10/11/2011 - 5:35pm

What people said above, but also this: if you have two players who do dramatically different things, as troy and alex smiths would, rotating them signals to the other team your intention on offense. This is the same reason why teams don't like to have one RB who's good at pass blocking and receiving, and one who's good at running: the personnel on the field dictates the play, and therefore the defense keys off personnel (which is easy) rather than something harder like the level of the linemen's helmet, the first step of the FB, etc.

A better thought would be rotating based on field location, for instance if you had a QB great between the 20s, but terrible in the red zone. ARI did this a few years ago with Warner and Leinart (albeit, a little misguidedly, as Warner is Kurt Fraking Warner), and I'd actually advocate strongly for Denver doing this with Orton/Tebow.

edit: IIRC, Warner came in for 2 min drills, not RZ, but similar concept (Warner was great, but turnover prone, which is less of a big deal in 2 minute situations since turnovers are nearly equivalent to failed drives). Situational rotation rather than schematic rotation.

by Keith(1) (not verified) :: Tue, 10/11/2011 - 5:45pm


Why is there not more of this?

by Arkaein :: Tue, 10/11/2011 - 6:56pm

But lots of teams do have situational RBs. The term "Thunder and Lightning" is almost a cliche.

Sure, every team would like to have multiple do-it-all RBs, but realizing that this usually impractical most teams have some diversity in their depth charts, at least a positions where players rotate.

D-linemen are often run stopping or pass rushing specialists. WRs are often deep threats or possesion receivers. And if you don't have a RB who is good at both catching screen passes and driving the pile, then you usually want at least one guy who can do each job.

There's a balance that needs to be struck between hiding your intentions and lining up the best players for the situation. In some cases teams can make known tendencies work to their own advantage by running changeup plays that counter tendencies and catch the other team by surprise (such as dropping a run stuffer back into a zone blitz). The idea hasn't been explored much with QBs, but if your team is 1-4 I think it's an idea worth trying.

I agree that rather than name Tebow the starter in Denver, it would make sense to have Orton be the main QB and Tebow be the short yardage and goal line specialist. Although each player would exhibit tendencies in the plays they execute, it would still give more options in designing an offensive gameplan. It also gives the opportunity to go with a "hot hand", if one QB is struggling and the other is having success.

by ChicagoRaider :: Wed, 10/12/2011 - 11:31am

With QBs, you may have a situation where one QB is doomed. Last season is was perfectly clear that Jason Campbell would get killed in the Steelers game. A pocket passer was not a realistic plan. They put Gradkowski in at half time and he failed too. But what if Gradkowski had gotten the first team reps? Would it have been better? After all, Gradkowski had beaten the Steelers the previous season.

by Keith(1) (not verified) :: Tue, 10/11/2011 - 5:41pm

I am posing this in a short capsule, but the rest of it would go as such: 1) You have a full offseason with which to practice; 2) Neither quarterback is of franchise-quality, and will most assuredly never be; 3) Neither quarterback is UFL-quality, and are reasonable players as your second-string on any "normal" team.

With that, I want to avoid the idea of a "Wildcat package." The question is multi-faceted, to be sure; however, the answer really cannot just be "finite practice snaps" and "unwillingness." If it is, that is a poor answer.

To take an actual NFL team, assume the team is Tampa Bay. They have two young quarterbacks, each with relatively different skill sets. Neither is particularly amazing "right now." Instead of marking a play as a run play or pass play, they could mark it as a Josh Johnson play and a Josh Freeman play. Since their skill sets are different, the defense will have to prepare for those differences, but each quarterback would have enough in common that the playbook is not "normal" and "Wildcat."

It seems difficult, but if you are the Bucs, or even the Broncos, this seems more reasonable than trotting out an unproven option guy 100% of the time and hoping for the best. Rather, develop a scheme to maximize his strengths -- both in and out of an option package -- and develop a scheme for your non-option guy.

If both are NFL-quality players, and they can make most of the throws necessary, and the "gadget" plays are limited, there seems to be no true downside.

by leviramsey (not verified) :: Tue, 10/11/2011 - 6:11pm

The Frankfurt Galaxy were very successful for a number of years when they were deploying a "two-headed quarterback monster" (would not surprise me if that's a translation of what the German press called it), perhaps most notably in 1999 when they had Jake Delhomme and Pat Barnes. The other NFL Europe teams copied the strategy for the next few years (I suspect under some pressure from the NFL so that 12 QBs could get a reasonable number of snaps instead of 6), but none were as effective as the Galaxy that year.

by Aaron Brooks Go... :: Wed, 10/12/2011 - 1:42pm

Were it not for a poorly timed case of the flu, Troy Smith would have been a starter for the other Harbaugh.

by bravehoptoad :: Tue, 10/11/2011 - 4:11pm

Kind of a reverse Manning. Look at the Colts without Manning! Look at the 49ers without Singletary! The more the 49ers win, the worse Singletary et. al. look.

A lot of emphasis has to be put on that et. al. Singletary was never a details guy; that was left for his coordinators. Of course he was the one who hired them, but never underestimate the power of Jimmy Raye to wreck an offense.

by Passing through (not verified) :: Tue, 10/11/2011 - 4:37pm

There has actually been a lot more turnover among the starting 22:

Braylon Edwards for Morgan (both are injured now)
Carlos Rogers for Nate Clements
Ray McDonald for Aubrayo Fanklin
Navarro Bowman for Takeo Spikes
Donte Whitner for Taylor Mays/Reggie Smith
Ahmad Brooks for Manny Lawson
Adam Snyder for Chilo Rachal (recently)
Alex Smith took over permanently (after battling Troy Smith last year)
Jonathan Goodwin for David Baas
Tarell Brown for Shawntae Spencer

That's just off the top of my head ;-)

by Viliphied (not verified) :: Tue, 10/11/2011 - 5:25pm

Yes, but only 3 of those names you listed weren't on the team last year.

by Viliphied (not verified) :: Tue, 10/11/2011 - 5:27pm

Er, 4

by commissionerleaf :: Tue, 10/11/2011 - 6:27pm

Braylon Edwards for Morgan (both are injured now)

A wash, since they're both injured, but Edwards is a major upgrade.

Carlos Rogers for Nate Clements

Minor upgrade.

Ray McDonald for Aubrayo Fanklin

I would be shocked if Ray McDonald was an upgrade, but these things happen.

Navarro Bowman for Takeo Spikes

Bowman is younger, but I recall Spikes getting good reviews last year for his play in SF, so I wouldn't think this is a big upgrade.

Donte Whitner for Taylor Mays/Reggie Smith

Ding ding ding. We have a winner. Whitner is a good player, and Mays/Smith were not (although Mays is so athletic you have to assume he'll turn up somewhere).

Ahmad Brooks for Manny Lawson

Lawson was talented but never a 3-4 player. Upgrade.

Adam Snyder for Chilo Rachal (recently)
Alex Smith took over permanently (after battling Troy Smith last year)
Jonathan Goodwin for David Baas


Tarell Brown for Shawntae Spencer

Probably a slight upgrade, but Spencer wasn't terrible.

by bravehoptoad :: Tue, 10/11/2011 - 8:03pm

Ding ding ding. We have a winner. Whitner is a good player, and Mays/Smith were not (although Mays is so athletic you have to assume he'll turn up somewhere).

Whitner hasn't played so much recently. He missed the Philly game, and was limited in the Tampa game, only playing on running downs -- and SF was in its nickel defense most of the game. Also, Smith has been playing a lot, and has been pretty good.

Ahmad Brooks for Manny Lawson

Brooks was on the team last year, and played on most passing downs.

Tarell Brown for Shawntae Spencer

Spencer has been playing, and is a starting corner when healthy.

Offensive personnel changed hardly at all. Edwards and Goodwin are the most noteable additions. Goodwin is a downgrade from Baas, I think, and Edwards hasn't played much.

On defense, Rogers and Aldon Smith have been the biggest additions. Whitner, like I said, hasn't played much, and the other starters were all on the team last year. A lot of those are just playing better -- McDonald, Brooks, Goldson, Smith, Soapoaga. They're playing more often, some of them, or in somewhat different positions. None of those are particularly high draft choices; the team has just been good at developing linebackers, DL, and safeties lately.

by greybeard :: Wed, 10/12/2011 - 12:29am

McDonald, Bowman, Brooks, Snyder, Brown, and Smith were with the team last year. Singletary could have started them. Why mention them as turnover?

There are basically 3 changes from last year wrt to new players:
Goodwin for Baas: so far not a favorable change
Whitner for Mays/Lewis/Smith : whithner was injured mostly so did not happen yet
Addition of Edwards: again injured, did not help much
Rogers for Clements: definitely an upgrade in pass defense. probably a downgrade in run defense but with the front 7 SF has that is probably negligible.

by Karl Cuba :: Tue, 10/11/2011 - 4:55pm

There's been quite a bit of turnover on the defense. Only Justin Smith, Patrick Willis and Dashon Goldson are starting at the same position as last year. They have added Whitner, Rodgers and Aldon Smith. This leaves five guys where you can really see how misguided Singletary was. Why weren't Brooks, McDonald, Bowman and Brown seeing more playing time last year and why is fangio able to get so much more out of them?

In his defense, are there even any die-hard fans who thought McDonald would be playing at the level he's at right now?

I always got the impression from Singletary that he knew what he wanted the team to do but had very little idea of how to coach the players to do it. It's easy to say, "minimise turnovers" but harder to direct your players so that turnovers are reduced.

by Karl Cuba :: Tue, 10/11/2011 - 5:00pm

I just looked at the playoff odds page and scared myself half to death. Just don't look 49er fans, it could go to your head.

by bravehoptoad :: Tue, 10/11/2011 - 8:07pm

Good lord. Absurd.

Isn't it a funny season, when we're five games in and two teams have a <3% chance to miss the playoffs, and another two have <15% chance to miss the playoffs?

by Leibniz (not verified) :: Tue, 10/11/2011 - 5:10pm

Totally agree. I also feel that the 49ers have been underestimated based on their atrocious 2010 win-loss record alone. Recall that they had five close games last year (within 3 pts) and lost four of them, some to teams that eventually made the playoffs (Saints, Eagles, Falcons). That seemed to indicate that the player material was actually pretty good, especially considering last year's jurassic playcalling and cavalier attitude toward gameplanning, as you point out.

by greybeard :: Tue, 10/11/2011 - 5:30pm

You can look at just the Tampa and Seattle games and make a decision.

49ers played their opening game against Seattle both last year and this year. They were out coached by Seattle last year. Despite the lockout they were not out coached this year.

When 49ers played with Tampa last year everybody was talking about how much difference in competence there was between the coaching staff of Tampa and 49ers. This year people sure don't talk about it.

That Singletary preferred Troy Smith to Alex Smith was another give away.

To me, the level of turnaround is not shocking at all. What was shocking to me was how bad Singletary was. I wanted the team to sign Singletary after his work as an interim head coach. I was wrong and figured out I was wrong as soon as he fired Mike Martz and lost Linehan to 0-16 Lions.

by Alex (not verified) :: Tue, 10/11/2011 - 9:37pm

Similar talent, but not the same.

Carlos Roger has been a significant upgrade over Clements (very significant).

Chris Culliver in his limited time has played very well and was a big reason the Tampa passing offense was so frustrated.

Ray McDonald being allowed to start now has turned up the notch on our defensive pressure in a huge way. He's playing like a pro bowler IMO.

Bowman has been a big upgrade over Spikes. He's arguably had a better year than Willis up to this point. Both have been great.

Aldon Smith has been a nice addition to the LB/DL rotation as well. Significant upgrade in talent there.

Edwards really has been a non factor as he's missed a few games.

Bruce Miller has been better at FB than Norris as well. The special teams have also been exceptional, and a big part of that is upgrades on the roster (Grant, Jones etc.).

Adam Snyder lost a bunch of weight after being inspired by the transformation he saw out of Boone and has looked so much better than he ever has.

But - those upgrades alone don't really explain the turn around and I think Harbaugh and his entire staff deserve huge credit. I don't know how many watched the Tampa game, but there were some brilliant play design and brilliant play calls all day long. It was like night and day compared to last season. Has he managed to win keeping it vanilla as he's installed more of the offense as the season progresses? I suspect this might be the case.

I have to give a lot of credit to the Niners assistants as well. I think they might have one of the most talented group of assistants around and it shows on game day.

by Jeremy Billones :: Wed, 10/12/2011 - 9:26am

One interesting aspect of the Carlos Rogers improvement:

INTs in Washington (78g): 8
INTs in San Fran (5g): 3

If the SF coaching staff under Harbaugh has taught Rogers how to hold onto the ball and make INTs, that in and of itself is a huge upgrade.

by johonny (not verified) :: Tue, 10/11/2011 - 3:56pm

Hmm Dolphins are 26th with the 2nd hardest schedule remaining. Here comes Andrew Luck.

by The Powers That Be :: Tue, 10/11/2011 - 4:04pm

I don't know; the Rams are giving up more DVOA to their future opponents.

by MC2 :: Tue, 10/11/2011 - 6:22pm

It would be pretty shocking if the Rams drafted Luck.

by Milkman (not verified) :: Tue, 10/11/2011 - 7:59pm

Yeah, but are the Rams really going to draft Andrew Luck? They kind of have their future QB.

I guess they could sell the #1 overall pick to the highest bidder, though...

by Alex (not verified) :: Tue, 10/11/2011 - 9:41pm

Having seen a decent amount of Bradford, I'm not really convinced about his ability to be an elite guy in the NFL. Luck? I'm more than convinced.

Bradford throws so many short passes it's ridiculous. And the team surrounding him is questionable as well, which isn't going to help his development. If he takes a big step back this year and Spags gets fired (I think that would be a mistake unless he completely loses the team) it is very conceivable they would draft Luck. If Spags is still around, I suspect they go for the boat load of draft picks Luck would net them. I wouldn't underestimate the value of Bradford in a trade though either. Some team probably would roll the dice on him and trade at least a first for him even if he struggles some this year.

by zenbitz :: Tue, 10/11/2011 - 4:01pm

While the niners are probably not the 2nd best -- or even 8th best -- in the league, they sure looked like it for the last 6 quarters.

by bravehoptoad :: Tue, 10/11/2011 - 4:03pm

I'm guessing the three most controversial ratings here will be San Francisco (2), the Jets (6), Minnesota (11), and Detroit (13). So let's get into some discussion of those four teams.

Oops...that would be FOUR most controversial ratings.

by Jerry :: Tue, 10/11/2011 - 6:43pm

Among the many controversial ratings...

by Stats are for losers (not verified) :: Tue, 10/11/2011 - 8:05pm

Nobody expected that.

by Joseph :: Tue, 10/11/2011 - 4:07pm

OAK just got through the hard part of the sched, while SD's is about to get harder. Can OAK win the AFCW?

BTW, looking at future sched, the numbers say that DET will be hard-pressed to stay even with GB. They smell like a 12-4/11-5 wildcard (which is nothing to be ashamed of).

by CuseFanInSoCal :: Tue, 10/11/2011 - 5:35pm

I doubt it. SD has only lost once in September, despite playing poorly. If the Chargers follow their normal trend under Norv Turner, they will improve a lot over the course of the season. And if they don't put themselves in a big hole to start the season, I don't think anyone in the AFC West can catch them.

by Will Allen :: Tue, 10/11/2011 - 4:14pm

If DVOA is right, the Vikings will beat the Bears next Sunday in Soldiers Field by about 24-17. If my eyes are right, the Vikings lose by about 17-14.

I still think the numbers are missing something about the inadequacy of an offense with McNabb's lack of accuracy, and a bunch of guys who don't get open and who don't have much in the way of ball skills. I do think the Vikings may have been hurt more by the lockout than some other teams, and that will fade somewhat as the season wears on, especially on the offensive line. If it gets to Thanksgiving, and the reason the Vikings don't have any meaningful games to play in December is because a Vikings db made the typical drop of an int in the closing minute of the Tampa Bay game, and because they lost to the Chiefs, in the season's first month, I'm gonna be irritated. I'd rather see thm go 3-13.

by DisplacedPackerFan :: Tue, 10/11/2011 - 4:26pm

I just got done watching the Vikings - Cardinals game on NFL Rewind using the condensed mode (love that) and yeah the Vikings still looked bad in that game. Peterson had a few beast mode runs and they capitalized on turnovers from an even worse looking Cardinals offense. Maybe that was because I watched it after doing the condensed NO-CAR game (which was an awesome game) and then following it up with GB-ATL (I'm a homer I like to watch the GB games a few times). Carolina and Atlanta both looked better than Minn to me on offense in their losing efforts.

Though I say don't worry about them not having meaningful games, the two losses they are going to suffer to the Packers in weeks 7 and 10 would have mostly made the rest of their games meaningless anyway. :) Of course GB not seeing MN any more after week 10 and not seeing DET for the first time until week 12 just feels like odd scheduling to me.

by Will Allen :: Tue, 10/11/2011 - 4:41pm

I don't think 10 wins guarantees a Wild Card spot this year, which is why an 0-4 start makes the rest of the season uninteresting; you just need to be better on offense than the Vikings are to finish 10-2, even against an easy schedule, and even then you probably don't make the tournament. If they had beaten the Bucs and Chiefs, or just one of those two, October and November would have been more interesting to watch McNabb skip one four feet in front of the receiver.

by Nathan :: Wed, 10/12/2011 - 8:34pm

Holy shit, I never noticed the condensed mode until now, this is revolutionary. I'm watching the Lions / Bears game and it actually manages to completely skip over all the Jaws commentary. That's worth the price of the game rewind alone.

by TomC :: Wed, 10/12/2011 - 12:54am

If the Bears LBs and safeties play like they did last night, AP will run for 500 yards, and the Vikes will win easily. Not saying it's gonna happen, but those holes Jahvid Best ran through were just ridiculous.

by Will Allen :: Wed, 10/12/2011 - 8:31am

There ain't a Calvin Johnson on the Vikings roster, or anything in the same universe. The Bears may as well play with 5 linebackers.

by Will Allen :: Wed, 10/12/2011 - 8:31am

There ain't a Calvin Johnson on the Vikings roster, or anything in the same universe. The Bears may as well play with 5 linebackers.

by DisplacedPackerFan :: Tue, 10/11/2011 - 4:14pm

Looking forward to watching the GB remaining schedule go up after next when when St. Louis is in the rear view mirror. DVOA of opponents after that game is removed goes up to an average of 4.2 with 6 games against teams with DVOAs in the teens and the negative DVOA's belonging to San Diego, Tampa Bay, Chicago, and Kansas City. Heck looking at current DVOA, Chicago is now the worst team they have played. Though in my little spreadsheet I swapped from DAVE to DVOA this week so my trending lines on opponents can get a little off though. It's not a huge a shift, it just makes the shifts a little more dramatic in either direction (teams that were trending down are still trending down and teams that were trending up are still trending up) the big exception in Minnesota who had pretty much been a flat line, they spiked a lot.

by Stones1981 (not verified) :: Tue, 10/11/2011 - 4:29pm

We're all the penalties the main reason Detroit had a negative DVOA against Chicago? Besides the penalties I thought they played a pretty good game.

by Kal :: Tue, 10/11/2011 - 4:37pm

They actually really didn't. They had one okay drive, then two huge plays (DVOA doesn't like huge plays) and then basically did nothing else the entire game. Their defense didn't do that well against one of the worst offenses in the league either, so that hurts them too. I think this was rather fair, actually; while Detroit ended up winning by 11 it never felt dominant at any point in the game and it always seemed like Chicago was right in the mix.

by mm (not verified) :: Tue, 10/11/2011 - 10:37pm

"(DVOA doesn't like huge plays)"

While your overall explanation is right, this statement is wrong. DVOA likes huge plays, just not as much as conventional statistics do. If you have a lot of big plays with few bad plays you get a really great DVOA. There are games like this sometimes (as a Saints fan, I remember Brees destroying the Patriots on Monday Night 2 years ago).

DVOA prefers teams that consistently have good plays over "boom and bust" teams. It isn't the "boom" that hurts, it's the "bust". Lots of unsuccessful plays will pull your rating down.

A better statement would be "A few huge plays doesn't offset many bad or average plays in the eyes of DVOA"

by Aaron Brooks Go... :: Wed, 10/12/2011 - 1:56pm

The argument is also that DVOA prefers three 20-yard completions to one 60-yard completion, despite the scoreboard being disinterested.

by ASmitty :: Wed, 10/12/2011 - 2:14pm

Fun Detroit Stat of the Day:

The Detroit offense has scored six touchdowns outside of the Red Zone. Two of those plays were from their own half of the field.

The Detroit defense has allowed only one touchdown outside of the Red Zone, a 25 yard TD pass to Dez Bryant.

There's something to be said for a defense that forces you to get into the Red Zone to score TDs, and an offense that can score TDs from anywhere. When you consider this alongside Detroit's turnover margin, you can see that they're winning the big play battle as a team.

by mm (not verified) :: Wed, 10/12/2011 - 7:05pm

"DVOA prefers three 20-yard completions to one 60-yard completion"

DVOA is an average; a 60-yard completion by itself will have a higher DVOA than the three 20-yard completions (which just average out to a 20-yard completion).

The correct statement is "DVOA prefers three 20-yard completions to 2 incomplete passes with one 60-yard completion".

DVOA rewards the success of a big play, but not enough to ignore failing plays.


Let's look at 2 offenses:

Offense 1: 2/3 of its plays gain no yards. However, 1/3 of its plays score a touchdown, no matter where it is on the field.

This offense will score a lot of points. It will have lots of highlight-reel plays, with multiple 80 and 90 yard TDs in games. They'll win all the ESPYs. It will generally have several scoring drives in a row. However, there will be about 20% of drives that go 4 and out (I assume they'll go for it every 4th down). These drives produce 0 points.

Still, a fantastic offense. Scoring TDs on 80% of drives is great.

Offense 2: Never turns the ball over, Never gets a penalty. Get's precisely 5 yards on every play it runs, scoring a TD if it's less than 5 yards from the End Zone.

This is a dull offense. It will have no highlights. However, it will score every single drive except for those at the very end of each half, where time may be a factor.

This is a better offense that Offense #1. Not because 5 yard gains are better than 80 yard TDS, but because 0 yard gains are bad plays, and bad plays will lead to bad drives. This offense has no bad drives.


A team that has 1 60 yard TD with 2 plays for no yards is offense #1. If your team fails on 67% of it's plays it will have some drives that produce no points; the scoreboard is most certainly interested in that.

A team that gets 20 yard completions every play is actually slightly better than offense #2 (because it will run out of time at the end of the half less often).

DVOA likes big plays; it just doesn't put up with failing plays as much as you do.

by Jerry :: Wed, 10/12/2011 - 11:22pm

You've summarized one of the misconceptions that many of us have about DVOA. An offense that always scores a touchdown on the first play of its drive, or a defense that always causes 3-and-outs, is going to have a fantastic DVOA. For real teams, what happens on the series (or even plays) that aren't so successful is going to have a large effect on DVOA.

by Aaron Brooks Go... :: Thu, 10/13/2011 - 1:26am

Doesn't this neglect the affect of total possessions?

Team 1 scores 80% of the time on very short drives. Lets call this team the Colts.
Team 2 scores 100% of the time on extremely long drives. Lets call this team the Vikings.

As a defense-independent measure, Team 1 is more successful if it manages 25% more possessions. But this is possible if it's scoring twice as rapidly.

by nat :: Thu, 10/13/2011 - 6:47am

Teams alternate possessions. If your fast scoring gives you 25% more drives, it gives your opponents 25% more drives, too.

by Led :: Tue, 10/11/2011 - 4:31pm

Interesting minor point: the Jets are top of the rankings in kickoff coverage with one of the lowest touchback rates in the league. Folk does not have a huge leg, but I think the Jets are intentionally not going for touchbacks and instead kicking for hangtime to pin teams inside the 20. Whatever they are doing, it's working so far. They are first in KO returns as well, with a TD by yet another different returner. Mike Westhoff is such an asset. I think rule changes on average favor the better coaches because they adapt quicker to figure out more advantageous strategies. Unfortunately for the Jets, Westy reportedly is retiring after next year.

by Will Allen :: Tue, 10/11/2011 - 4:31pm

I also think the numbers are about right on the Lions, so far. A great defensive front, backed by a secondary with issues, and some playmakers on offense paired with a qb with occasional accuracy issues and a o-line with problems. I think Stafford and the dbs could improve, though, and, in particular, if the latter happens, they are going to get some short fields as the defensive linemen bludgeon people. The upside for the Lions is pretty impressive, I think.

by Verifiable (not verified) :: Tue, 10/11/2011 - 5:27pm

I agree completely. The lack of running game, last night not withstanding, is also a concern. A positive I do see is Stafford's pocket awareness, he gets rid of the ball effectively to avoid sacks, which is agood skill with the OLine in Detroit.

by LionInAZ (not verified) :: Tue, 10/11/2011 - 9:33pm

Whatsecondary issues? If you're going to criticize a team, at least be specific about it.

by Anonymous1 (not verified) :: Tue, 10/11/2011 - 4:35pm

I must admit to finding it interesting how "well" you rate the Buffalo defense. That group is like a carbon copy of NE's last year, only not quite as good at actually stopping anyone without a turnover. Yet NE last year at this time was something like 30th in the league.

I understand where you are coming from, I just don't think tip drill pick 6s are all that predictive. It takes a pretty bad unit to get their hands on 4 interceptions and still allow 500 yards, as Buffalo has done twice this year.

by Kal :: Tue, 10/11/2011 - 4:39pm

Not really. Yards are a pretty meh way of measuring defenses, especially without counting situational value. I agree that the INT rate is not indicative of actually amazing defense and that in general it would be better if INTS could somehow be credited to tip or pure catch or whatever, but at the same time they've been able to do a fair amount of control of decent to good offenses.

by Anonymous1 (not verified) :: Tue, 10/11/2011 - 4:56pm

They have? Oakland gashed them, NE abused them up and down the field except for about 4 plays and Philly also moved the ball very well despite a putrid OL.

I'm not saying they are the worst defense ever, I just don't see the ~20% discrepancy between Buffalo and what NE was trotting out there last year.

by RickD :: Tue, 10/11/2011 - 6:05pm

Well, there's the turnover aspect and there's the yardage aspect. The Bills lead the NFL in INTs with 12 - the Pats only have 7. (They are even at 3 fumble recoveries each - NFL.com doesn't track fumbles forced as far as I can tell.) The Pats are giving up the most yardage per game at 326.6, the Bills are giving up 283.4 (which is still bad).

The points/game is remarkably uniform across the AFC East. The Pats are giving up 23.8 while the Bills are giving up 24.0. The Jets (!!!) are giving up 25.0 and the Dolphins are giving up 26.0.

So by this measure, the Pats have the best defense in the AFC East. They just have a bad habit of giving up long drives, especially when they have a large lead.

Some of these things will even out over the long run. If the Bills' rating is inflated due to luck, they probably won't continue to maintain such a high rating. If the Jets' offense starts producing, perhaps the Patriots won't be penalized for having a middling effort against them.

FWIW, I think the Pats' defense should improve if Haynesworth keeps playing, and should improve more when Mayo comes back.

by Verifiable (not verified) :: Tue, 10/11/2011 - 5:29pm

Depends on who tipped them, tipped by the defense is usually a good play, by the offense is usually luck.

by Anonymous1 (not verified) :: Tue, 10/11/2011 - 5:34pm

I've seen two tipped by the offense and another two that went off of OL's helmets.

by John Doe (not verified) :: Tue, 10/11/2011 - 7:36pm

As far as I know DVOA counts all picks equally, so that is probably most of the perceived discrepancy between this years Bills and last years Patriots.

That said, if you throw the ball and hits one of your linemen in the head, you have made a poor throw. While I agree that good passes that are batted up by the receiver are not the QBs fault, I think it all comes out in the wash when you consider that dropped interceptions aren't factored into DVOA either.

by Anonymous1 (not verified) :: Tue, 10/11/2011 - 7:43pm

Agreed. I'm not saying I want them to adjust the system, I'm just making a statement that the Bills' defense isn't as "good" as these numbers suggest. Frankly, other than the interceptions they are quite terrible, and it just seems to me that many of them aren't predictive of future turnover success.

by CraigoMc (not verified) :: Tue, 10/11/2011 - 7:47pm

I got into a bit of a tiff about this issue here a few weeks ago, and you summed it up perfectly - there's no real way to objectively decide which INTs are a quarterback's "fault." FO does the exact right thing by penalizing all of them. In most cases, good and bad luck will roughly balance out, and that's better than turning an objective rating into subjective film study.

by AnonymousA (not verified) :: Tue, 10/11/2011 - 4:44pm

The top 8 teams here are:


Over at advancednflstats Brian Burke has (purely objective rankings, no adjustments):


And Sports Illustrated has (purely subjective rankings, no math):


I'm not sure which of these I like best. I think you'd have to eliminate TEN and HOU based on injuries to cornerstones, but we'll see.

by Deelron :: Tue, 10/11/2011 - 5:36pm

Seeing Dallas as the top team in any ranking gives me pause to spend the time to even consider anything from that source.

by AnonymousA (not verified) :: Tue, 10/11/2011 - 8:04pm

It looks a bit weird, but BB's stats and analysis have consistently been the best on the 'net (including an excellent debunking of FO's curse of 370 and associated "overuse" theory), so I'm inclined to give him some credit. I think the Cowboys' variance is also very high right now, which may mask their average level of play being excellent.

by t.d. :: Tue, 10/11/2011 - 8:06pm

Well, Dallas could easily be 4-0 (or, 0-4), so it isn't that crazy. They also should soon have both Bryant and Austin on the field, healthy, and if that happens, they immediately become the heavy favorite to win the East, considering that they've held their own without one of those guys every game so far

by Deelron :: Wed, 10/12/2011 - 3:29pm

That's kind of what does make it crazy. Being easily 0-4 says to me you're not the best team in the league, or even top 5.

by t.d. :: Thu, 10/13/2011 - 9:10am

well, their opponents are 13-2 against everybody else, with only the Jets having lost. At least this week their opponent is only 4-1.

by BaronFoobarstein :: Tue, 10/11/2011 - 7:31pm


by MJK :: Tue, 10/11/2011 - 4:46pm

OK, explain. Last week, New England's defense had looked terrible through four games and had a +15% DVOA. This week, they faced a fairly average offense (~-3% by DVOA). Said offense had ten offensive possessions (discounting the garbage time drive at the end when the Jets were down by 9, with no timeouts and just one minute on the clock).

On 7 of those 10 possessions, the Pats defense held them to a 3-and-out. Not held them to a couple of first downs, or a FG, but a 3-and-out. On the 8th, a bad special teams play let the Jets start at the Patriots 20. The Pats gave up a TD. The Pats also gave up two long TD drives where they looked terrible. Yet all in all, a pretty good defensive showing. Not allowing your opponents to move the ball at all on 70% of their possessions is pretty darned good.

And yet the Pats defensive DVOA DROPPED to +19%. I don't think it's a function of opponent adjustments changing...Miami, Buffalo, and San Diego all stayed roughly the same, and Oakland actually improved in Week 5 relative to Week 4. So why does this good defensive performance warrant a weakening of their defense?

Is it just a matter of the fact that a good defensive performance makes your get less chances to show it off, so for the 7 drives when they played well they only had 21 plays to show off good defense, whereas the two drives (plus the short one) where they played poorly they had 25 plays to show it off (because they kept giving up yardage in small chunks), so overall, DVOA seems to think that on slightly more than half the defensive snaps the Pats offense was bad?

This could point to an area where DVOA could be improved for defense, and may relate to why defensive DVOA doesn't correlate to wins as well as offensive DVOA. By raw counting, the Jets offense succeeded on 25 plays and failed on 21. But all those successes gave them only 21 points...essentially three successful drives. From a drive standpoint, the Jets succeeded completely on three of their drives, and failed completely on seven of their drives. And so only scored three times despite touching the ball ten times.

I have to think a little more about this...

by RichC (not verified) :: Tue, 10/11/2011 - 5:30pm

Its not just that they dropped, they dropped 4.5%, which is pretty substantial. Essentially, DVOA thinks their game was roughly +40%, which makes absolutely no sence.

DVOA doesn't think the defense had a bad game, it thinks the defense had their worst game of the season. It think they had a catastrophically bad day.

by GrandVezir :: Tue, 10/11/2011 - 5:51pm

It's simple: DVOA is per play, not per drive. Since the three-and-out drives were over and done, the missing plays from those drives didn't help the Patriots' DVOA.

by RichC (not verified) :: Tue, 10/11/2011 - 6:22pm

SO, essentially, You can stop opponents 7 times, and then give up two long drives, and you're the worst defense in the league.

Makes a lot of sense.

We should be questioning these sorts of things, not trying to explain them.

by Tom Gower :: Tue, 10/11/2011 - 11:21pm

Jets offensive drives, successful plays v. unsuccessful: 2/1, 2/1, 1/2, 0/3, 10/3, 3/0, 0/3, 0/3, 0/3, 6/5, 2/2 (+spike). Totals: 26/26, for 50%. The Patriots defense played a good third quarter, when they had the three consecutive three-and-outs with no successful plays, but on par, that's not a particularly successful day on defense but those successful plays occurred relatively often on third down.

Let me put it this way: the Titans had three quarters in the first four games where they ran more successful than unsuccessful plays on offense. [EDIT, misread my notes]: The Jets had two on Sunday.

by RichC (not verified) :: Wed, 10/12/2011 - 8:29am

Success rate isn't what I asked about Tom. You guys don't give Team-by-team listings of success rate, so there's no way to see if thats average, above average, or below average.

DVOA seems to think its catastrophically bad, in the +35% range, but there's absolutely no method transparity, so we have no idea of whats going on.

Also, success rate uses completely arbitrary endpoints. The numbers change if you use 66% on 2nd down instead of 60%. Which is a big problem when the NFL gives you numbers in 10% chunks, that are basically +/- 10%.

by Kal :: Wed, 10/12/2011 - 10:33am

FEI uses the concept of drives instead of plays and does a pretty good job of predictive value. It's something that DVOA should look into; for a defense a counting stat is putrid.

Another way to put it: You can say that Brady has a great playoff record or that the Pats have a better playoff record than a lot of teams but one reason they do have that is because they haven't been to the playoffs every year and thus didn't get a chance to play (and lose) in those playoffs. Is that 'better' than going? Clearly not, but counting stats like that downgrade that sort of value. Same thing is happening here; a 3 and out is only considered (at best) 3 unsuccessful plays, but for a defense that's as good as it can get short of forcing a turnover. Why would that be punished?

by Tom Gower :: Wed, 10/12/2011 - 1:13pm

Put it this way: success rate rewards the ability to put teams in 3rd-and-long more than it does stopping them in 3rd-and-short.

I mentioned the Titans as point of comparison, since they're the only team whose success rate I have calculated. Through the first 4 weeks, they ran 94 successful plays and 139 unsuccessful ones for an overall success rate of 40.3%. They ranked 6th in Offensive DVOA, though of course DVOA is more inclusive than mere success rate.

I frankly don't care where particular teams are ranked, but there is a non-bad reason to think the Patriots did not have a great defensive game against the Jets. You may disagree with that reason, but it's there. Personally, I haven't watched the game yet and have no opinion.

by MJK :: Wed, 10/12/2011 - 3:36pm

Put it this way: success rate rewards the ability to put teams in 3rd-and-long more than it does stopping them in 3rd-and-short.

I'm not certain that this is a good quality in a metric, because it favors one defensive philosophy over another. There are two ways to play defense...you can either be super aggressive, with blitzing, chucking the receivers close to the line, playing a lot of man coverage, crowding the box, etc., which tends to create negative offensive plays with greater frequency, change short gains into zero gains, but increases the chance of giving up big chunks of yardage.

Or you can lay back, play more zone defense, limit your blitzing, play a cover-2, and focus on tackling well and keeping short gains as short gains, and force the opposing offense to execute correctly over and over again as they move down the field. This will tend to give up a lot of short plays, but limit the chances of big plays.

In your example, the former approach will tend to put teams in 3rd and long, but increase the chance of converting it, while the latter approach will tend to yield a lot of 3rd and shorts, but counts on the fact that eventaully the other team will screw up converting one of the many 3rd and shorts and you'll end the drive.

Success rate (and defensive DVOA) will obviously favor the former approach when evaluating a defense. I'm just not sure, objectively, that an evaluation system SHOULD favor the former approach, since you're just as screwed when Calvin Johnson or Wes Welker catches an 80-yard TD pass as you would have been if you had given up ten 8-yard plays.

I'm suggesting that this could be part of the reason why defensive DVOA doesn't correlate as well to wins as offensive DVOA.

by MJK :: Wed, 10/12/2011 - 3:40pm

Another thought...success rate might be a BETTER way of evaluating a defense than looking at total success points, because what matters to a defense is not how much yardage they give up on a given play so much as can they be successful three times in a row. Which is directly tied to success probability on a given play.

For example, I think in a lot of situations it's better to give up 10 or 11 yards on 1st down than it is to give up 9, especially in the red zone. Because 9 yards gives the other team an extra two chances to hurt you. Success points obviously reward the offense more for the 11 yard gain, but the defense would just as soon get a fresh set of downs to try to stop the offense.

by MPavao (not verified) :: Wed, 10/12/2011 - 1:52pm

Not all three-and-outs are created equal. A defense can give up a 5 yard run, a 4 yard run and then an incomplete pass. That is clearly not as successful as a defense forcing two 1 yard runs and an incomplete pass. The second defense did a better job.

by Kal :: Wed, 10/12/2011 - 2:34pm

Sure. But we're not comparing all 3 and outs; we're comparing (for example) a 3 and out with two successful plays (5,4 yards then a stop) to a 10-play drive with a lot of negative plays (incomplete, incomplete, 10 yards, repeat). The latter I believe would look 'better' to DVOA as far as a defense is concerned even though the former is far more important to a game's actual outcome.

And this somewhat makes sense, too. The former shows the 3rd down luck; a 3rd and 1 is very often converted, so not doing it is highly unlikely and not repeatable. Stopping people over and over again on 1st and 2nd only to fail on long 3rd downs over and over again is also lucky, and should regress to the mean. So from a purely predictive standpoint one is much better than another at predicting future outcomes.

But at the same time, I suspect we could find ways in which a defense had more non-successes (such as a 3 and out) where that drive was actually less valuable than a long drive with repeated success on long 3rd down plays. And that's clearly wrong; when a drive fails on all three tries that should be worth significantly more than a drive where it fails only on first and second down, both on predictive and actual value.

by Kevin M (not verified) :: Tue, 10/11/2011 - 4:55pm

Knocking Singletary only makes sense if you think the talent they had last year really underachieved. I'd say they're getting unsustainable play from Alex Smith, in addition to improved offensive line play (should be expected with more experience from the 1st rounders). Heck, the biggest difference in their improved special teams play is kickoff coverage. That might have something to do with the new rules (Akers is getting touchbacks at 5 times the rate Nedney did in 2010). Their defense was actually solid last year, so the additions (Rogers, Smith) might be making them even better. Harbaugh is definitely a better coach than Singletary, but I don't think one guy was horrific or the other guy is re-inventing the wheel.

by RickD :: Tue, 10/11/2011 - 5:11pm

Knocking Singletary also makes sense if you think he made bad coaching decisions regardless of whether the team was underachieving or not. And there's more to coaching than simply the question of whether a team is underachieving or not. Harbaugh has the same players Singletary had, but they are playing better as a team not because they underachieved last year, but because they've learned how to play better this year. Alex Smith, in particular, is somebody very likely to profit from Harbaugh's influence. That doesn't mean he underachieved last year, but that he's improved.

by bravehoptoad :: Tue, 10/11/2011 - 6:21pm

I don't think you're analyzing things that well. Alex Smith has not been asked to do much this year. Sunday was his best game, and he had 19 attempts. For most of this year, the o-line has stunk.

by chemical burn :: Tue, 10/11/2011 - 5:08pm

My eyes don't deceive me: Philly played much, much better on defense this week (especially in the second half) - with opponent adjustments they jump from solidly 32nd, dead last in the league last week, to 26th this week.

The lessons: Rolle should have been the LB starter from week 1. Chaney deserves his job as starting MLB and gets better every game. Parker (game-losing penalty aside) should be the starter, not Babin. (Sorry, his sacks don't come anywhere close to making up for his abysmal run defense. Put him in on 3rd and long and forget about it.)

Philly's defense is headed in the right direction - I think more than any team in the league, they were hurt by the lockout. They really needed the off-season to put the pieces in place and get their shitty, shitty d coordinator time to figure things out.

Too bad everything has gone wrong for them in ridiculous ways every game and they couldn't even be treading water at 2-3. 1-4 is just too big of a hole. I think they'll end 9-7 or 8-8 and be the team no one wants to play in the second half of the year...

(Also, if Buffalo and SF are the best the league has to offer, there are no good teams this year. Both of them barely slipped by the worst Philly team in over a decade...)

by C-Weezy (not verified) :: Wed, 10/12/2011 - 2:09pm

It also appeared to me that in the second half, the Eagles tightened their defense line, and as a result they stopped the run better. I wonder if this indeed happened, will it continue for the rest of the year?

by chemical burn :: Wed, 10/12/2011 - 10:26pm

Yeah, you can't see from the t.v. angles, but they appear to have abandoned the 9 gap and just let PAtterson and Parker do what they do. If they go COle, Patterson, Jenkins, Parker for the rest of the year and forget the 9 gap, their defense will be fine. Seriously, the rest of NFC East should be praying that Babin (the NFl sack leader) continues to be the Eagles' starter...

by RickD :: Tue, 10/11/2011 - 5:07pm

But DVOA is telling us that Detroit is not playing like an undefeated team right now,

I'm not sure about that. They do seem to be undefeated after all. You'd have been better off using a different adjective, like "dominant."

and the Vikings are better than their 1-4 record,

Bill Parcells would disagree. "You are what you are." But you might get me to believe that they are better than a typical 1-4 team.

and those broad differences between conventional wisdom and DVOA are much more important than the small difference between 11th place and 13th place.

This reads like Step 2 of the Underpants Gnomes Business Plan. You start with some reasonable statements, and you have a logical goal you want to reach, but the middle part is a bit shaky.

Face it, guys, your system still has glitches in it.

by Will Allen :: Tue, 10/11/2011 - 5:12pm

Your last statement would seem to imply that there exists some people at FO who think it possible that a glitch-free system could be created. I doubt this is the case.

by Keith(1) (not verified) :: Tue, 10/11/2011 - 5:26pm

And semantically speaking, his correction to Aaron's second sentence essentially says the exact same thing as the original (because, derp, "than" denotes simile in either sentence). Likewise, the first change is irrelevant because it implies similarity between the ability of an undefeated and a dominant team, not necessarily restricting a relationship between the two, which both sentences are doing.

He just picks because he is "better at math than 99% of the people that read Football Outsiders," and is doing his best 13-year-old attempt at trolling. Just sort of ignore him. He is mostly nonsense.

by Keith(1) (not verified) :: Tue, 10/11/2011 - 5:27pm

Err -- his correction to Aaron's first sentence.

by Keith(1) (not verified) :: Tue, 10/11/2011 - 5:28pm

Ninja edit: "second change".

I should probably register, but meh.

by RickD :: Tue, 10/11/2011 - 6:12pm

No, "better than their 1-4 record" does not mean "better than a typical 1-4 team."

The former implies that they really should be in a different category (i.e. with more wins) while the latter implies that they are in the correct category, but are better than most teams in that category.

At least, that's how I think of the difference.

"Better than *** record" is a typical cliche that implies a team should have more wins than it does.

This may seem like semantics to you, but the difference makes sense to me.

by tuluse :: Tue, 10/11/2011 - 6:17pm

I find it humorous you are arguing the semantics of better/worse than a record, while actually using the word "glitch" incorrectly.

by RickD :: Tue, 10/11/2011 - 6:26pm

I'm fairly sure I'm using the word "glitch" correctly.

I was trying to explain my usage of language. So far nobody has addressed the main thrust of my argument. Instead of actually addressing my criticism of Aaron, I see people instinctively rallying to his defense by misconstruing what I say, inventing straw men, and your little snide comment about language.

So laugh it up. You are performing the intellectual equivalent of dotting an 'i'. I hope that made your day.

by tuluse :: Tue, 10/11/2011 - 10:00pm

A glitch would mean there is something done incorrectly causing the system to produce erroneous results. Not a systemic flaw causing a system which is working correctly to produce results that aren't what you want them to be.

by RickD :: Tue, 10/11/2011 - 10:18pm

I'm pretty sure I was arguing something was being done incorrectly to produce erroneous results.

by tuluse :: Wed, 10/12/2011 - 2:38am

So you think the people at FO have inputted something incorrectly or the computer is calculating incorrectly?

It certainly sounded like you thought the system of how DVOA works is wrong to me.

by RickD :: Wed, 10/12/2011 - 10:32am

When you work in mathematical modeling, the model itself is where the glitches are.

That is not "incorrect usage" of the word "glitch". It's fairly common.

I'm sorry if that usage doesn't agree with your understanding of the word, but, quite frankly, you're still not adding much to the conversation.

by Keith(1) (not verified) :: Wed, 10/12/2011 - 10:44am

In post 81 (your post), you said: "The former implies that they really should be in a different category (i.e. with more wins) while the latter implies that they are in the correct category, but are better than most teams in that category.

At least, that's how I think of the difference."

That second part is what matters. You are supposing that everybody is using your language, and that everybody must use your language, but you cannot accept when other people who know language as well as or better than you are not using your language. What "[you] think" and how people use the word colloquially to you does not make it wrong.

So, back to the original point: what Aaron said initially is correct, as stated, as his usage was "not incorrect usage;" the fact that you had to continue to beat a language point seriously diminishes your other points.

The conversation is not really about the data at this point, either way. It is about defining meaning in words that people are using, and determining intent. You feel you know their intent, but nobody knows your intent, so you must be right. That is well and good when you are speaking to, and only to, your children (if you have any). Otherwise, as somebody else said, you just look like a jerk without a point.

Pro tip: calm down.

by Whatev :: Wed, 10/12/2011 - 4:14am

You have previously claimed to be "a scientist" and also to be "better at math than 99% of the people on this site." So surely you must be familiar with the concept of sampling variation and hence of the difficulty of interpretation on a small sample. It may be stretching it to ask you to be familiar with clustering, as I was not aware of this when I was in physics, but it may be useful to understand this also as it is relevant here.

by RickD :: Wed, 10/12/2011 - 10:37am

This isn't a question of sampling variation. The issue at hand is not whether, in reality, Detroit is a much, much better team than Minnesota or not. The issue is what the most reasonable interpretation should be of the data we actually have.

Nor is this a question of clustering. DVOA isn't a clustering algorithm.

by Whatev :: Thu, 10/13/2011 - 9:14am

But where is the interpretation? DVOA spat back two values. You are upset because the result of the calculation of the means shows that Minnesota has a slightly higher mean assessment where the observations occur at the per-play level, based on the data we have and the algorithm we have. Your claim is that this result implies that the algorithm is faulty, but the variances are very large. In other words, the result returned by DVOA is NOT inconsistent with the idea that Detroit is a much better team than Minnesota. You couldn't conclude that with confidence from the result. The point I'm trying to make is that the most reasonable interpretation of the data we actually have is: too soon to call.

And what does being a clustering algorithm or not have to do with anything? The point is that the outcomes of the plays are not independent of each other. That means that you have less data than you'd think, and that therefore sampling variation is a bigger problem.

The thing is, if you don't agree with me that sampling variation is a relevant concern, we're just not going to see eye to eye. To my mind the most damning flaw of DVOA is that (even with a whole season of data) it often cannot separate 1st from 2nd from 3rd in any kind of very meaningful way--the samples are STILL too small. For individual players, the scores are furthermore context dependent, so that statements regarding the specific quality of players and teams just get watered down to broad categories--and mostly, those are things you could've figured out without doing statistical analysis.

by Aaron Brooks Go... :: Wed, 10/12/2011 - 2:11pm

I'm seeing a lot of ad hominem, and not a whole lot of addressing his argument.

by Jovins :: Thu, 10/13/2011 - 3:45am

DVOA is an attempt to understand the true talent level of teams in a predictive sense. It is impossible to argue that Detroit has had better final results than Minnesota for the first 5 games of the season. But DVOA isn't ranking teams based on what they have done - its a predictive measure. Minnesota doesn't fit the profile of a typical 1-4 team, nor does Detroit fit the profile of a typical 5-0 team.

Taking a deeper look at the data, Detroit's offense and special teams are what drag its overall ranking down. Personally, I feel that Detroit has a better offense than what DVOA credits, but DVOA sees what I can only assume to be an average number of successful plays. Looking at the data further, Detroit's rushing offense is 30th in the league, whereas Minnesota has the best rushing offense. However, Jahvid Best has actually been relatively successful as a runner. It is Jerome Harrison and Keiland Williams, who have combined for -38 DYAR on 33 carries. What DVOA doesn't know is that Detroit's offense at its best is far better then what it is given credit for, and that in high leverage situations Detroit's offense ought to perform far better (at least in the running game) then expected, simply by virtue of having the best personnel on the field.

by Whatev :: Thu, 10/13/2011 - 9:04am

Where's the ad hominem? The statements in quotations were his own words. If there is any respect in which I am being snide here, it's because he cites his scientific training, but I don't recall getting a response to my previous question about what type of scientist he is (relevant because it turns out that "math" is not one big thing, and different fields of science, perhaps surprisingly, use different mathematical tools).

You're free to feel that I didn't address his argument, but as can be seen above, he was able to identify my argument and respond to that, so evidently we've agreed that there is something there to discuss.

by RickD :: Tue, 10/11/2011 - 6:21pm

I was chiding Aaron for defending the anomalies of this week's DVOA ratings. His defense consisted far more of "no, really, the ratings are correct!" than of "well, let's see what our statistics are telling us now." To be blunt: I don't think Minnesota is appreciably better than their 1-4 record and it's a bit silly to trust their statistics on this issue.
Whether Aaron "thinks that a glitch-free system could be created" is only the issue to the extent that he tries to gloss over flaws as being unimportant. He's imposing his own bias on the analysis when he says "no, trust me, the Lions have not played significantly better than the Vikings."
He can either admit the result is fishy or fall back on the impossibility of creating a perfect system. When he stands by the rankings, it hardly seems fair for you to bail him out by saying "everybody knows that no perfect system is possible" when that's exactly the opposite of what Aaron is doing!

by Keith(1) (not verified) :: Tue, 10/11/2011 - 7:33pm

Of course he is defending HIS system. Why else would it exist? He created it, therefore he has a reasonable interest in it being correct. That is not that say he says the system is infallible; in fact, all of the writers here know the system is a non-perfect system, but they still, within reason, support that it is correct.

You are splitting hairs of language that need not be split, yet you say that Aaron is doing just that as well. You cannot infer his intent through simple text. But here is my intent: you are quite ridiculous with your demands that he create a system based on what you think. You should create that system and go post on RickDOutsiders.com and let us know how great that works out.

In the interest of civil discussion, they are constantly reshaping the system to match the public perception of what is correct and incorrect; however, that is done in the offseason. For now, you get analysis such as "Detroit is not playing like an undefeated team" and "Minnesota is playing better than a 1-4 team." Those sentences essentially say -- to everybody who knows any English except for you, and anybody who knows the purpose and context of this site -- "Detroit is overplaying their ability based on this non-perfect system" and "Minnesota is underplaying their ability based on this non-perfect system."

by RickD :: Tue, 10/11/2011 - 10:47pm

Aaron was presented with a situation where a 1-4 team was rated 11th in his system while a 5-0 team was rated 13th. Rather than simply say something to the effect that this was an unexpected result that would cause him to question his system, he made the argument that there really wasn't much of a difference in what the two teams had accomplished so far.

I disagreed with the latter implication. I think it's quite clear to casual observers that the Lions have played much better so far than the Vikings have. If the Vikings are doing something that is increasing their DVOA score without helping the team win games, then that is something Aaron should investigate, rather than simply continuing to defend DVOA.

I watched Detroit dismantle the Bears last night. They're playing like an undefeated team.

If you think "I am being ridiculous" to ask Aaron to provide a better justification for ranking Detroit behind Minnesota, there's not much I can say. I am not asking Aaron to "create a system based on what (I) think." I'm asking for his system to live up to the billing of doing what he says it does.

But rather than getting so meta- about things, I would prefer it if somebody, somewhere, would put a better effort into explaining why Minnesota should be ranked ahead of Detroit. That would be far more interesting to me than another comment that consists largely of personal attacks.

by Keith(1) (not verified) :: Tue, 10/11/2011 - 11:18pm

Forget where they end up as a list. Forget the 11th ranking and the 13th ranking. Minnesota has been close in every game and have had a lead in every game. They have lost flukily a number of times. Watching them play without a final result has them playing better than their record indicates. The Lions have essentially been the complete opposite in terms of fluke games. I am not sure if you were watching a replay of Lions-Chiefs, but at no point was Detroit dismantling Chicago. They won on two huge plays and some superior line play against the awful offensive line.

We could say that Minnesota should have beat at least Tampa Bay and Detroit and that Detroit should have lost to Minnesota and Dallas. Those were aberrant in any normal year, especially to happen to the same teams. Looking at it in that light, you see two "slightly above average" teams, not an undefeated team and a bottom-dweller.

Put all that together, and each of these teams are doing things the system likes. Those things correlate well with winning on a larger scale than just five games. The system does a relatively good job of identifying trends that major media publications miss. The biggest example of this has been the Eagles -- they have been "average to above average" many years based on record (10-6 and 9-6-1), and yet have fared well in DVOA, and then made it far into the playoffs. Just because things do not match up exactly does not mean the system is wrong on a larger scale; it just means the "system as created" has evaluated something as instructed. And actually dissecting the data in a logical sense makes it clear that the results as currently returned are not completely implausible.

by Aaron Brooks Go... :: Wed, 10/12/2011 - 2:19pm

Forget where they end up as a list. Forget the 11th ranking and the 13th ranking. Minnesota has been close in every game and have had a lead in every game. They have lost flukily a number of times.

Try this sentence on for size:

Detroit has been close in every game and have had a lead in every game. They have won flukily a number of times.

This statement merely confirms that Detroit and Minnesota, stat-wise, aren't terribly far apart. The argument that Minnesota would win on a neutral field is somewhat specious. They have played, and at Minnesota (traditionally a house of horrors for the Lions): Detroit won.

The eye test suggests that Detroit occasionally beats itself and is probably playing below theirt potential, whereas Minnesota has been kept in games by great ST plays and defenses forgetting to tackle Adrian Peterson. Once those get corrected at halftime, Minnesota has no other options.

by Keith(1) (not verified) :: Wed, 10/12/2011 - 4:41pm

Detroit has been "close" in some games, sure; but to say they "have had a lead in every game" because they have won every game is literally correct, but not in some very technical sense. By definition, you have to have a lead to win. However, merely saying that belies how those leads happened. They were down 20+ points multiple times; that is not a close game. That they came back in the second halves of those games and eventually won those games says a lot about the team...in the second half.

You can say all day that good teams win close games, but research has shown that luck plays a major factor in close wins (as defined by less-than-a-touchdown margins). Luck has it that Minnesota managed to lose three games after holding 17-point or greater leads. I will grant that they should have only won two of those games, as the game against San Diego never really seemed to be a winning effort from the Vikings.

To compare basis numbers:

The Vikings are converting 35% of third downs while giving up 45%. They are gaining 5.4 YPCarry while giving up 3.3 YPCarry. They are completing 57% of passes for 6.4 YPCompletion while giving up 63% for 7.4 YPCompletion. They are giving up more passing touchdowns than rushing touchdowns while scoring more rushing touchdowns than passing touchdowns.

The Lions are converting 34% of third downs while giving up 32%. They are gaining 4.0 YPCarry while giving up 4.8 YPCarry. They are completing 64% of passes for 7.8 YPCompletion while giving up 67% for 6.3 YPCompletion. They are giving up more passing touchdowns than rushing touchdowns while scoring more passing touchdowns than rushing touchdowns.

So, in other words, they are really being outplayed in the run game by the Vikings and outplaying the Vikings in the passing game. They are outplaying the Vikings in defensive third-down conversion percentage (a pretty good indicator, to be certain), but are converting at about the same rate on offense. I would say it is a push on pass defense, with the Vikings slightly outplaying the Lions in run defense.

All that together shows that the Lions would probably win in a close game against each other, as the Lions' offensive strength is the Vikings' defensive weakness, but you cannot write off the Vikings run offense against a lackluster Lions run defense, even though people have been purporting their run defense as being stout. What really happened was the Vikings collapsed and/or the Lions had an amazing comeback; choose whichever side of the argument makes you feel better.

That DVOA sees Minnesota as a "better" team is not really something that should be concerning after five weeks. But the Lions are not greatly outplaying any team, and probably are just going to end up being an above-average team at the end of the year.

by ASmitty :: Wed, 10/12/2011 - 5:25pm

How about drive stats?

Detroit's offense is fifth in the NFL in points/drive. Detroit's defense is third in the NFL in points per drive. Special teams-wise, they have converted 100% of their field goals and extra points, and they have the third best starting field position on offense and the third best best starting field position for their defense.

They are second in the NFL in terms of fewest touchdowns allowed per drive, but they are fifth in terms of most touchdowns scored per drive.

You can micro-breakdown the stats all you want, but in the end, the game is still offense as a whole versus defense as a whole, with a special teams kicker. Detroit scores touchdowns, and they don't let you score touchdowns. They haven't allowed a touchdown longer than 25 yards all season long, but have six scores over 25 yards themselves. They bend, don't break, and burn you with big plays;it's not a complicated formula for success.

But by all means, I hope people continue to doubt them. I bet on the Lions every single week, and over the last 13 games they are 12-0-1 against the spread.

by Keith(1) (not verified) :: Wed, 10/12/2011 - 6:11pm

Oh. Here I thought we were discussing whether the Vikings could realistically be near the Lions. You thought I was doubting the Lions.

When it comes down to it, yes, the Lions are playing above average football, not undefeated football. That is the crux of the argument.

Just so you know, the drive stats indicate that the Lions are not playing undefeated football. They are playing great football, but not undefeated football. There are less than a handful of teams that grade out better than they do, and the difference is not significant (good for the Lions). However, the Vikings do not grade out too far from the Lions, and are actually closer to first than they are to last, and are graded above average.

As an aside, the top team in the NFC by those stats are the Packers, but not by much, so that has to be good news for Lions fan. (The Pack are followed by San Francisco and then Lions.)

by Aaron Brooks Go... :: Thu, 10/13/2011 - 1:34am

You argued with the point that the Lions have held the lead in every game by asserting that leading at the end of the game is irrelevant. What else could you have been doing besides gainsaying the Lions?

by Eddo :: Thu, 10/13/2011 - 10:40am

I don't think he's saying that leading at the final whistle is irrelevant, just that it's not really a factor in DVOA.

I've always thought of it this way: since DVOA is per-play based, not per-sixty-minutes based, it thinks that, hey, yeah, the Lions took the lead with 20 seconds left (hypothetical example); now the Vikings get to run out their roughly-equally-successful per-play offense, and should get a chance to take the lead. DVOA doesn't care that the game ended right then, it only cares that, on a per-play basis, the two teams were basically the same.

by Will Allen :: Tue, 10/11/2011 - 9:05pm

I think you have literacy issues. Either that, or you are simply a jerk for saying that that somebody has to "face it" by admitting that their system has a glitch, when in fact, the people who write here have saId, time and again, that their metrics will NOT produce a perferct evaluation of team strength. I myself find the ranking of the Minnesota offense particularly suspect, and said so above. That isn't the same thing, however, with falsely implying that a claim of perfection has been made. To explore what the metrics may be missing is one thing, and being dishonest about what the writers of this site have claimed is another.

by RickD :: Tue, 10/11/2011 - 10:32pm

Wow, you guys all really heart Aaron. "Falsely implying a claim of perfection" is, by the way, a particularly rich argument. At no point did I do any such thing.

Now please, stop making personal attacks against me and get back to the original point. Which team has played better, 5-0 Detroit or 1-4 Minnesota?

Detroit 27 Tampa Bay 20
Detroit 48 Kansas City 3
Detroit 23 Minnesota 20 (OT)
Detroit 34 Dallas 30
Detroit 24 Chicago 13

San Diego 24 Minnesota 17
Tampa Bay 24 Minnesota 20
Detroit 23 Minnesota 20 (OT)
Kansas City 22 Minnesota 17
Minnesota 34 Arizona 10

These two teams not only have two common opponents, that Detroit beat and which beat Minnesota, they also have a head-to-head matchup, that Detroit also won. Detroit also has wins over Dallas and Chicago while Minnesota lost to San Diego but beat Arizona handily.

With all of this in mind, Aaron is defending the notion that the two teams have had roughly equal success this seaon - or rather, that the Vikings have played slightly better than Detroit has.

I'm sorry, but this is an instance where the problem is not with the commonsense interpretation of the data, but rather with whatever DVOA is doing with it (and DVOA continues to remain a black box for Football Outsiders outsiders).

It is particularly noxious for you to say that I've made a "dishonest" claim about "what the writers have claimed" when this claim on my part is entirely in your head! I've already explained what I was trying to do with my first comment.
But apparently "you have literacy issues."

Either that our you are simply a jerk.

by dbostedo :: Tue, 10/11/2011 - 11:37pm

Well whether you are ultimately correct or not, this phrase :

"Face it, guys, your system still has glitches in it."

can easily be taken to mean that you are refuting a claim (by some unknown person) that the system has no glitches in it.

It's the "face it" part. That's usually used to indicate that someone has to accept something that they were previously unaware of, or had previously claimed the opposite of. And, of course, the writers are this site are completely aware that there are issues with the system.

by Tim R :: Wed, 10/12/2011 - 4:37am

I agree that the Lions have been a better team than the Vikings thus far. However everyone getting worked up over early season DVOA ratings is a waste of time. I generally take all ratings with a pinch of salt until around week 8 when opponet adjustments are fully incorporated and there is a greater sample of games.

by Will Allen :: Wed, 10/12/2011 - 8:35am

O.K., you've made it clear; you don't know the meaning of the sentences you write.

by bravehoptoad :: Wed, 10/12/2011 - 10:59am

Detroit has won a game by 3, a game by 4, and a game by 7.

Minnesota has lost games by 3, 4, 5, and 7.

Each team has had a single blowout against a bad team.

What there is leading you to believe that Detroit is significantly outperforming Minnesota? You do know that luck plays a factor in these games, right?

It seems odd that you're using such an enormous and crude hammer to ridicule a system that works on the unit of the play. You're putting way too much emphasis on W-L.

Of course we like Aaron. Some of us have been coming here for eight years. DVOA has its problems, but we're used to more intelligent criticisms than those based on "commonsense" [sic].

by ASmitty :: Wed, 10/12/2011 - 11:35am

It's not just crude W-L, it's common opponents:

The "bad team" Detroit blew out, beat Minnesota.

Detroit beat Minnesota, at Minnesota.

Detroit beat Tampa Bay at Tampa Bay, Minnesota lost to Tampa Bay in Minnesota.

Minnesota is 1-4 with a schedule involving three home games, Detroit is 5-0 with a schedule involving two home games. Detroit leads the NFL in point differential.

I definitely see the argument for Minnesota having a standing in the rankings that is superior to their W-L record, but it's not like the argument against them being ranked higher than Detroit is a pure W-L argument. Pretty much every way in which Detroit and Minnesota's paths have crossed, Detroit has come out looking better.

by Keith(1) (not verified) :: Wed, 10/12/2011 - 11:43am

They only "come out looking better" if you only consider the final score. That is something that needs to be remembered.

by ASmitty :: Wed, 10/12/2011 - 11:56am

Really? The DET/TB game was only close if you look at the score, Detroit won that pretty handily until some late game prevent offense by TB. And Detroit was on the road, while MIN lost, at home.

The KC game was not close, at all.

The DET/MIN game was close, granted.

Detroit still leads the league in point differential. Not just W-L. They've still played 60% of their games on the road, as opposed to 60% at home like Minnesota. I'm not sure what more people want.

Personally, I think Detroit should be 4-1 right now. They blew out KC, handled TB and CHI convincingly enough, and totally fluked out against DAL. The MIN game is the only controversial one, but on balance I thought Detroit outplayed Minnesota in that game.

I do think Aaron's point in the article that Detroit is hurt because their offensive starts are slow and fast starts allow teams to dictate the game going forward is interesting. But does that really hurt Detroit? Pass rush hype aside, I think the Detroit defense wants you to run the ball, not throw it, and Detroit's offense is going to pas it no matter what.

What happens typically when you fall behind? The other team runs it and you have to throw it. It's not like Detroit has to rip up the playbook if they fall behind. If anything, they might struggle with a lead because they can't chew up clock and their corners can be beaten deep if you protect the QB.

Like I said in a different post, we'll know more about this team over the next few weeks when they're home against SF and ATL.

by Nathan :: Wed, 10/12/2011 - 4:18pm

holy shit you guys

by Dan :: Tue, 10/11/2011 - 9:46pm

Minnesota has outscored their opponents by 5 points so far (111-106), which is not typical for a 1-4 team. It's not a surprise that DVOA & Aaron think that they're better than their record.

by ASmitty :: Wed, 10/12/2011 - 10:20am

And Detroit leads the NFL in point differential.

by Anonymous1 (not verified) :: Tue, 10/11/2011 - 5:20pm

Are you saying that it is a good thing for a team to have gone into halftime with a deficit three straight weeks, two of which by 20+ points?

Or do you believe that it is impossible to have a record incongruous with your level of play? If so, why come here at all? Just list the teams from 1-32 based on their record and be done with it.

by RickD :: Tue, 10/11/2011 - 10:50pm

It's better than finishing a game four straight weeks with a deficit.

by Anonymous1 (not verified) :: Wed, 10/12/2011 - 5:46pm

Fair enough. So, again, if the final score is all the only relevent factor, why are you here?

by Aaron Brooks Go... :: Thu, 10/13/2011 - 1:37am

If your proxy for wins doesn't, in fact, predict wins, then perhaps the problem is with the proxy and not with winning.

by liberty_JAC :: Tue, 10/11/2011 - 5:28pm

There is only so much you can do to explain the relevant phenomena and predict outcomes in an open system.

No hypothesis is ever perfect, or without its glitches; especially in a system this unpredictable. There are always going to be discrepancies.

Just my thoughts.

by Jonadan :: Tue, 10/11/2011 - 9:17pm

As a Lions fan, I think if we replayed this year over and over and over most of the time we'd come out 4-1. The only team we played that I feel we're substantially better than was the Chiefs. The Vikings are also kind of a mess, but AP AD so yeah. But I think Tampa, Dallas, and Chicago are all close enough to us in the "pretty good but flawed" category that the Lions "should" have lost one of those games (probably Dallas).

Not that I'm complaining.

On the other hand, there's always a significant difference, especially early in the season, between DVOA and "actual wins". What I'm beginning to wonder is if DVOA is actually at its most predictive when extrapolating to "next season"'s probable (especially negative) regressions? Exhibit A: 2010 Chiefs -> 2011 Chiefs.

"When you absolutely don't know what to do any more, then it's time to panic." - Johann van der Wiel

by jimm (not verified) :: Wed, 10/12/2011 - 1:19pm

I've enjoyed reading the back and forth between RickD and others that resulted from the original "glitch" comment.

What I took from it:

1) There is a huge amount of luck in football. DVOA is an attempt to quantify what isn't luck and therefore predictive. That is really hard to do...but very interesting.

2) Most debates end up being about ego defence. Certainly this one seems to have gone down that path in a huge way.

3) The people in this debate all seem pretty darn intelligent to me.

I've been in such debates myself, probably with some of the same folks in this one...when I'm in the middle of one of them I think I'm trying to put forth the best info as I know it, but really it ends up being - I'm right and you're wrong and my comments will be biased to supporting that goal. That's not really much use. Too much wasted energy and a lot less learned.

Anyway - thanks for the debate - I rather enjoyed it.

by panthersnbraves :: Wed, 10/12/2011 - 3:55pm

sometimes these seem to be "religious" debate, with neither side able to convert the other.

by bravehoptoad :: Thu, 10/13/2011 - 10:50am

Can we expect the FOMBC to descend on Detroit now?

by MilkmanDanimal :: Tue, 10/11/2011 - 5:25pm

Out of curiosity, does the game charting track "Completely open WR", even more so than good ol' Hole in Zone? I swear I've seen more WRs catch easy TDs this year than ever before (easily attributable to the lockout/short offseason), and I'm curious if it's that much more pronounced this year. It seems like there's at least 2-3 cases a week where the secondary completely blows the coverage and the WR waltzes in untouched.

by chemical burn :: Tue, 10/11/2011 - 5:37pm

Been watching Philly play a lot?

by LionInAZ (not verified) :: Tue, 10/11/2011 - 9:38pm

I was thinking that when I saw the Broncos leave Greg Jennings uncovered on a straight route to the end zone.

by Aaron Brooks Go... :: Wed, 10/12/2011 - 2:29pm

That defense worked really well for the Lions last year.

by panthersnbraves :: Wed, 10/12/2011 - 11:24am

Maybe the defense figures that the WR will drop the pass - that seems to be the situations with the Panthers' #17.

by QQ (not verified) :: Tue, 10/11/2011 - 5:50pm

I still propose that DVOA does not give enough credit to Huge, Explosive Plays by Offenses. Some teams are actually good enough to be able to consistently produce them, for example GB basically averages a 50+ yard TD every game.

Perhaps the 2 most Explosive Offenses the NFL has seen, 98 Vikings and 99 Rams, only were able to produce Offensive DVOA's of 19.9% and 12.4%. It seems hard to believe that this Year's Raiders really have a better Offense than either of those 2 teams or that this year's Giants and Steelers have a better Offense than the 99 Rams

by ammek :: Wed, 10/12/2011 - 5:19am

You need to make a couple of adjustments. Firstly, offense has become more efficient since the late 1990s league-wide. Secondly, we're not even one-third of the way through the season yet.

The 1999 Rams offense has been debunked here before. I'll try to do the same for the 1998 Vikings.

Firstly, you state that it was one of the two most explosive offenses ever seen. Let's stick to pass offense here. The Vikings had 62 pass plays of more than 20 yards, and 21 of 40 or more. There were two other offenses the same season which had 60 pass plays of 20+ yards: Atlanta and San Francisco. In 2011, there are eight teams on track to match the 1998 Vikes in this category. Last season, the Chargers, Broncos and Eagles all topped 60 pass plays of 20+ yards.

Regarding pass plays of 40+ yards, it is true that the Vikings' total has only been matched twice — by the 2000 Rams (not the 1999 version) and the 2009 Eagles (who had more 20+ yard plays than Minnesota, too). But plenty of recent teams have been within two or three such plays. You want DVOA to hoist the 1998 Vikings over, say, the 2004 Eagles on the basis of two long passes — about 0.3% of pass plays and 2% of pass yardage?

And that's just the passing game. Minnesota's 53% unadjusted VOA in 1998 is in the top dozen for passing DVOA all-time — and that's despite throwing a thoroughly average 16 interceptions, and being quite lucky with recovering its own fumbles. It takes a downward adjustment in DVOA for facing some weak defenses (Wannstedt's Bears, the AFC Central, etc). But what brings the figure for total offense down is the running game.

Robert Smith was a boom-and-bust runner, never more so than in 1998. He averaged 4.8 yards per carry, yet finished with roughly 0% DVOA. The Vikings managed five runs of 40+ yards, tied for second in the league, but Smith was 25th in success rate.

Overall, the 1998 Vikings' offense (run and pass) didn't even lead the league in yards per game or yards per play. It managed just one more play of 20+ yards than the 1998 Niners, but gained 44 fewer first downs and 500 fewer yards. Yet despite DVOA's alleged 'bias' against big plays, Minnesota's unadjusted offensive VOA was still a little higher than San Francisco's.

by Aaron Brooks Go... :: Wed, 10/12/2011 - 2:39pm

Robert Smith was a boom-and-bust runner, never more so than in 1998. He averaged 4.8 yards per carry, yet finished with roughly 0% DVOA. The Vikings managed five runs of 40+ yards, tied for second in the league, but Smith was 25th in success rate.

Overall, the 1998 Vikings' offense (run and pass) didn't even lead the league in yards per game or yards per play.

These seem like mutually inconsistent positions.

Smith was bad because despite leading the league in yards per play, he was inconsistent.

The Vikings offense was bad because despite being consistent, it didn't lead the league in yards per play.

You can't have it both ways. I think the criticism is that DVOA discounts offensive plays that result in touchdowns in favor of consistent offensive yardage production, whereas the opposite is true for defenses. It doesn't much care about yards but strongly rewards red zone success -- it values low points allowed over yards allowed.

by madvillain (not verified) :: Tue, 10/11/2011 - 6:19pm

can someone explain to me the cliff's notes version of why DVOA doesn't like explosive plays on offense. Thanks. My guess is it doesn't think they are repeatable?

Dunno, In general DIPS theory was widely accepted "as is" in 2004, and is now widely recognized as a flawed but interesting theory that still has lots of work to be done on it.

Somethings that seem well fleshed out in sabermerics take decades to fully refine and things that are taken as universal truths in one generation are often overturned in the next.

by Arkaein :: Tue, 10/11/2011 - 7:05pm

It's a misnomer to say DVOA doesn't like big plays. It likes them best of all, it just doesn't see a 60 yard play as being twice as impressive as a 30 yard play, or four times as impressive as a 15 yard play.

What DVOA prefers is large numbers of good plays to a few great plays mixed in with a lot of bad or mediocre ones. Lots of good plays is more likely to indicate sustainable quality, while a small number of great plays is more likely to be a fluke.

Think of a 10 yard completion that goes the distance because a safety whiffs on a tackle. The receiver deserves credit for the completion with a bonus for breaking a tackle and extending the play, but whether he gains another 10 yards or another 50 depends as much on the position on the field and the specific alignment of the defenders on that play as it does on his own execution.

When teams or players can consistently produce big plays, and mix those with several more solid plays, while avoiding turnovers, they tend to produce the best DVOAs possible.

by CraigoMc (not verified) :: Tue, 10/11/2011 - 7:43pm

I think the basic concept behind DVOA - measuring plays within context of down, distance, and position - is fundamentally sound, but you have to wonder whether it's being executed properly. Sure, a 50-yard bomb isn't as predictive as 5 ten yard slants, and probably isn't worth five times as much as any one of those plays.

But I'm beginning to think that DVOA really is underestimating the value of explosive plays. Without knowing exactly how much more a 50 yard gain from the 20 on 1st and 10 is compared to a 10 yard gain in the same situation, I can't say for sure. But FO's own research shows that there is a substantial difference in scoring expectancy from 70 yards out and 30 yards out, that probably isn't being properly credited by the system. Yes, first downs are incredibly important, but the baseball analogy doesn't really hold up - outs are far, far more important. I think DVOA has tunnel vision in that it's focusing too much on whether a play makes a team more likely to get a first down, as opposed to more likely to score.

by MJK :: Tue, 10/11/2011 - 10:15pm

Thinking about New England's drop this week in defensive DVOA has me thinking that it's even simpler than that. Let's say you get about 50 offensive plays in a game.

If you have one big play of 80 yards, you get credit for it...but it's still only one play. So one play out of, say, 50 you might have in the game was really good.

If you have ten 8-yard plays, those are also pretty good. For the reasons above, they are better than 1/10 as good as the 80 yard play (it's fair to say so, because the 80-yard play could just as easily have been a 40 yard play had the LOS been the opposing 40 instead of your own 20). However, now 10 plays out of 50 were pretty good.

Since DVOA is averaged over a lot of plays, a string of decent plays will end up giving you more of a boost than one massive play.

I wonder if there's some way to correct for context of a play to a drive... I.e. an 80-yard bomb isn't valued as just one play, but rather as a play that REPLACED a bunch of shorter plays that the team didn't have to execute to get the same result. Conversley, a 3-and-out forced by the defense isn't just 3 good defensive plays, but counts something for all the other defensive plays the defense didn't HAVE to make to stop the drive.

This is still congealing in my head, so I know it's a crappy idea right now. But I'm trying to figure out if there's a way to take drive context into account, the same way DVOA currently takes first down context into account.

by MJK :: Tue, 10/11/2011 - 10:28pm

I just had an idea...

Currently, you get "success points" based on how successful a play is towards getting you closer to (1) getting the next first down, and (2) scoring. That's compared to the league average for the given situation you're in, and then adjusted for defense and non-predictive events to get DVOA. The problem is, I suspect, that it is possible to get more "success points" by stringing together ten 8-yard plays than throwing one 80-yard bomb. Or it might be possible for the DEFENSE to get more negative success points by playing pretty well on a lot of snaps, but occasionally giving up a first down conversion and having a long drive that still never produces points, than they would have if they had just forced a simple 3-and-out.

Not sure if either of this is the case without seeing the exact success point tables that Aaron uses, but from what I know of DVOA it is possible.

What I would suggest is changing the success point system to scale all successes so that the most points a drive can ever produce is a fixed value (say...7?). In orther words, set up your DVOAish system so that it is drive based, not play-based. You can still look at every play, but it's worth at most a fraction of 7 points that relates how much you moved yourself towards that goal of ultimately getting 7 points. You can do this with probabilities of scoring. Say you have 1st and 10 from your opponents 40, and the probability of a team scoring the TD in this situation is determined to be 41%, and the probabilty of getting a FG is *8%. So your expected points for this drive is 0.41*7 + 0.48*3. You get a 6 yard un, now having 2nd-6 from the 34. Now your probability of TD is 42% and FG is +9%, so your expected points for this drive is now 0.42*3. The difference in the expected points is the point value of that play. You now compare THAT change in value to the league average performance in that situation, and correct for defense, etc.

This way, if you start from your own 1-yard line and string together 11 9-yard plays and score a TD, or if you throw one 99-yard pass, the sum total of success points for both plays will be almost 7. You've automatically compensated the value of the bomb for available field position, and haven't over-valued those small continuous plays.

A defense that gets a 3-and-out will get more points for those 3 plays than one that gives up a longer drive but still has a bunch of good plays in there.

I don't know for sure, but it's possible a fix of this kind could correlate better with wins than DVOA, especially for the defense...

by Jerry :: Wed, 10/12/2011 - 6:25am

Let's see...

Drives leave you with fewer data points than plays, which matters in a universe as limited as an NFL season. You also lose the ability to separate runs and passes, or assign value to individual players.

Treating plays as fractions of drives addresses those issues. However, let's say a team starts a couple of drives on its own 20. In each case, they pick up five yards on the first play, which should be the same VOA/DVOA. Should subsequent plays on the drive affect how we view that first one?

I think Brian Burke at advancednflstats.com does stuff with expected points and win probabilities. You'll have to spend more time there than I'm willing to right now to figure out whether he's doing some of what you're talking about.

To be clear, I'm just trying to think along with you, and these are the things that immediately come to mind. There's certainly a useful discussion to be had here.

by Sean McCormick :: Wed, 10/12/2011 - 12:53pm

Yes, Burke measures each play by where the play ended up and then calculates the expected points you would derive from that field position and/or the likelihood of winning the game based on score, down and distance, time, and field position.

by Aaron Brooks Go... :: Wed, 10/12/2011 - 2:43pm

That sounds like FEI.

by MJK :: Wed, 10/12/2011 - 3:27pm

I confess I'm not super familiar with FEI, because I don't follow college ball. I should read up on it more...

by Keith(1) (not verified) :: Tue, 10/11/2011 - 11:42pm

Think about it like this -- you get some value relevant to DVOA based on plays, and those plays individually have context. Our context: the game is tied, fourth quarter, two-minute drill.

1st and 10 from the 20 :: throw for 80-yard touchdown (15 points).
1st and 10 from the 20 :: throw for 70-yard completion (10 points).
1st and 10 from the 20 :: throw for 60-yard completion (10 points).
1st and 10 from the 20 :: throw for 50-yard completion (10 points).
1st and 10 from the 20 :: throw for 40-yard completion (10 points).
1st and 10 from the 20 :: throw for 30-yard completion (10 points).
1st and 10 from the 20 :: throw for 20-yard completion (10 points).
1st and 10 from the 20 :: throw for 9-yard completion (8 points).
1st and 10 from the 20 :: throw for 8-yard completion (6 points).
1st and 10 from the 20 :: throw for 7-yard completion (4 points).
1st and 10 from the 20 :: throw for 6-yard completion (2 points).
1st and 10 from the 20 :: throw for 5-yard completion (0 points).
1st and 10 from the 20 :: throw for 4-yard completion (-1 points).
1st and 10 from the 20 :: throw for 2-yard completion (-2 points).
1st and 10 from the 20 :: throw for 2-yard completion (-3 points).
1st and 10 from the 20 :: throw for 1-yard completion (-4 points).
1st and 10 from the 20 :: throw for 0-yard completion (-5 points).

In our fake example, you get a bonus for a big play (20+ yards), and a bonus for a touchdown on that big play. All other successful plays are awarded positive points. All negative plays are awarded negative points. A 50% play is awarded nothing, as it is exactly average. (This is not exactly what DVOA does, but stay with me.)

From that we can gather that a bunch of short gains that are positive will cumulatively be better than a big, explosive touchdown. So, say you have 1st and 10 from the 20 and throw for 0 yards; you now have -5 points. On the next play, you throw for the touchdown; you now have -5+15, for 10 points. Now, say you throw for 50 yards, for 10 points. Then you throw five 6-yard passes, for another 10 points. After the touchdown bonus, you get a couple more points, for a 20+ point total. That is slightly better than even the one long play, worth 15 points, if it happened on 1st and 10 from the 20.

Now imagine every down and distance, game situation, clock situation, score, team, and location, and each of those have individual attributes all conflating to make your "DVOA points." Every play is graded by what every other team is doing in those situations. If Team A scores a touchdown in that situation, yet it takes Team B seven plays from different downs and distances, game situations, and clock situations, either could still get more overall points for their successes. It just depends on what happened.

That is why negative plays are so negative and boom-bust teams look worse than they actually are -- the negative plays are such offsetting events that a team that plods "successfully" down the field by throwing and running for boring yards is more likely to be high in DVOA than, say, a team that throws for a 99-yard touchdown four times a game and gains no other yards.

by Eddo :: Wed, 10/12/2011 - 10:52am

Good explanation, MJK. In your example, even if the 80-yard play was ten times as valuable as each 8-yard play, your overall DVOA will still look worse. If you have 20 total plays in each scenario, you still have 19 plays with negative value in the one-long-play case, but only 10 with negative value in the ten-shorter-plays case.

by Aaron Brooks Go... :: Wed, 10/12/2011 - 2:46pm

There are times I think Leroy Hoard authored DVOA.

by RichC (not verified) :: Tue, 10/11/2011 - 10:11pm

I agree with you that there's a problem here, and its very similar to the problems with DIPS. Both DIPS and DVOA are drastically better evalution methods than what we had before, but at some point, people here kind of just accepted that they worked.

They stopped questioning whether or not they were working correctly, and instead started trying to describe why things are the way they are within that system.

That's a very, very bad thing. Its a fundamental shift in the philosophy, and its not a good one.

by Eddo :: Wed, 10/12/2011 - 10:56am

You actually bring up some useful points. I'm not sure that Aaron & Co. have evaluated DVOA at its most base level (as opposed to just tweaking the specific coefficients) when things look weird. Maybe they do, and just don't write about it. I couldn't say.

But you're going about it all wrong. You sound condescending with your comments, which will indeed prompt people to get defensive as opposed to approaching your points with an open mind. And the vast majority of your points are specific to the Patriots' defensive ratings. It could very well be that the Patriots' defense (like Matt Cain or Mark Buehrle with DIPS) is just an outlier to an otherwise quality system. Or it could be indicative of a more systematic failure. But we need more than the Patriots' defense being strangely rated as a data point.

by MJK :: Wed, 10/12/2011 - 3:26pm

The Patriots defense this week is actually a very good example of what might be going on in the system because it was so bipolar this week. Out of 9 drives where the Jets had the potential for a long drive, the Patriots utterly shut them down on 7 of them, and utterly fell apart on 2 of them. DVOA thinks that is a terrible defensive performance, and yet drive efficiency thinks it's a very good defensive performance. So we have a point where two objective metrics disagree sharply on how well a team played, which means that one game is an excellent tool for understanding WHY the different metrics disagree, and identifying ways that either one could be improved.

The question is: if a team shuts down offenses most of the time, but falls apart completely occasionally, is it a good defense? Kind of like a pitcher who is lights out until he gets his confidence shaken, and then he just starts giving up HR after HR...but only once in every 8 or 9 starts or so. (Like Dereck Lowe a number of years back).

by RickD :: Tue, 10/11/2011 - 6:51pm

A bit off-topic: have we heard from RaiderJoe since the death of Al Davis?

by Stats are for losers (not verified) :: Tue, 10/11/2011 - 7:11pm

I think many just assumed that they were the same person

by RickD :: Tue, 10/11/2011 - 10:51pm

That's going to be my conclusion if we don't see RaiderJoe again.

by Staubach12 :: Wed, 10/12/2011 - 1:56am

Honestly, I laughed when people said that on Saturday. Now, however, I am actually considering the possibility.
RJ, if you are still among the living, please let us know.

by Aaron Brooks Go... :: Wed, 10/12/2011 - 2:57pm

And if he's not -- this is one of the greatest trolls ever.

by pedropolis :: Wed, 10/12/2011 - 3:31pm

I read through the comments just looking for him. You know, if they were the same person it would explain his encyclopedic knowledge of the Raiders through the decades...

by The Hypno-Toad :: Thu, 10/13/2011 - 4:22am

Found him
I'm glad Raiderjoe is still around... But I have to admit that a part of me fell in love with the idea that he had actually been Al Davis the whole time.

by t.d. :: Tue, 10/11/2011 - 7:44pm

Detroit is interesting, because they are winning and have definite areas of strength, but they're still really incomplete. Maybe I'm too influenced from watching the Dallas game, but that offense is Cal Johnson/Brandon Pettigrew and a lot of crap. The intro talks about comparing their start to the Vikings' start, but we've already seen that matchup: the Vikings dominated for much of the game, then blew it. Stafford looks ok, but he's not near elite yet.

by ASmitty :: Wed, 10/12/2011 - 10:18am

Much of the game? I'm pretty sure the Vikings dominated half of that game, then got dominated for the other half. At home.

In fact, 25% of Minny's net yardage for the game came on two plays, both in the first half. The game was much closer in the first half than the score indicated, and the next half and OT bore that out.

I'll start taking the criticisms of the Lions more seriously when they struggle against someone at home, which may well happen the next two weeks. In my opinion though, other than the final W or L, what happens on the road, stays on the road. I think Detroit is looking excellent at home so far, and for all the "just a few big plays" talk I hear about the Bears game, I don't really hear anyone talking about how incredibly improbable Jay Cutler's performance was in that game as well.

If it were played again, I doubt Best goes off like that, but I also doubt Cutler completes as many long passes on a dead sprint and avoids committing any turnovers under the sort of durress he was in. Sports counter-factuals have to work both ways, you can't just eliminate the improbable plays by one team. Cutler made a handful of big, borderline impossible plays as well.

by Will Allen :: Wed, 10/12/2011 - 12:09pm

I think a lot of Cutler's heroics may be due to the Lions biggest weakness, which is defensive backs who do not close on the ball well. Now, this is a weakness that can get obscured when a defensive front is rushing the passer really well, especially against a qb who lacks Cutler's mobility, so it doesn't necessarily have to be a flaw that comes to the forefront consistently. You get a game where an offensive line has just fairly decent day in pass protection, with a decent qb, however, and you're looking at a defense that will give up 35 points.

Minnesota, to me, looks like it is ranked too high, and if I were to guess why, it is because the game may be shifting even more towards making a passing attack critical to winning. It may be even more unlikely now, compared to even 5 years ago, to construct a defense that can win games consistently, despite the team's offense being unable to throw downfield, thus DVOA is being too generous to teams which have the Vikings deficiencies in the passing game. Over the next couple of years, we may see if my suspicions are correct.

Detroit ranking doesn't strike me all that off-base, and I say that as someone who is rooting for them. They have a great defensive front masking significant deficiencies in their defensive backfield. They have a monster wide receiver, and some other good ones, along with a running back with above-average talent, masking a qb with accuracy issues and a so-so, at best, offensive line. I might put them at ninth or tenth right now, which isn't that far from where DVOA places them. It really is hard to be an elite team in the NFL anymore, if you don't have an elite qb.

by Will Allen :: Wed, 10/12/2011 - 12:19pm

I'll repeat, however, that the Lions' defciencies in the defensive backfield and at qb, could be things that change dramatically over the course of the season, given the age of the players. They might be much better come January.

by ASmitty :: Wed, 10/12/2011 - 12:22pm

I don't have much disagreement with that.

I mentioned it up-thread, but I think Detroit's success coming from behind is almost predictable. When teams get ahead and try to chew clock, they run the ball right into the strength of the defense and stop trying to beat the CBs deep, which is where Detroit is weak. Offensively, it's not like the Detroit cares if they're behind or not. When they're ahead it's almost scary though, because they can't run the ball and their soft shell defense is very, very soft.

I really can't overstate Cutler's game on Monday though. He looked like he was throwing some of those passes with his eyes closed, but then the ball would end up accurate enough to land in a shot glass. Anything less than that performance, and the game would have been a rout, if for no other reason than the overwhelming field position advantage Detroit would have enjoyed.

by Will Allen :: Wed, 10/12/2011 - 12:25pm

Like I said in the quick reads thread, if you switch qbs' teams in that game, the Lions likely win 38-0.

by Aaron Brooks Go... :: Wed, 10/12/2011 - 3:00pm

That said, if you switched the teams' DLs, the Bears win 38-0.

by Will Allen :: Wed, 10/12/2011 - 3:57pm

Yeah, but that would be nine Lions for, what, six Bears?

by nuclearbdgr :: Tue, 10/11/2011 - 8:10pm

Do my eyes deceive me? Is the Packers ST DVOA actually positive?

by DisplacedPackerFan :: Tue, 10/11/2011 - 8:41pm

It is!

Not only that the only negative aspects are the punting and punt return teams. I still have a lot of faith in Masthay, but since he never punts anymore the few bad punts he had earlier are still dragging him down. He was good to excellent for more than half of last season, there is reason to believe he will return to that form.

Punt returns I'm not sure about. It's close to 0 (so average) Cobb has ability, the blocking I still wonder about. Same with kickoff returns, the blocking has still been below average on that and Cobbs 108 yard return is still carrying a lot of weight.

by rggd_clws (not verified) :: Tue, 10/11/2011 - 9:20pm

I was just wondering - SF and BUF have shot up the rankings - how predominant do wins over PHI affect that? BEating a team who started the season with higher expected wins?

I'm just morbidly wondering if the Bird's blitzkrieg of atrociousness is skewing the rankings.

by LionInAZ (not verified) :: Tue, 10/11/2011 - 9:41pm

Now THAT is an elephant in the room here.

Are the Eagles screwing up DVOA by not being as dreamy as expected?

by theshadowj :: Tue, 10/11/2011 - 10:16pm

Unless I'm not understanding what you're saying, I think you guys are confused about how DVOA works. It doesn't take into account any preseason projections or expectations of a team, it only looks at the plays on the field. DAVE does take into account preseason projections, but I don't think that's what you were talking about.

by MC2 :: Tue, 10/11/2011 - 10:23pm

As I understand it, the opponent adjustments are based strictly on this year's DVOA. They are not based on DAVE (or on anything else that takes the preseason projections into account).

by TomC :: Wed, 10/12/2011 - 12:59am

Very minor detail, but I'm mildly curious how the Bears' unadjusted VOA is higher than their DVOA when they've played the 6th hardest schedule so far (average opponent DVOA of +10.1%).

by andrew :: Wed, 10/12/2011 - 11:42am

As DAVE is only used early on, and increasingly less so as the early part winds down, I think this is a useful stat to track how well a team is exceeding, living up to or failing expectations. The further into the season you go, the more we'd start facing reality that the bad teams really are bad, and maybe beginning to believe the good teams are that good. Ranking DVOA - DAVE, you get...

Team Expectations
BUF 10.10%
SF 10.00%
TEN 9.80%
OAK 8.10%
WAS 7.20%
CIN 5.30%
MIN 4.60%
GB 4.30%
BAL 4.20%
DET 4.20%
DAL 4.20%
NO 2.10%
NYG 1.70%
DEN 1.60%
HOU 1.50%
SEA 1.30%
CAR 0.60%
NYJ 0.40%
TB 0.00%
NE -0.60%
ATL -1.60%
KC -2.50%
ARI -2.60%
CLE -2.90%
IND -3.50%
PIT -5.10%
CHI -5.40%
SD -6.80%
PHI -8.40%
JAC -8.80%
MIA -10.40%
STL -10.90%

Thus some teams, despite being good, are leaving people somewhat underwhelmed (Patriots, Steelers) while seattle and carolina are bad, just not as bad as we thought they would be.

by jpg30@earthlink.net :: Wed, 10/12/2011 - 2:42pm

I like your idea, Andrew. It seems like a good starting point for looking at teams that may be regressing in coming weeks. Can a team like Philadelphia, for instance, with their superior personnel continue to clock in at -8.4% week after week? For me, it makes their game at Washington (+7.2%) this week look a lot like the "Regression Bowl."

by Anonymous1 (not verified) :: Wed, 10/12/2011 - 5:33pm

Living up to the #1 DAVE is "leaving people somewhat underwhelmed"? Defensively, syre, but the offense and overall team is quite good. You need to add another layer to your analysis if you want to draw any conclusions from the data.

by Nathan :: Wed, 10/12/2011 - 5:59pm

From Matt Bowen's article on what's wrong with the Bears defense:

The safety position. It is a serious issue in Chicago right now. Poor angles to the football, bad tackling and the inability to read route schemes. That’s two straight weeks of below average play from Brandon Meriweather and Chris Harris was the safety victimized by C.J. last night. It would be the same results if the Bears went to a Cover 1 scheme or brought more pressure. The safeties just aren’t accountable for Lovie right now.

Sound familiar to any Pats fans besides me?

by Aaron Brooks Go... :: Thu, 10/13/2011 - 1:41am

It does, but part of the problem is a certain amount of schematic rust.

The Bears' Cover-2 is predictable. There are standard ways to beat it. And those ways *always* work, because the Bears never adjust.

Best got an 88-yd run and a 44-yd run, from the exact same play, because that play beats Cover-2. When you never need to even think about another scheme appearing, it's easy to run those Pavlovian plays and make the LBs drool in the wrong direction.

by greybeard :: Thu, 10/13/2011 - 2:33am

I think blaming the defensive scheme for those runs is unfair. Both runs happened because the LBs could not peel off and Harris attacked the wrong gap instead of waiting the RB to come to him. If your players are not good you will loose with any scheme.