Week 12 DVOA Ratings

Week 12 DVOA Ratings

by Aaron Schatz

The top teams remain the same in this week's Football Outsiders DVOA ratings, with Houston still leading the way despite their rash of quarterback injuries, followed by Green Bay and New England. Below that there were some larger moves, particularly because of two games. Baltimore and New Orleans move up, while San Francisco and the New York Giants move down.

Once again this week, we've done two versions of our playoff odds report. The second version attempts to estimate the drop in weighted DVOA we would expect when a team has to use a backup quarterback -- or now in Houston's case, a third-string quarterback. Despite these adjustments, you might think the odds we have listed for Green Bay to go undefeated are a bit low. Right now, we have this at just 11.9 percent on the "injury-adjusted" version of the playoff odds report. There are two issues here. One, of course, is that we have Green Bay listed as just another very good team, not the kind of dominant team we would usually be expecting to challenge for an undefeated season. Second, the formula we use to figure out games in our playoff odds report is based on the last few years of results, and the fact is that upsets are a lot more likely than many fans realize. Good teams beat very good teams all the time, and even average teams beat very good teams.

Here are the chances the formula gives to Green Bay in each game when figuring out whether to assign the Packers a win or loss in each simulation:

  • Week 13 at NYG: 51%
  • Week 14 vs. OAK: 73%
  • Week 15 at KC: 69%
  • Week 16 vs. CHI: 67%
  • Week 17 vs. DET: 68%

So the Packers are technically favored in every game, yet their chances of winning all five are very low. Even if you were to increase the odds of winning each game by ten percentage points, the probability of winning all five would still be only 24.0 percent. So even in the home stretch, going 16-0 looks very difficult.

Here are the same chances given for Indianapolis to win in each of its five remaining games:

  • Week 13 at NE: 5%
  • Week 14 at BAL: 5%
  • Week 15 vs. TEN: 46%
  • Week 16 vs. HOU: 38%
  • Week 17 at JAC: 18%

* * * * *

So, why have the Denver Broncos suddenly gone 5-1 over their last six games after beginning the season 1-4? Is it Tim Tebow, improvement in the running game, or improved defense? That last element is getting a lot of credit from smarter fans. What if I told you it wasn't quite true?

If we look at the Denver Broncos defensive numbers before and after Tim Tebow took over in Week 7, we see almost no difference:

DEN defense Total Rk Pass Rk Run Rk
Weeks 1-6 7.7% 19 19.1% 21 -5.0% 16
Weeks 7-12 4.6% 19 13.6% 22 -6.5% 12

Now, part of the reason why there's no difference is that the Broncos got clobbered by Detroit 45-10 back in Week 8. So what if we set the line after that game instead of after six weeks? Well, now we see a little bit of improvement, but still nowhere near as much as you might expect.

DEN defense Total Rk Pass Rk Run Rk
Weeks 1-8 8.3% 21 19.5% 25 -4.4% 14
Weeks 9-12 2.1% 15 10.4% 20 -8.1% 14

Now let's look at the offense. Look at the running game before and after Tebow took over as quarterback, and you'll see where the real improvement in Denver has been:

DEN offense Total Rk Pass Rk Run Rk
Weeks 1-6 -8.1% 27 -4.8% 28 -8.9% 26
Weeks 7-12 1.0% 18 -8.1% 23 11.6% 7

First of all, in case you are wondering, Denver's pass offense DVOA has gone down overall but up in ranking because the overall leaguewide passing DVOA has gone down since those first few ridiculous weeks that started the season. Still, the real change here is in the running game. We've written about this a couple times, including a few years ago in one of the first pieces we ever did for ESPN, but the threat of a running quarterback really does open things up for the rest of the running game. Clearly, that's what has happened in Denver. Some of the improvement is Tebow's own rushing ability, but a lot of it is an improvement from Willis McGahee because opponents have to constantly watch out for Tebow carrying the ball himself. When some commentators have said the option looks seem to be working in Denver, this is what they mean. Denver's offense has improved more than Denver's defense, and the biggest reason for the improvement really is Tim Tebow.

Of course, the other element in Denver's 5-1 record is pure, old-fashioned luck. Five of Denver's six wins this year have come by seven points or less, including two in overtime. If Nick Novak hits a 53-yard field goal -- or if Norv Turner doesn't go all conservative on that drive before the field-goal try -- this conversation seems a bit different.

* * * * *

These are the Football Outsiders team efficiency ratings through 12 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. WEIGHTED DVOA represents an attempt to figure out how a team is playing right now, as opposed to over the season as a whole, by making recent games more important than earlier games.

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

To save people some time, please use the following format for all complaints:

<team> is clearly ranked <too high/too low> because <reason unrelated to DVOA>. <subjective ranking system> is way better than this. <unrelated team-supporting or -denigrating comment, preferably with poor spelling and/or chat-acceptable spelling>

1 HOU 35.1% 1 36.7% 1 8-3 23.8% 4 -8.9% 7 2.4% 9
2 GB 27.1% 2 25.8% 2 11-0 37.6% 1 12.3% 26 1.9% 10
3 NE 25.2% 3 25.2% 3 8-3 33.5% 2 10.9% 23 2.6% 8
4 NYJ 22.6% 5 22.6% 5 6-5 4.4% 18 -13.6% 3 4.6% 3
5 BAL 21.3% 9 21.1% 7 8-3 6.3% 15 -19.4% 1 -4.3% 29
6 PIT 20.7% 7 23.6% 4 8-3 17.4% 5 -2.0% 9 1.3% 13
7 NO 19.6% 11 18.5% 10 8-3 31.2% 3 11.2% 25 -0.4% 21
8 SF 19.4% 4 19.6% 8 9-2 1.2% 20 -9.7% 6 8.6% 2
9 CHI 19.3% 6 21.9% 6 7-4 -2.6% 22 -10.8% 5 11.1% 1
10 ATL 17.3% 10 18.6% 9 7-4 10.3% 10 -6.9% 8 0.0% 16
11 DET 9.6% 13 9.3% 11 7-4 5.1% 17 -10.8% 4 -6.4% 31
12 TEN 8.9% 15 6.2% 16 6-5 7.4% 13 2.7% 12 4.2% 4
13 DAL 8.4% 12 8.8% 12 7-4 10.6% 9 0.6% 11 -1.6% 25
14 NYG 7.1% 8 6.3% 15 6-5 13.3% 7 6.2% 18 0.0% 18
15 CIN 6.8% 16 7.1% 13 7-4 7.6% 12 3.6% 15 2.8% 7
16 PHI 5.3% 14 6.8% 14 4-7 11.0% 8 7.0% 19 1.3% 12
17 OAK 1.8% 18 0.7% 17 7-4 7.1% 14 5.3% 16 0.0% 17
18 MIA -2.7% 17 -1.1% 18 3-8 0.5% 21 2.8% 14 -0.4% 20
19 BUF -4.0% 20 -9.4% 22 5-6 10.0% 11 12.9% 28 -1.0% 22
20 DEN -5.6% 19 -3.8% 19 6-5 -2.8% 24 6.1% 17 3.2% 6
21 TB -6.7% 21 -9.3% 21 4-7 3.1% 19 11.0% 24 1.2% 14
22 SD -11.3% 23 -12.0% 24 4-7 6.1% 16 14.3% 30 -3.1% 27
23 JAC -11.7% 26 -9.1% 20 3-8 -22.2% 31 -14.1% 2 -3.5% 28
24 CAR -12.0% 28 -12.0% 23 3-8 15.4% 6 21.7% 32 -5.8% 30
25 WAS -14.0% 22 -16.1% 27 4-7 -12.2% 27 0.3% 10 -1.5% 23
26 CLE -14.1% 25 -13.4% 26 4-7 -7.0% 25 7.1% 20 0.0% 19
27 SEA -14.4% 24 -13.0% 25 4-7 -11.9% 26 2.7% 13 0.2% 15
28 MIN -16.7% 27 -18.5% 28 2-9 -2.7% 23 12.5% 27 -1.6% 24
29 KC -25.5% 30 -23.7% 29 4-7 -18.1% 30 8.8% 21 1.4% 11
30 ARI -26.4% 29 -27.1% 30 4-7 -16.9% 29 13.1% 29 3.6% 5
31 STL -36.1% 31 -34.7% 31 2-9 -25.4% 32 9.0% 22 -1.6% 26
32 IND -42.5% 32 -44.6% 32 0-11 -16.1% 28 19.5% 31 -7.0% 32
  • 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 HOU 35.1% 8-3 40.8% 8.2 2 -2.4% 27 -4.3% 25 16.8% 24
2 GB 27.1% 11-0 29.3% 9.1 1 -2.1% 26 2.4% 15 3.0% 3
3 NE 25.2% 8-3 21.9% 8.1 3 2.0% 12 -13.8% 30 10.6% 15
4 NYJ 22.6% 6-5 18.5% 6.6 11 3.9% 7 -6.0% 26 14.8% 23
5 BAL 21.3% 8-3 18.9% 7.8 4 6.6% 3 -15.0% 31 24.2% 30
6 PIT 20.7% 8-3 21.4% 7.5 6 -1.0% 22 -7.6% 29 14.6% 22
7 NO 19.6% 8-3 19.3% 7.5 5 -2.5% 29 1.4% 16 13.5% 19
8 SF 19.4% 9-2 21.9% 7.2 8 -0.3% 21 -18.4% 32 5.3% 6
9 CHI 19.3% 7-4 16.0% 7.1 9 4.0% 6 -7.0% 28 9.1% 13
10 ATL 17.3% 7-4 14.0% 7.2 7 -0.2% 20 4.9% 10 2.6% 2
11 DET 9.6% 7-4 11.1% 6.5 12 2.4% 9 4.1% 11 7.7% 8
12 TEN 8.9% 6-5 12.9% 6.2 14 0.2% 19 -0.7% 21 17.9% 25
13 DAL 8.4% 7-4 11.0% 6.3 13 -1.1% 23 -2.7% 23 18.0% 26
14 NYG 7.1% 6-5 7.8% 6.2 15 -1.8% 24 10.5% 5 19.4% 27
15 CIN 6.8% 7-4 14.6% 6.6 10 0.6% 17 2.9% 13 1.0% 1
16 PHI 5.3% 4-7 4.2% 5.7 17 1.4% 15 0.0% 18 12.9% 17
17 OAK 1.8% 7-4 6.6% 5.9 16 2.8% 8 -0.6% 20 25.7% 31
18 MIA -2.7% 3-8 -3.0% 5.6 18 2.2% 11 10.2% 7 12.9% 16
19 BUF -4.0% 5-6 -4.3% 5.3 20 4.6% 4 2.9% 14 30.2% 32
20 DEN -5.6% 6-5 -3.7% 5.5 19 2.4% 10 -0.4% 19 7.5% 7
21 TB -6.7% 4-7 -15.4% 5.0 21 14.5% 1 -2.0% 22 20.7% 28
22 SD -11.3% 4-7 -6.3% 4.3 27 1.1% 16 3.4% 12 4.0% 4
23 JAC -11.7% 3-8 -14.9% 4.4 25 9.2% 2 -6.9% 27 8.6% 11
24 CAR -12.0% 3-8 -9.8% 4.5 24 -2.5% 28 11.7% 3 14.4% 21
25 WAS -14.0% 4-7 -14.4% 4.4 26 -6.0% 32 8.7% 8 13.8% 20
26 CLE -14.1% 4-7 -6.3% 4.5 23 -4.4% 30 11.5% 4 5.2% 5
27 SEA -14.4% 4-7 -11.1% 4.6 22 0.3% 18 -3.7% 24 9.6% 14
28 MIN -16.7% 2-9 -15.1% 4.1 28 4.2% 5 5.7% 9 13.5% 18
29 KC -25.5% 4-7 -23.2% 3.3 30 -1.8% 25 13.0% 2 22.5% 29
30 ARI -26.4% 4-7 -18.0% 3.4 29 -5.0% 31 1.2% 17 7.9% 9
31 STL -36.1% 2-9 -36.9% 2.5 31 1.8% 13 10.4% 6 8.1% 10
32 IND -42.5% 0-11 -46.5% 1.7 32 1.7% 14 15.8% 1 9.0% 12


347 comments, Last at 04 Dec 2011, 9:18am

2 Re: Week 12 DVOA Ratings

How does special teams DVOA factor in fumbles on returns? The Jets have, I think, 5 fumbles on kick returns, losing a fair number of them. They've been devastating. And yet the Jets are still 3rd in ST. It's true that aside from the fumbles the ST have been excellent, but the fumbles are a pretty big deal.

5 Re: Week 12 DVOA Ratings

There is a bigger difference between the Bears (1st) and Jets (3rd) in special teams DVOA than the Jets and the Rams (26th).

And this is during a year when Gould has missed more kicks than any other year in his career. Which I think you can give Maynard some credit for that. He was an outstanding holder.

38 Re: Week 12 DVOA Ratings

Huh? Gould has only missed two kicks this year. He is 23 for 25, which is 92%. That is the highest percentage of his career. Every other year he has missed between 3 and 6 FGs, and 89.7% was his previous season high. He also is 5-5 on 50+ yard FGs this year. Those 5 50+ yard FGs match the number he had made in his career before this season.

I would say Gould is having his best season. And that doesn't even take into account his outstanding results this year on kickoffs.

199 Re: Week 12 DVOA Ratings

I thought he was having a particularly good year. He's always been super-accurate on short-to-medium kicks but anything over 50 has been iffy. His accuracy on those long kicks this year has been very impressive. That last field goal on Sunday - I think it was a 53-yarder? - was almost dead center between the uprights and would have been good from 10 yards further.

3 Re: Week 12 DVOA Ratings

How did you calculate probability of winning? Is this just a simple equation using DVOA?

88 Re: Week 12 DVOA Ratings

Right. The only way I was able to duplicate Aaron's results in the right order was by using unweighted DVOA which ignores the QB situations. The playoff odds supply DVOA which does not ignore these situations, and results in a slightly different order.

343 Re: Week 12 DVOA Ratings

Brings up an interesting question: which teams have the biggest DVOA advantage at home?? That would make a great article for next week.

Completely unrelated question: how do unrecovered forced fumbles affect defensive DVOA? I recall seeing a stat on MNF that New Orleans has forced something like 18 fumbles but only recovered 4!?!?

4 Re: Week 12 DVOA Ratings

What is the statistical likelihood of NE's weighted DVOA being exactly the same as its unweighted?

6 Re: Week 12 DVOA Ratings

The more statistically honest question is "What is the statistical likelihood that at least one NFL team's weighted DVOA would be exactly the same as its unweighted DVOA through week 12?"

46 Re: Week 12 DVOA Ratings

The even more statistically honest question is "What is the [trend in the] statistical likelihood that at least one NFL team's weighted DVOA would be exactly the same as its unweighted DVOA at any time in the season after weighted DVOA starts being calculated?"

18 Re: Week 12 DVOA Ratings

No real idea, but perhaps not too unlikely since NYJ and CAR managed the same feat this week?

7 Re: Week 12 DVOA Ratings

Is it possible to add something to the playoff odds about the distribution of wins needed to get various seeds? For example, if the 6 seed in the NFC would have a record of 9-7 40% of the time, 10-6 50% of the time, 11-5 2% of the time and 8-8 3% of the time? (Numbers are hypothetical, of course, but that would give a feel of how good teams would need to be to get into the playoffs.)

8 Re: Week 12 DVOA Ratings

I know the NY Jets were covered in a previous article, but it still shocks me to see them so high. Maybe they'll get lucky and Mark Grossman will get them to the Superbowl, like the 2006 Chicago Bears.

9 Re: Week 12 DVOA Ratings

New Orleans should be 8-3.

It still amazes me that one of Green Bay, New England and New Orleans is the 3rd best offense in the league, and that one of Rodgers, Brady and Brees is having the 3rd best season by a QB this year. Brady is on pace for 5275 yards with 40 TDs and might not get a single MVP vote.

70 Re: Week 12 DVOA Ratings

It still amazes me that people don't seem to have noticed the explosion in passing game numbers since both the 'Polian rule' and the 'holding on offense clarification' and seem to think that impressive statistical performance under these rules relates to players being the best ever. Most of the top DVOA QB seasons are now held by post Polian players but their performances are routinely credited to passing game 'improvements' rather than 'easier rules for QBs'.

76 Re: Week 12 DVOA Ratings

I agree that rule changes matter. But some of it is that the truly all-time-great QBs aren't evenly distributed - they "clump" randomly.

1983-1985 brought us some truly amazing QBs. 1998-2001 brought us some more. But who was drafted in that 13-year gap between Randall Cunningham and Peyton Manning?

Well, Brett Favre. OK. That's one. Who are the second and third best QBs drafted in that time period? Go on, name them! Troy Aikman and Mark Brunell? Steve McNair and Drew Bledsoe?

Are these guys really supposed to be the successors to Dan Marino and Steve Young? Are they really supposed to be the predecessors to Tom Brady and Peyton Manning?

Some eras have better QBs than others. We are in an era of great QBs right now. Just sit back and enjoy it.

113 "All Time Great" QBs.

It still amazes me that people don't seem to have noticed the explosion in passing game numbers since both the 'Polian rule' and the 'holding on offense clarification' and seem to think that impressive statistical performance under these rules relates to players being the best ever.

The old-time greats are both systematically denigrated (by highly biased official stats like passer rating) and culturally denigrated, as a whole lot of football fans seem to think no player could ever have been a great if not seen on an ESPN highlight show.

No old-time great is more denigrated, for some strange reason, than Namath -- who's numbers hold up damn well against anybody's, playing today or any other time -- but the rest are as well, Unitas too, all of them. The difference in appreciating the players of the past seen between baseball fans and football fans is really striking.

The rule changes that *really* opened the passing game were in 1978. The rest since then have only incrementally increased passing. But the 1978 changes not only made it much easier to pass, they changed what won games, the type of passing that won, shifting it to short, safe (boring) passes, from the dramatic long balls of yore, and substituting safe, short passing for former rushing, padding the numbers of all later-era QBs, no matter how mediocre.

Since then the great passers of before then have been exterminated on the "all time great lists" and pretty much in the minds of fans who think the only good kind of passing is today's, and that today's stats somehow properly measure the quality of passing back then.

The NFL Hall of Fame actually published an All Time Top Passers ranking list that had all the top 17 QBs (and 29 of the top 30!) as playing after 1978, with Mark Bulger #5 all time. Can you imagine if the Baseball Hall of Fame ever published a ranking list like that? Baseball fans would burn the place down. Football fans go, well, the QBs of today of course are all the best ever, right?

125 Re: "All Time Great" QBs.

Actually, I think football fans gave a collective bemused shrug if they even knew such a list existed.

130 Re: "All Time Great" QBs.

And I doubt anybody, even including people named Bulger, thinks that Mark is the 5th best QB ever.

BTW, Jim, that was nice work. I suspect that much of the negativity about Namath was and is a reaction to the hype at the time. If he'd played in, say, Houston instead of New York, he'd probably be viewed more accurately.

258 Re: "All Time Great" QBs.

BTW, Jim, that was nice work.


I suspect that much of the negativity about Namath was and is a reaction to the hype at the time. If he'd played in, say, Houston instead of New York, he'd probably be viewed more accurately.

Yes, I think there's something to that. One gets a huge amount of hype, glory and fame, later comes the reaction and blowback. Too far each way.

But I don't think Joe can complain overall. "Houston Joe" or "Minnesota Joe" just wouldn't have been the same.

275 Re: "All Time Great" QBs.

I'm pretty sure Joe is content with his football career. I suspect that "Beaver Falls Joe" would have been happy enough anywhere he'd played.

302 Re: "All Time Great" QBs.

Namath also threw a lot of interceptions and his last good year was probably '71 or '72 (that's based strictly on memory, but I think it's probably a pretty accurate guess). Obviously, his bad knees were a significant factor in his early decline.

132 Re: "All Time Great" QBs.

I fully agree that there were bigger changes in 1978 than anything else. I am suggesting that the last five or six years or so seem to have an inordinate amount of elite numbers for QBs and this needs to be born in mind alongside how the old guys get stiffed in general.

156 Re: "All Time Great" QBs.

A big thing that gets missed in the evolution of QB's is the evolution of QB development at a very young age. In the last 10 years there has been an explosion of year round specialty among young quarterbacks with individual coaching and elite passing camps. Beyond that there is a plethora of pro-style and spread passing offenses not only all the way down to the high school level, but even the junior high level.

it's easy to point to the evolution of the rules as the reason QB's dominate today's game. It's obviously a component of it but you also have to look at the sheer amount of time that young players at the QB position put in at a very young age with elite coaching.

It all goes back to the Malcolm Gladwell idea of thousands of hours or practice. Elite young QB's of today, like Matt Barkley and Andrew Luck, literally have put in more time at the position with a higher level of specialized coaching than their predecessors of 30 years ago.

30 years ago a young QB was playing multiple sports. Matt Barkley, a potential elite QB, has seen specialized 1 on 1 coaching since he was 8 years old.

165 Re: "All Time Great" QBs.

Jimmy Clausen is a perfect example of how concentrated work can take a QB and turn him into an NFL elite ... ok, I just can't finish that with a straight face.

252 Re: "All Time Great" QBs.

and don't forget video games. Any under 30 QB has been playing Madden, or a lot of other Playstation XBox Wii products ever since they were a little kid-- these games basically took off in the mid 1990s. It can't help but aid their hand-eye coordination and reaction time that is duplicate to some extent on the field.

157 Re: "All Time Great" QBs.

I will take this opportunity to once again argue that Tarkenton is reasonably seen as the best ever at the position. Played his career under the old rules. Played the first 2/3 of his career, and spent his physical prime, with markedly inferior teammates, yet still put up numbers that hold up extremely well today. I love me some Unitas, some Montana, some Otto Graham, some etc., etc.,. Now imagine how those guys would be viewed today if they hadn't benefitted from competent teammates until they were well into their 30s.

164 Re: "All Time Great" QBs.

You link to a study that compares Namath's five best years to the five best years of a number of other QBs.

So great, we can throw away Namath's worst eight seasons and..ta da, Namath was excellent! The problem is that you don't have to cherry pick seasons for the other QBs on that list. If we want a good measure of how good a QB was over the entirety of his career, we really cannot simply cherry pick which seasons to look at.

That study is also flawed because it engages in the practice of "converting numbers from one era to another" or "adjusting for era". I'm not sure what this is supposed to mean. Is it supposed to mean that a QB who was at the 98th percentile in 1973 would have been at the 98th percentile in 2003? I know it's hard to compare performances across eras, but the solution isn't just to do a linear transformation of stats. Everybody in the NFL is far more athletic today than they were in Namath's day. Linemen are bigger, defenders are faster, defenses are more complex...it's just not meaningful to do this kind of transformation.

Yes, Namath threw for a lot of yards. He also threw with poor accuracy compared to the leading QBs of his day. He threw a lot of picks. He had a 13-year career and threw more picks than TDs in 11 of those 13 seasons.

256 Re: "All Time Great" QBs.

You link to a study that compares Namath's five best years to the five best years of a number of other QBs.

So great, we can throw away Namath's worst eight seasons and..ta da, Namath was excellent! The problem is that you don't have to cherry pick seasons for the other QBs on that list

Dude, you've got it backwards. Namath only played seven healthy seasons. The "cherry picking" favored Brady, Elway, Montana and the rest who got to pick their best 5 of up to *13* full healthy seasons. That's a big advantage for *them*.

Best seasons to best seasons, Namath was fully their match. That's by yards, by yards/attempt, by AYA including the cost of picks. Deal with it.

(Hey, nobody pretends that Namath had more than about six good years because of all the injuries he had. But if you want to argue "Namath crippled and playing 4-game seasons wasn't as good as other top QBs healthy, so he sucked!", go for it! Be proud.)

He also threw with poor accuracy compared to the leading QBs of his day...

Guy, you read but refuse to comprehend. Unitas was MVP two times with both personal and league *low* accuracy numbers, completion percentages, because what won games back then was *yards per completion*, and he had huge Y/C those years. Throwing deep as he did during those great seasons reduces "accuracy". Do you think the MVP voters were wrong because his completion percentages were poor compared to his peers, and to his own other seasons, those years? :-)

Namath's yards/completion in 1972 was *four standard deviations* better than average -- and that is what *won games* in those days. Yards/completion contributed *twice as much* to winning compared to "accuracy" then. What wins games has changed since then, look and see and try to comprehend.

You are the perfect example of what I was talking about, fans who don't comprehend how the game has changed and think the old-time greats sucked because they didn't pile up the numbers that matter to you -- while you have no idea of what the numbers were that mattered *to them.*

Yards-per-completion had double the importance to winning of accuracy, but you ignore huge, dominating yards-per-completion and talk accuracy.

He threw a lot of picks.

Did you see the AYA number in the table were he ranks ahead of, Montana, Elway, etc? That includes the cost of picks at -45 yards each.

When Namath played the TD-Pick ratio was 1 to 1.4, today it is 1 to 0.7. To say "he threw a lot of picks" compared to the QBs you know, without realizing only half as many picks are thrown today by everybody because the passing rules are so much easier, is dim. You have to adjust for changing norms like that, which is what that table does.

And if you bothered to read to the end of the story you know that the regressions that determine how the components of the QB stat line correlate to winning, in each QB's own time, say Namath's stat line in 1972 was at a .767 winning level, while the highest number in 2010 was .714 by Brady.

And yeah, that .767 was with his "low" accuracy and all those picks.

188 Re: "All Time Great" QBs.

In reply to by Jim Glass (not verified)

While I agree that the "old timers" aren't as valued, they really aren't in any sport other than baseball. (bad example - Lebron James not wearing #23 any more because it was Jordan's number, but he takes #6, worn by Bill Russell AND Julius Erving. wow)

But using Namath is a poor example. 50% passing completion for his career is OK, and most years of his career he threw more INTs than TD. (over the career, almost than 50 more ints) Namath never led the league in passing percentage, and only once in passing touchdowns, yet 4 times in interceptions.

Johnny Unitas had a slightly higher percentage (54%), and threw about 40 more TDs than INTs. Thats a huge difference.

Namath also had a career post season record of 2-1. Granted, one of those wins sealed his legacy. I know not as many teams made the playoffs back then, a QB as dominant as you seem to want to think Namath was should have made the playoffs more than twice. Unitas - 5 playoff appearances.

Finally, lets look at career record. Namath actually had a losing record for his career.

Namath was a very good quarterback, no question, I'm not arguing that. But he's a bad example of a player who's unfairly denigrated. Think about it this way - if he was so great, why did it take him 3 years to get into the Hall of Fame? Unitas - first ballot.

BTW - Namaths most famous pass attempt, directed at Suzy Kolber, also fell incomplete.

220 Re: "All Time Great" QBs.

Note: This is a reply to the discussion as a whole, not specifically to the unverified bouch.

It isn't just the Polian rule that has been helping quarterback performance. As pointed out above, one major factor is the rule change in the late 70s that made it easier for receivers to run free into the secondary. There have been two other changes that have boosted the productivity of the passing game; it has been made easier for quarterbacks to throw the ball away, preventing both the loss of yardage taken on a sack and the risk of injury to the passer. Additionally there have been changes to protect both quarterbacks, via the 'Brady rule', and the new rules to protect receivers. If you put the 2011 Packers/Saints/Pats offense against the 85 Bears under 85 rules then I'd expect those teams to dramatically change their offenses because their qb would probably not finish the game if they stuck with their 4 and 5 receiver sets. Conversely, if the 85 Bears played today without changing their style I'd expect a personal foul every drive, which would limit the efficiency of their defense. That the game is safer for offensive skill players means that they will be more productive.

Another important factor is the evolution of coaching and schemes. Sid Gillman and Don Coyrell created some of the first modern passing systems and some teams run these to this day and are still effective. Then Bill Walsh installed another evolutionary element as the 49ers stretched the field horizontally while June Jones and Mouse Davis thrilled with the run and shoot. Today many of the better offenses incorporate many of these elements into their schemes, standing on the shoulders of the giants that preceded them while adding in the increased use of shotgun snaps and the no huddle perfected by Tom Moore. All of these factors will serve to make offenses more effective.

The addition of the helmet radios will also have helped the offense over the defense. As well as getting the play to the quarterback the best organised teams will be able to throw in a few coaching points and while the defense has the same opportunity it is more influential to get these notes to the qb, who is responsible for at least half of the passing game than to one middle linebacker or safety. The ability to get clear signals in quickly will also help offensive performance.

One question I have concerns the helmet radios. I'm pretty sure that the rule is supposed to be that the coach can only speak to the qb if the offense huddles but I've seen many occasions this year where the offense doesn't huddle, the defense is unable to substitute and the qb is stood holding his hands over the earholes on his helmet, clearly trying to listen to his coach (I noticed this with the Tyler Palko against the Steelers as they ran their no huddle last week). I like the no huddle but only if it's run on a fair basis. If you don't huddle then you shouldn't be able to use the radio.

All in all I think it's easier for quarterbacks to achieve high levels of production now than it was in the past and direct statistical comparison of quarterbacks from today and as recent as the 1980s is not quite fair on the earlier players.

234 Re: "All Time Great" QBs.

The 70s rules changes which outlawed the head slap, and gave offensive linemen more latitude with regard to the use of their hands, and the extension of their arms, were just as important, in my opinion, as the restriction on when receivers can be hit. Give Reggie White and Bruce Smith the head slap, and they might have each ended up with 300 sacks.

249 Re: "All Time Great" QBs.

The Brady rule really isn't a new rule so much as a rewriting or clarification of an existing rule.

When Brady was hit low by Bernard Pollard, hitting the QB low was already a foul. That Pollard wasn't flagged for it is still something that baffles me.

Before it was the Brady rule, it was the Carson Palmer rule. But apparently not all the refs got the message.

251 Re: "All Time Great" QBs.

No, it was changed. The Carson Palmer rule was that you can't intentionally go low on a QB, but once you've been blocked to the ground you were allowed to hit a QB low. The Brady rule got rid of that exception.

259 Re: "All Time Great" QBs.

The LB who calls the defensive signals also has a radio in his helmet.
Why should huddling or not huddling make a difference?

264 Re: "All Time Great" QBs.

Well I always equated the use of no huddle with the advantage of preventing the defense from substituting, with the downside being that your qb has to call his own plays.

10 Re: Week 12 DVOA Ratings

Speaking purely as a Broncos fan, it's been gratifying to see some sane commentary here and other threads regarding Tebow and the Broncos this last week or so. Six weeks ago I felt like I was taking crazy pills.

184 Re: Week 12 DVOA Ratings

The funniest part has been that the crazy talk has been coming from both sides. The Tebow fanatics are over-the-top but some of his detractors are just as bad the other way.

But so far my favorite moment has been on a radio show where the host--a rabid Tebow supporter--suggested that Fox should be fired for being too conservative and running Tebow too much. Meanwhile, I think he deserves huge credit for catering the offense so much to what Tebow does well.

11 GB vs. NYG

Wait, your formula thinks Green Bay only has a 51% chance of beating New York this week? Considering that Vegas has them as -350 favorites, I'm going to assume these calculations are pretty valueless.

12 Re: GB vs. NYG

In reply to by Scizzy (not verified)

Vegas lines simply reflect the amount of money being bet on each team by the populace. If you dropped $1 billion on the Giants (barring the red tape involved), they would suddenly become the overwhelming favorite.

26 Re: GB vs. NYG

In reply to by nibiyabi

I'm getting fed up of seeing this argument written. The Vegas lines represent what odds compilers believe the correct probabilities are. Lines are not created or moved according to where muggish, populace money is bet.

A good bookmaker looks at 'who' is betting, not how much is being bet. If you think that shrewd, professional punters are betting on your line, you would want to move it. But these people do not represent the populace.

29 Re: GB vs. NYG

In reply to by BJR

Odds makers usually set the line so that an equal amount of money is bet on both sides to guarantee that they make money.

49 Re: GB vs. NYG

In reply to by tuluse

No. Oddsmakers put out the price they think represents the correct probability on each team winning. If that results in an equal amount of money being bet on each team then great. If not, then they run a liability. But so what? As long as it is just regular, mug punters who they are better informed than, they will always win money in the long run. However, if they recognise punters betting that are shrewd and sharp, and will win money in the long run, then they would move a price.

It's all about the 'who', not the volume.

83 Re: GB vs. NYG

In reply to by BJR

This is true of some bookmakers, but others aim to reduce variance, and most current Vegas bookmakers lean towards the latter position. They generally do not want to tell the casino's CFO (who, remember is likely to be used to table games and slots where the sheer number of trials means that daily/weekly variance is basically nil) that the wrong sides won and the loss wiped out blackjack's profit for the week.

Dedicated bookmakers (e.g. offshore) are much more likely to lean to the former position, though even there, it's difficult to say. Pinnacle, for instance, has such a history of opacity that it's difficult to tell whether their odds are based on their opinion or on balancing action: talk to 50 people who've played there or watches their moves and you're likely to get 50 different answers. My best guess is that, judging by how Pinnacle reputedly recruits (mostly financial (especiallly option) traders) and how they're reputedly structured (the trader on a given game is risking his own capital alongside Pinnacle's) is that it varies from market to market and trader to trader.

The who is important, though: any bookmaker worth his salt profiles his clientele and might, for instance, consider a $1000 bet from a sharp to be the same as $15k in aggregate square money when it comes to deciding to move the line.

As a philosophical point, though, it's not necessarily a great thing for a bookmaker who handicaps better than other bookmakers to noticeably shade the line even most of the time: sharp bettors pick up on this and simply bet the bookmaker's lean elsewhere at better prices. End result is that the bookmaker doesn't get as much action as he could by sticking closer to the herd and even given a dramatically higher proportion of square action, is likely to end up with less overall profit.

115 Re: GB vs. NYG

In reply to by BJR

"Odds makers usually set the line so that an equal amount of money is bet on both sides to guarantee that they make money."

No. Oddsmakers put out the price they think represents the correct probability on each team winning. If that results in an equal amount of money being bet on each team then great. If not, then they run a liability. But so what? ...

Because they are in business to make money, not run liabilities needlessly = incur losses without reason.

But, Dudes, it doesn't matter. Because 99.9% of the time it is the same thing -- either way, the Vegas line is the best advance estimate of winning and by how much.

If they weren't the same thing, then either way it would be easy for wagerers who aren't dufuses to beat the Vegas line -- and it ain't.

This has been documented a thousand times over, supercomputers have a difficult time staying even with they Vegas line, and *nobody* beats it systematically, it is the best estimate there is. (Any method of beating it systematically would immediately be adopted by those who set it -- end of that success right there.)

In reality the book usually sets a line estimated to draw balanced action, then moves it as wagers come in to keep the balance. That's why spreads move during the week. Because they make money from from the vig -- balance attracts more wagers coming in, thus more vig for them, and leaves them with no *risk of loss* because they don't have either side of the bet.

Who the heck isn't happy with income coming in with no risk of loss? It's as good as running a casino. Often, it *is* running a casino. Why would they take a side on the bet for no reason and risk a big loss?? They aren't setting spreads to show how well they can predict football games, they are doing it to make money.

But sometimes they *do* have a reason to take a side, and do so. Often on "special event" action, much less so for routine games. Then the wagering is unbalanced and they take a risk. But for a business reason. Vegas is reputed to have lost money on the wagering on the last few Super Bowls -- but you can bet, so to speak, that it made it back in other ways.

So you are both right.

But either way, the Vegas line is the best estimate of future results, period. Either way. Whatever is setting it at the moment. That's all anyone needs to know.

311 Re: GB vs. NYG

In reply to by Jim Glass (not verified)

That's an excellent summation. To reinforce just how hard it is to beat the house, an astonishing number of bettor advisory services have winning percentages in the 20-30% range (which doesn't stop them from charging a thousand bucks for a year''s subscription). Also, I think that's another thing that flies in the face of the notion that the Packers aren't really good enough to be 11-0. Vegas, unlike FO, liked Green Bay a lot from the beginning of the season. Plus, the Packers are one of very few teams (and maybe the only one this year) popular enough to force bookmakers to inflate the line routinely, yet I believe they're 8-2-1 vs. the spread.

150 Re: GB vs. NYG

In reply to by BJR

I'm a little dubious on your arguement. It's not as if the casino writes down your name when you make a bet - how do they know if you're a "sharp" or a knucklehead like me?

153 Re: GB vs. NYG

In reply to by Kevin from Philly

I would think that Vegas books know who the "sharps" are. They're the ones betting large sums. Bookies would have to be stupid (and judging by the money they take in, they're not) to fail to note that information.

EDIT: In a sense, sharps are identified by the amount they're betting, not by their name and address. If all the large bets a book takes are on one side of the line, they'll adjust the line so that side is less favorable, the "public" bettors be damned. Remember, the goal of casinos is to get the same amount of money bet on each side of the line, not the same number of bets.

159 Re: GB vs. NYG

In reply to by Eddo

Maybe the casinoes would notice large bets, but if I'm betting REAL big, wouldn't I be more likely to bet with one of my Itallian friends, who I don't have to pay up front?

183 Re: GB vs. NYG

In reply to by Kevin from Philly

And against whom you have no recourse if they don't pay you.

224 Re: GB vs. NYG

In reply to by dryheat

Well, you take the same risks with offshore web sites run by "legitimate" corporations too, no? Besides, if I'm that sharp, the casino would eventually bar me, which is the equivalent of not getting paid on future bets.

Unfortunately, I don't suppose I'll ever be considered a threat to any sports books' finances, but it's fun to think about.

213 Re: GB vs. NYG

In reply to by Eddo

I would argue that sharps are identified by the timing of their bets. Whether they are getting the best number possible during the week.

Or so I have heard/read regarding the people who have been banned from certain online books.

216 Re: GB vs. NYG

In reply to by Eddo

If the 'sharps' were sharp, they wouldn't be betting against a spread that is designed to split the betting pool evenly. When the spread is successful, a chimp could do as well as a 'sharp.' And the spread IS successful.

50 Re: GB vs. NYG

In reply to by BJR

Why would they put all the resources, time, and effort into the possibility of making more money (while risking a big loss, i.e., gambling) when they can simply run software to make sure the money bet on each side is equal to therefore guarantee a huge profit on every single game, every single week?

204 Re: GB vs. NYG

In reply to by nibiyabi

I will rephrase your question: "Why would they want to make even more money than their competition?"

31 Re: GB vs. NYG

In reply to by nibiyabi

Your comment implies Vegas does not set a line until significant money is bet on the game. This is, of course, utterly untrue. The Vegas odds makers have a very sophisticated box of statistical tools at their disposal that predict both the game outcome and betting action. The opening line is set thusly.

Changes in the line reflect betting action. Bookmakers care not a whit about outcomes, only about the division of betting.

That Vegas and FO differ on a single game means next to nothing. The sample size is the smallest possible value. FO does not anticipate betting action. DVOA is a very flawed predictor of wins and losses. That does not make it worthless.

53 Re: GB vs. NYG

In reply to by BroncosGuyAgain

This assumes that Vegas tries to set the line at the 'correct' spot. Not true. Vegas doesn't care if the odds reflect an even chance for each team to cover.

If Vegas feels that a particular team will be more popular among bettors (higher national profile, etc.) they will often set a line to entice betting on the other team. Vegas is only interested in the line that will bring in the most money on both sides.

69 Re: GB vs. NYG

In reply to by Scott P. (not verified)

That still requires knowledge of what the actual win probability is. But it also requires knowledge about betting trends for less predictive factors such as the teams playing and win loss records.

Even if the goal of Vegas is simply to split the money bet down the middle for both teams it still requires the initial line to be set based on some fairly solid win probabilities, because actual bettors, at least in the aggregate, have a pretty decent idea of who will win each game.

96 Re: GB vs. NYG

In reply to by Arkaein

Most odds services with which I'm familiar with send out two lines to their clients: an expected result line and a line based on expected betting patterns.

Originating books then take those into account and open at generally low limits, reflecting their lack of confidence in the lines. Because generally people betting openers are more sharp than square (e.g. they understand that the lines are necessarily more opinionated at this time), this betting action is watched especially closely and moves are frequent (of course, a lot of this action is merely trying to set up scalp and middle opportunities for later in the week, so it's not based on an opinion of true probabilities but on a prediction of where the line will go... the analogies with value investment and technical trading are apt).

Another factor is possibly future exposure, which can affect the odds in individual games. Apocryphally, (this comes from Manteris's book) in order to get a convention group from Minneapolis to stay in the hotel (I think it was the Hilton) in the spring of 1987, the execs decided to give everybody a free $100 bet at the sportsbook. A large number of these were placed on the Twins to win the World Series and the result was a lot of exposure that could not otherwise be reduced on the Twins winning the Series. As the Twins (who were not considered a particularly strong team heading in and so had fairly long odds offered) got into the AL West race and then made the playoffs, the name of the game for Manteris became attracting as much action against the Twins on a game-by-game basis to hedge the futures risk. Even though the Twins won the World Series (and the net loss on the futures was quite large), they soaked up enough Cardinals action on the games that the Twins won to break even overall. I can see that possibly being a factor here: the Packers are a public team there may well be decent net exposure on Packers to win the Super Bowl and/or NFC. Since a loss to the Giants, ceteris paribus, should reduce the chances of paying out on those futures (e.g. one way the Giants can win is by knocking Rodgers out...), taking a loss from offering +EV Giants ML bets might be offset by the decline in mark-to-market future exposure to the Packers.

97 Re: GB vs. NYG

In reply to by leviramsey (not verified)

Truth is, I think everyone's missing the boat on the GB-NYG game. The current line, i believe is 7-- and I think that is LOW. 8.5 or 9 is probably a truer reflection of the two teams' ability now, even granting the Giants a home field edge. 8.5 or 9 is so far froma 51-49 probability as to be a joke.

104 Re: GB vs. NYG

In reply to by Paul M (not verified)

I'm not going to join the argument about how bookmakers set the line, but I DO think it would be very interesting to see FO's estimated chances of each team winning, every week.

214 Re: GB vs. NYG

In reply to by Tom W (not verified)

That is part of the premium service.

228 Re: GB vs. NYG

In reply to by jebmak

No it's not (unless it's somewhere I haven't found). They give their picks vs. the spread and straight up, and they rank them according to confidence, but they don't give the actual percentages.

But they do give results on their picks vs. the spread since 2008, and they've been over .500 every year. And their higher-confidence picks this year are far over .500 vs. the spread (data isn't broken out for prior years). How realistic their listed spreads are (whether you could actually bet those lines), I don't know.

261 Re: GB vs. NYG

In reply to by Arkaein

If the line were set on realistic win probabilities, then why would the line shift depending on the wagers? We all know that the initial line set by Vegas is not the line that will show up on game day!

309 Re: GB vs. NYG

In reply to by LionInAZ (not verified)

Presumably for the reasons others have stated: to keep the money on both sides even.

As long as the betting public has a bias, the money bet will be lopsided across the betting line.

Vegas has to set a reasonable line to start with, otherwise smart bettors will take advantage of this line to make easy money. However, if Vegas wants to hedge its bets by balancing both sides, then it will have to adapt to the betting public. Although I'm not well versed in the specifics, the actual strategy seems to find a place in the middle, to keep the line relatively balanced while not allowing too much edge to smart bettors.

Also consider that conditions change after the initial line is set. A player getting hurt in practice mid week will change both the actual and publicly perceived win probabilities.

13 Re: GB vs. NYG

In reply to by Scizzy (not verified)

Seems odd for the Colts percentages too. I realize that home field is a definite factor but the Colts have a 38% chance of beating 1st ranked Houston with only an 18% chance of beating 23rd ranked Jacksonville?

19 Re: GB vs. NYG

In reply to by Mr. Show (not verified)

The Colts play the Texans at home while they play the Jaguars in Jacksonville.

64 Re: GB vs. NYG

In reply to by RickD

Plus the Colts are playing with the same second string QB they've had all year while the Texans are now have either a fifth round rookie, the recently signed Kellen Clemens, or (hide your children) Jake Delhomme!

21 Re: GB vs. NYG

In reply to by Scizzy (not verified)

It's a home game for the Giants. And really, the Giants are not a bad team. And their strength (passing game) matches up well to GB's weakness (pass defense).

I'm sure GB would be favored more often on a neutral field.

I don't think you appreciate the difference between how Vegas sets lines and what's going on at this site.

23 Re: GB vs. NYG

In reply to by RickD

GB is a terrible matchup for NYG. The Giants secondary regularly gets torched (heck even Charlie Whitehurst has torched them) and have a QB that historically is Turnover Prone vs a GB Defense that is Great at Intercepting Passes. Not surprisingly, these factors played out exactly as one would suspect during Week 16 Last Year.

NYG is slow in the secondary and has no playmakers outside of their Front 4 on Defense. Good luck trying to matchup against GB's Receivers with their 3 Safety Defense

32 Re: GB vs. NYG

In reply to by QQ (not verified)

The Giants beat the Pats and the Pats are constructed in a similar manner to the Packers - explosive offense and weak defense.

I think people are suffering from Most Recent Game bias.

42 Re: GB vs. NYG

In reply to by RickD

NYG was getting torched by Good Offenses long before The Saints game. Heck even terrible Charlie WhiteHurst/Tarvaris Jackson killed them. NYG can't cover anyone and QBs that can get rid of the ball have a field day.

This game will be a less extreme version of the Week 16 Game Last Year. I see a 38-27 type win for GB

167 Re: GB vs. NYG

In reply to by QQ (not verified)

The Pats have the #2 offense in the NFL and yet the Giants were able to hold them in check. That was my point.

The game against the Seahawks was an anomaly. You cannot judge the Giants by that game alone. If they played that poorly every week Tom Coughlin would have died of a stroke already.

58 Re: GB vs. NYG

In reply to by RickD

This is a misconception. It's true that the Packers and Patriots share the common attributes of an elite offense and suspect defense, but they are not "constructed in a similar manner" at all.

The Patriots rely on mismatches at Tight End with the crafty Welker mixing it up with the short game. There's not a diversity of ways they can beat you offensively. The Packers, on the other hand, are quite possibly the most diverse offense in the NFL. They have a wealth of guys that can move the ball on you. The Packers offense is deeper than the Patriots.

Also, while they both bleed yards, the Packers D has more teeth. They've got a number of playmakers on that side of the ball.

134 Re: GB vs. NYG

In reply to by Mavajo (not verified)

While the Patriots employ the two TE set more frequently and the Packers don't, I don't think that really means they are that different. Both teams use the pass to set up the run and both teams try to exploit mismatches. The Patriots work the inside of the field more and the Packers work the outside more. That doesn't mean they are dramatically different. Does that difference somehow negate the strength of the NYG D, its pass rush? I don't think so but who knows.

As for the Packers having more "playmakers" on D, apparently not. The reality is that everyone has been gashing them this year for whatever reason and Eli is a pretty good QB. Giants have a decent shot at the upset, although not likely

158 Re: GB vs. NYG

In reply to by BSR

Yeah, they've been gashed so much that they are #3 in the NFC in points given up. The most overrated defensive stat is yardage given up. The Lions had a bunch of yards last Thursday in the first half but didn't get inside the red zone. And yes, the Packers do have more "playmakers" on D. They perennial lead the league or are at the top of the league in interceptions-- they have playmakers on D where the Patriots don't. No one can argue that. Eli has a tendency to throw interceptions. It's not a good match up as last years 45-17 score can attest.

166 Re: GB vs. NYG

In reply to by Abe (not verified)

...and the Pats have given up even fewer points than the Packers have (223 to 227).

So both defenses tend to give up lots of yardage without giving up many points.

171 Re: GB vs. NYG

In reply to by Abe (not verified)

Yeah, they've been gashed so much that they are #3 in the NFC in points given up.

According to whom? NFL.com has them joint 4th in the NFC, tied with Atlanta and behind Dallas, Washington, and San Francisco.

I also note that you specify NFC, which conveniently omits the fact the AFC has 8 of the top 10 (and 9 of the top 11) defenses by points allowed. Green Bay is joint 13th. (As for the entire Patriots/Packers aspect, the Patriots and their supposedly awful "no playmaker" defense have allowed fewer points than the Packers. Both give up a lot of yards, but are middle of the pack in points allowed.)

143 Re: GB vs. NYG

In reply to by RickD

The problem is that I think the Packers more resemble the Saints than the Patriots.

24 Re: GB vs. NYG

In reply to by RickD

Trust me, if Vegas was routinely this far off on a game (literally 75% vs. 51%!) there would be no such thing as Vegas because every sportsbook would be broke a million times over. Trust me, the way Vegas sets lines, which are very difficult to beat on games which have immense amount of actions, is far more rigorous then whatever spit out this result. Honestly, if you think the Packers are really going to lose this game half the time, you should rush to your local bookie because you'll make a fortune.

33 Re: GB vs. NYG

In reply to by Scizzy (not verified)

If you think the Packers are really going to lose this game half the time, you should rush to your local bookie because you'll make a fortune.

Why? How many times are they going to play this Packers-Giants game?

40 Re: GB vs. NYG

In reply to by RickD

If the playoff odds simulator is actually correct in the odds listed here, then there's a strong pattern of Vegas underestimating the chances of underdogs winning straight-up. So you'd actually have many, many chances to make a fortune.

186 Re: GB vs. NYG

In reply to by RickD

It's the same point. You just didn't get what the first guy was saying.

43 Re: GB vs. NYG

In reply to by RickD

Oh, the rest of those percentages are similarly terrible. Though you're right about the sample size, of course.

39 Re: GB vs. NYG

In reply to by Scizzy (not verified)

Quite seriously, what you're doing is making an argument to authority. It's not like the method that Vegas has is more rigorous than that FO is doing. Vegas is simply trying to maximize profits, which is a far different goal than maximizing the probability of being correct.

It's quite possible that a poor estimate of the win probability will bring a higher expected return for a given casino than the most accurate estimate would be. Vegas wants to have as much action as possible while minimizing risk. These are two different goals, and neither of them equals "finding the optimal estimator for win probability."

I wouldn't be surprised either way at what the outcome in NJ is next week. Would I put the Packers' probability of winning at 51%? Probably not. Accuscore has it at 70%, which seems more reasonable. But I wouldn't argue against FO simply by an argument to authority.

54 Re: GB vs. NYG

In reply to by RickD

I guess you could say I'm making an argument to authority, but it's an authority that has proven its legitimacy over literally millions of trials. If those percentages are right that would imply that people could beat Vegas for huge returns on their investment in ways that would be observable to onlookers. Since nobody is that good at making money betting on sports, I think it's safe to say that FO's numbers, which are literally untested and originate in a black box, are wrong on this one.

61 Re: GB vs. NYG

In reply to by Scizzy (not verified)

It's not untested. Every year there are sites that compare how various advanced stats do versus Vegas. Here's one right here:


75 Re: GB vs. NYG

In reply to by bravehoptoad

Interesting, but in terms of the argument at hand (whether Vegas or FO has a superior formula) we would need to see initial (rather than final) Vegas lines. I'm assuming these are final lines which were skewed by betting patterns.

55 Re: GB vs. NYG

In reply to by RickD

It's not an argument from authority - it's the efficient market hypothesis. Vegas oddsmakers are actually staking their own money on their odds being unbeatable (or mostly unbeatable.) Since Vegas doesn't go bankrupt, either their odds are unbeatable, or nobody has noticed any areas of obvious, glaring bias.

73 Re: GB vs. NYG

In reply to by Yaguar

I wouldn't say "nobody". Some people may make money on betting at football. A more appropriate statement, would be "not enough" people know how to game the system to make a dent in the vast masses that lose money.

121 Re: GB vs. NYG

In reply to by Yaguar

Note as well that any *systematic* method of beating the Vegas line will of course immediately be adopted by Vegas in setting the line -- end of that story right there.

That puts the kibbosh on Football Outsiders or ANY publicly known method of computing wagers and setting odds beating the Vegas line right there. Ain't happening, ain't gonna happen (for longer than it takes a gnat to blink).*

So if you figure out a way to do it, keep it secret to yourself and be careful not to win enough to draw attention to yourself.

* Of course, the same logic debunks all those who claim they can tell you a way to "beat the stock market", or any other public market that professionals trade in, for real estate, commodities, forex, whatever. If they can beat the market, why the hell aren't they making a fortune beating the market instead of wasting their time talking to you about it? (And if they are offering to *sell to you* their secret of beating the market, then it is pretty obvious they think they can make more money from their "secret" via sales-to-rubes than from market-beating.)

187 Re: GB vs. NYG

In reply to by Jim Glass (not verified)

100% correct. Whenever anyone comes seeking you out bragging about how they can beat Vegas or the stock market, you should run because they're either a thief or a moron. No doubt there have been people who have done both, at least for periods of time, but they were smart enough not to publicly announce their discovery and thus instantly doom it.

If FO had any confidence in their system, they wouldn't even be admitting that it existed. And if the thing had proven to work, this site would be long gone because they'd be gambling full-time and rolling around in their winnings.

192 Re: GB vs. NYG

In reply to by markus (not verified)

Unless, maybe just maybe, their system isn't intended to "beat Vegas". Perhaps they're just genuinely interested in why teams win football games, and in detailed analysis of what makes football teams tick. Perhaps, horror of horrors, they actually enjoy the sport of football WITHOUT needing to gamble on it.

Crazy talk, I know.

209 Re: GB vs. NYG

In reply to by markus (not verified)

Or perhaps the amount of capital they would need in order to survive the variance of the results is in excess of what they currently have or can raise.

172 Re: GB vs. NYG

In reply to by Yaguar

The efficient markets hypothesis is like the theory of evolution: it only says that the fittest have survived, it doesn't make any predictions about the future, except to say that in the future, the fittest will survive.

As fallacies go, the argument from authority isn't terrible, especially when relating to the field of expertise of the authority, but it's still an argument that has its weaknesses. Basically it stays true for a long time until it's not true any more.

290 Re: GB vs. NYG

In reply to by RickD

Except the efficient markets hypothesis is wrong on the basis of the investor hypothesis, the markets can stay irrational longer than you can stay solvent.

82 Re: GB vs. NYG

In reply to by Scizzy (not verified)

I, for one, will be surprised if the Giants do not at least cover.

328 Re: GB vs. NYG

In reply to by Scizzy (not verified)

I guess you skipped the above discussion. Vegas makes money as long as they equalize the bets on each side, regardless of how their lines differ from actual likelihood of outcome.

14 Patriots

This has been a very interesting DVOA season, seemingly with more "doesn't jibe with my eyes" ratings and weekly reaction to games.

Take this week, for instance. NE takes a team that had a respectible 10% DVOA on the season behind the woodshed and their rating improves marginally. The Jets struggle against a Buffalo team in freefall (both on the field and in DVOA) yet their total DVOA remains largely unchanged despite being in the same general vicinity as NE's to start the week.

In fairness, I don't believe the Young-led Eagles were any great shakes, but DVOA thought they were good and still didn't react to a total trouncing. Hell, Houston couldn't pull away from a Jax team that DVOA thinks is terrible and their total DVOA rose by 0.1% more than NE's did!

Another comparable is the NO/NYG game. Sure NY was rated a little higher (roughly 8% than Philly) and perhaps the performance was a little better, but NO leaped up by 5% after Mondays' game, compared to 1.1% for NE. Yes NE allowed a lot of yardage, but NO allowed only a single yard less than and NY also scored 4 more points than the Eagles.

Just a head scratching season from me.

15 Re: Patriots

In reply to by Anonymous1 (not verified)

A head scratching season FOR me. I had nothing to do with it's creating, either on the field or in the DVOA spreadsheets.

20 Re: Patriots

In reply to by Anonymous1 (not verified)

DVOA has no idea that Vick and Schaub are injured.

37 Re: Patriots

In reply to by Karl Cuba

I know, but that has very minor relevence to the discussion at hand. If anything, that should actually *help* NE's rating, since NE's defense would have allowed that much yardage to a better quarterback. Contrastingly, it should *hurt* the Texans' due to the inferior QB play.

52 Re: Patriots

In reply to by Anonymous1 (not verified)

Beyond a certain point, you have to shrug and say "DVOA likes the Jets."

We know DVOA doesn't like long drives against a defense, even if they result in zero points. The Eagles had a 78-yard drive against the Pats that resulted in zero points.

DVOA isn't the only measure that dislikes the Pats' defense. Their yardage/game is still horrible. But they are excellent w.r.t. points allowed. Well, "excellent" is probably an overstatement, but they are in the top 10. I think this is fluky statistically.

What concerns me more is that the Jets seem to bump past a lot of teams (including the Ravens, who pasted them) based on the strength of their special teams. Are special teams really that important? It's true that the Jets have a good return game, and you could argue "yardage is yardage," but return yardage necessarily comes before an offensive drive that itself needs to score points (unless you have Hester or Peterson scoring TDs).

The sad part of this is how it's all going to resolve itself. The Jets will get the #6 seed, go to Houston, and then beat the Texans with Sage Rosenfels or Vinny Testaverde or Dan Pastorini at QB. And this fluke result will validate what otherwise smells like an overrating of the Jets.

71 Re: Patriots

In reply to by RickD

I like the sound of that Jets scenario! However, given the tie breakers, they're only going to make the playoffs if they win at least 4 of the last 5 games (and they probably need all 5). They would have to play pretty well to pull that off, which would largely validate their current DVOA. They could also play poorly and lose to the Giants, Iggles and Dolphins, causing their DVOA to plummet and solving the problem a different way.

137 Re: Patriots

In reply to by RickD

I understand the Jet issues, which is why I brought up Houston and NO as well. The lack of consistency between their respective responses to this past week is staggering.

175 Re: Patriots

In reply to by RickD

"I think this is fluky statistically."

The Patriots have been ranked better in Points Allowed than Yards allowed almost every year that Bellichick has been there. It may not totally be the defense, but its certainly not a fluke.

180 Re: Patriots

In reply to by RickD

But they are excellent w.r.t. points allowed. Well, "excellent" is probably an overstatement, but they are in the top 10.

They're 11th.

242 Re: Patriots

In reply to by Andrew Potter

And 16th in points per drive, which is a much better measure.

151 Re: Patriots

In reply to by Anonymous1 (not verified)

Perhaps you weren't paying much attention to the game. The Saints rushing attack was stellar. And they did score two more touchdowns. And they didn't go down 10 points at the beginning of the game.

160 Re: Patriots

In reply to by milo

Yes, as I said before, they played a little better. But DVOA really doesn't care about falling behind or whether the success was due to running or passing.

162 Re: Patriots

In reply to by Anonymous1 (not verified)

DVOA cares a great deal about success in the running game. NO had a better running DVOA in this game than CAR(#1) or PHI(#2) have put up in any single game this year.

207 Re: Patriots

In reply to by milo

You aren't getting my point. A success is a success is a success, it doesn't matter whether it is running or passing.

It has been well documented that DVOA prefers an 80 yard drive of ten 8 yard plays over two 40 yarders. I don't see anything indicating that a DVOA is more impressed by ten 8 yard runs than they are by 10 8 yard passes. It isn't like they grade out running and passing equally regardless of how many attempts were made by each.

227 Re: Patriots

In reply to by Anonymous1 (not verified)

It has been well documented that DVOA prefers an 80 yard drive of ten 8 yard plays over two 40 yarders.

No, it hasn't, because it's not true, or true only to a very slight extent.

What DVOA does prefer is an 80 yard drive of ten 8 yard plays over two 40 yard plays and eight 0 yard plays.

231 Re: Patriots

In reply to by bravehoptoad

Pedantic. And still not relevent to the primary reason for my original post.

265 Re: Patriots

In reply to by Anonymous1 (not verified)

Well, as long as your original post doesn't rely on "well documented" fictions, then I have no trouble with it.

267 Re: Patriots

In reply to by bravehoptoad

Yes, I misstated that (though the intended comment was clearly as day to everyone but you, and below you'll also see my error was unintendedly correct as well). I hope you get exquisite joy out of pointing out trivial flaws while offering nothing of substance.

269 Re: Patriots

In reply to by Anonymous1 (not verified)

I'm with bravehoptoad on this one. It's not a trivial flaw. It's one of the most commonly repeated misconceptions across all the comment threads on this site, and is so commonly misstated that it's become erroneously accepted by many as fact. Given how frequently misunderstood and misstated it is, it's well worth correcting at every opportunity.

317 Re: Patriots

In reply to by Andrew Potter

But it is trivial when it has no bearing on the conversation at hand. Both NO and NE had their share of big plays and both had their share of long, sustained drives. It isn't like the one team had 7 three and outs and 5 80 yard TDs or anything. So, yes, in this instance they were trivial.

Beyond that, BHT's later comments were snarky and unnecessary. Had he simply said, "I feel I should point out that this statement is actually incorrect..." with a less arrogant delivery, the conversation would have gone in a different direction.

331 Re: Patriots

In reply to by Anonymous1 (not verified)

I didn't think it was either trivial or pedantic. In any case, there's a far, far more intelligent conversation taking place on this very topic just a few posts down.

238 Re: Patriots

In reply to by bravehoptoad

DVOA DOES fundamentally prefer an ten different 8 yard gains over 2 40 yard plays. Not only does 8 successful plays to more to offset negative plays later in the game (for example look at the Jets high DVOA vs NE thanks to super long drives) but also the long drives accumulate the Red ZOne Bonuses that 2 different 40 yard plays would not gain.

266 Re: Patriots

In reply to by QQ (not verified)

Not only does 8 successful plays to more to offset negative plays later in the game....

That's exactly what I'm disputing. There's no evidence for it. DVOA likes big gains. DVOA likes small gains, as long as they're successful. DVOA likes big gains more than it likes small gains. What DVOA doesn't like is unsuccessful plays.

Not even DVOA's creator knows why the Jets had such a high DVOA for the Jets-Patriots game. You don't know why, either.

...but also the long drives accumulate the Red ZOne Bonuses that 2 different 40 yard plays would not gain.

True. It's just not going to be that important.

268 Re: Patriots

In reply to by bravehoptoad

The creator does pretty much why the Jets had such a high rating. Their Offense had basically as many if not more Successful plays than Usuccessful since their few long drives offset all the negative plays from the 3 and Outs.

It also has been stated many times that 8 10 yard gains is worth more than 1 80 yard gain since DVOA sees the 8 10 yard games as more predictive/repeatable

273 Re: Patriots

In reply to by QQ (not verified)

It also has been stated many times that 8 10 yard gains is worth more than 1 80 yard gain since DVOA sees the 8 10 yard games as more predictive/repeatable

This is wrong. The 80 yard play will have higher DVOA than any 10 yard play.

However, if the 80 yard gain team has 7 0 yard gains then it will look bad.

284 Re: Patriots

In reply to by tuluse


DVOA considers *1* 80 yard play worth more than *1* 10 yard play (any other result would be madness)

But it does not consider *1* 80 yard play equivalent to *8* 10 yard plays.

Basically meaning the NYJ could finish with a higher DVOA from a series of grinding 8 play drives, and finish with a higher DVOA than the Packers, who converted a Hail Mary on every play, despite half the points, half the total yards, and 1/8th the yards per play -- all at the same success rate. This is not only non-intuitive, it's damned near non-defensible.

287 Re: Patriots

In reply to by Aaron Brooks G…

If the Packers converted a Hail Mary on every play, they'd have a much higher VOA than a Jets team that was taking eight plays per drive. (Opponent adjustments don't matter here.) If the Packers tried Hail Marys on every play and converted 1/8 of them, that would probably result in a worse VOA than the Jets' successful 8-play drives. This makes sense, or at least much sense as this extreme example can, since that 1/8 successful Hail Mary strategy will produce TDs on 41.4% of 4-down series, while the grinding strategy is producing TDs on 100% of series.

288 Re: Patriots

In reply to by Aaron Brooks G…

Nevermind, Jerry's explanation was better.

291 Re: Patriots

In reply to by tuluse

I used to agree with you on this, tuluse - I thought the only reason a boom-or-bust team would do worse than a team that steadily marched down the field was because of those busts. But based on what I saw about how DVOA is calculated in the Jets thread I think there might be more to this criticism of DVOA than I thought. 8 10-yard plays might actually help a team's DVOA more than 1 80-yard play.

Let's say that Team A and Team B play identical games, except for one 80-yard scoring drive. Team A has eight plays for 10 yards each and a touchdown on that drive, and Team B has just one play for an 80 yard touchdown. Other than that one drive their games are play-for-play identical. Which team has a higher DVOA for the game?

I think the answer is Team A. These numbers aren't exactly right, but based on the data in the Jets thread and on the methods page, on 1st and 10 it looks like that the baseline is about 0.8 success points, a 10-yard gain is worth about 2 success points, and an 80-yard touchdown is worth about 6 success points (I'm least sure of this last number). For that drive, Team A had a total of 16 success points and a total baseline of 6.4, and Team B had a total of 6 success points with a total baseline of 0.8. DVOA is points over baseline divided by baseline, so, just looking at that one drive, Team B obviously has a higher DVOA - 650% (5.2/.8) vs. 150% (9.6/6.4) for Team A.

But let's look at the whole game. Say that they both played average the rest of the game, 40 success points with a baseline of 40. Then for the full game Team A had 56 success points out of a baseline of 46.4, for a DVOA of 20.7%. Team B had 46 success points of a a baseline of 40.8, for a DVOA of 12.7%. So the 8-play drive does do more to help the team's offensive DVOA than the 80-yard touchdown, even without having any extra 0-yard plays. (The 80-yard touchdown would have to be worth over 9 success points for the two teams' DVOAs to be equal.)

This is just a rough calculation and I left out a couple things, but I think a more thorough calculation would tilt things even more in favor of Team A. I didn't give them a red zone bonus (which would increase the value over baseline of 2 of their plays by 25%, without changing the baseline), and I didn't count their touchdown play as being worth more than any other 10-yard gain (it looks like a 10-yard touchdown is worth over 4 success points, with a baseline still near .8, based on the numbers for the Jets' 9-yard touchdown). Once you take those into account, Team A's DVOA for the game would be more like 30% rather than 20%.

294 Re: Patriots

In reply to by Dan

That's interesting. It looks like there is some opportunity cost DVOA-wise for a team that scores really quickly.

However, I think there is basically no way a team which has an 80 yard scoring play doesn't end up running more plays in some other drive, which should eliminate that problem.

Actually, how did you estimate success points for an 80 yard TD play? In the Jets game, they got 5.3 success points for a 21 yard TD. That would mean teams only get .7 points for the other 60 yards. The Jets had 3 other plays from 19-22 yard gains which yielded 2.9-3.75 success points. So if those 60 yards were discounted by 1/3, that should yield about 8.3 success points for an 80 yard TD. Which would give a DVOA of 18.4% in your example. If you look at yards past the first ten, a 20 yard gain gives 1 point for the extra 10 yards. If that continued, an 80 yard TD would yield 11.3 success points. That would give a DVOA of 25.7%. The true answer is probably somewhere in the middle of those two.

Here's the other thing though, the team with 8 yard gains, has 8 players of above average play, the team with the one 80 yard TD has 1. Again it is the *other* plays from the game bringing their average down. By your own calculations an 80 yard TD is worth at least 650% DVOA. The problem is that the team wasn't not above average on any other play.

298 Re: Patriots

In reply to by tuluse

The methods page says:

We then expand upon that basic idea with a more complicated system of "success points." A successful play is worth one point, an unsuccessful play zero points. Extra points are awarded for big plays, gradually increasing to three points for 10 yards, four points for 20 yards, and five points for 40 yards or more.

So the success points don't keep increasing linearly - they level off after 40 yards. The passage makes it sound like a 99-yard play is only worth 1 success point more than a 20-yard play, so since the 21-yard Jets TD was worth 5.3 that suggests an 80-yard TD should be worth about 6.3 success points. The passage goes on to say that there is a bonus for a touchdown play, and looking at the Jets' touchdown plays, it looks like that bonus is around 2 success points, so another estimate is that an 80-yard TD is worth about 7 points (five points for gaining 40+ yards plus two points for the TD). My guess is that the number is somewhere between 6 and 7 success points, which makes only a small difference in my calculations (7 pts. would give Team B a 15.2% DVOA in the example in my previous comment rather than 12.7%).

The key thing in my example is that we're holding constant what the team does on the rest of its offensive plays - we're just replacing an 8-play 80-yard TD drive with a single 80-yard TD pass - and their offensive DVOA is going down. Even if they were above average the rest of the time, this would still happen. Let's say that the team had a 25% offensive DVOA the rest of the game (which would be good for 4th in the NFL this year) instead of a league-average 0% DVOA. They still had a baseline of 40 points for the rest of the game, but they gained 50 success points. Then Team A would have a DVOA of 42.2% for the full game (66 success points, baseline = 46.4) and Team B would have a 37.3-39.7% DVOA (56-57 success points, baseline=40.8). They're closer together, but Team A still does better in DVOA.

And I haven't been giving Team A credit for its touchdown or its red zone bonus in these numbers. I've looked at the numbers some more since my last post, and I think I'm understating Team A's performance by 3.2 success points and about 6% DVOA. It should be about 48.8% in this example, and 27.3% in my original example.

The conclusion of this line of argument is that DVOA is underrating teams that are good at getting big gains, and overrating teams that are bad at getting big gains. The Jets' longest play in their game against the Patriots went for 22 yards and they are last in the league this season in number of 40+ yard plays (they have 1), which I think is a big part of why DVOA likes them more than conventional stats do. Although maybe this is a feature of DVOA that makes it more predictive (even though it has a worse fit to past results), since big plays could be fairly random. I wonder if Aaron has tried varying the success point values to see what makes them most predictive of teams' performance. The numbers that he gives on the methods page are suspiciously round, which makes me think that he just came up with success point values that seemed plausible to him rather than basing them systematically on the data (although there might have been some kind of testing at some point).

300 Re: Patriots

In reply to by Dan

I tend to agree; I particularly tend to agree with your statement that DVOA's tendency to "underrate" teams that get a lot of big gains and "overrate" teams that don't is more of a feature than a bug. It seems reasonable that the ability to consistently string together a long series of successful plays is more predictive of future success than the ability to hit a single long play, if only from the "sample size" point of view. Even a blind Blaine Gabbert finds the occasional Mike Thomas.

332 Re: Patriots

In reply to by DGL

And also in this case, "big gains" is 40+ yards, right? That wouldn't be the usual lower limit when people think of big gains, I don't think.

295 Re: Patriots

In reply to by Dan

Nice work. To take it a step further, let's say that each team has an identical season except for team A scoring one touchdown on an 8-play drive every game while B scores on one play. (We'll assume that this has no effect on the defensive numbers.) We don't expect DVOA to recognize that this is a context where A's 128 plays are actually equivalent to B's 16; it's just going to see that team A has 112 more successful plays in their season than team B, even if B has those 16 huge plays. The exception would be the unlikely case where both offenses are good enough that the 10-yard plays are worse than normal, in which case those drives would bring the DVOA down.

The larger point may be that DVOA works better as the sample gets larger. (Shocking, isn't it?) Single plays, drives, or even games may have results that seem wrong, but as they're accumulated over the course of a season, they become more believable.

ETA: tuluse and I continue to agree.

321 Re: Patriots

In reply to by Dan

I don't want to get too involved, but the ten-80-yard-play TD drive is more valuable than the one-80-yard-play TD drive. You're allowing your defense to rest, and taking up more of the game clock, so your opponent has less time to counter your TD(*).

(*) While in some cases, the latter might not be ideal (if you're down by multiple scores), it is on the whole a slightly-preferred scenario, I would say.

339 Re: Patriots

In reply to by Eddo

This is debatable. Fundamentally, the Better the team, the more important it is to maximize the number of possessions, therefore the 80 yard TDs are better since they lead to more total possessions in the game.

For the weaker team that wants to minimize possessions, the 10 play drive is better, but the better team should prefer the 80 yard strike

233 Re: Patriots

In reply to by Anonymous1 (not verified)

DVOA: 45.5%|67.7%
ODVOA: 65.1%|78.8%
OPDVOA: 136.7%|100.6%
ORDVOA: 0%|78.2%
DDVOA: 24.7%|13.2%
DPDVOA: 42.2%|33.2%
DRDVOA: -15.9%|-23.0%
STDVOA: 5.0%|2.2%

Are one of these ratios not like the other? Which one? The D in DVOA for both opponents is close, so presumably the VOA's in this game compare in similar ratios. I don't know the value of any play (but I can read what is considered successful or not) in terms of VOA or DVOA, but I think I can see that the most significant difference in DVOA performance between NE and NO this past week is Offensive Rushing DVOA. DVOA thinks this was NO's best performance of the season. DVOA thinks this was NE's fourth best performance.

247 Re: Patriots

In reply to by milo

Thank you for posting this. I had little doubt that NO's rushing would be superior, and the magnitude is only marginally surprising.

That said, NE didn't really run that much, so the weight of that 0% shouldn't be too heavy on the overall performance (and the 22% gap seems reasonable, considering DVOA thinks NO's defensive performance was a little better).

That said, I'm still left wondering why NO improved on their entire season DVOA by 4% more than NE on the merits of that single game. 22% better in 1/11 of the season seems like 2% would seem more reasonable (not that it is ever safe to average averages). I also scanned through NO's schedule and it isn't as if any of their prior opponents did great this week making an earlier game appear more impressive in retrospect. Nor do I see the opposite happening with NE.

Whatever, none of this really matters. I guess I'm going to have to accept the divergence between my eyes and the numbers.

16 Re: Week 12 DVOA Ratings

I think the odds of an upset are being overrated. Very, very few people would give KC a better than 3 out of 10 chance of beating GB.