Here are the rankings headed into week 15. Click on the headers to sort. Raw efficiency stats are in the second table below.
| RANK | TEAM | LAST WK | GWP | Opp GWP | O RANK | D RANK |
| 1 |  DEN | 1 | 0.71 | 0.49 | 4 | 3 |
| 2 |  SF | 2 | 0.69 | 0.50 | 3 | 1 |
| 3 |  CAR | 4 | 0.63 | 0.52 | 7 | 7 |
| 4 |  SEA | 5 | 0.62 | 0.51 | 5 | 6 |
| 5 |  HOU | 3 | 0.60 | 0.48 | 8 | 8 |
| 6 |  NE | 8 | 0.58 | 0.50 | 1 | 31 |
| 7 |  NYG | 6 | 0.56 | 0.53 | 6 | 22 |
| 8 |  ATL | 7 | 0.56 | 0.49 | 13 | 20 |
| 9 |  WAS | 9 | 0.55 | 0.51 | 2 | 24 |
| 10 |  GB | 11 | 0.55 | 0.50 | 12 | 13 |
| 11 |  CIN | 10 | 0.54 | 0.47 | 18 | 11 |
| 12 |  DET | 12 | 0.53 | 0.50 | 9 | 18 |
| 13 |  STL | 13 | 0.52 | 0.53 | 14 | 12 |
| 14 |  CHI | 14 | 0.52 | 0.51 | 27 | 2 |
| 15 |  PIT | 16 | 0.51 | 0.48 | 24 | 4 |
| 16 |  DAL | 18 | 0.50 | 0.52 | 15 | 17 |
| 17 |  TB | 15 | 0.50 | 0.51 | 11 | 25 |
| 18 |  MIA | 17 | 0.49 | 0.51 | 21 | 16 |
| 19 |  NYJ | 19 | 0.49 | 0.51 | 25 | 9 |
| 20 |  NO | 22 | 0.47 | 0.53 | 10 | 27 |
| 21 |  BAL | 20 | 0.47 | 0.48 | 17 | 19 |
| 22 |  BUF | 23 | 0.47 | 0.48 | 19 | 21 |
| 23 |  CLE | 24 | 0.45 | 0.47 | 26 | 10 |
| 24 |  SD | 21 | 0.45 | 0.49 | 30 | 14 |
| 25 |  IND | 25 | 0.44 | 0.45 | 16 | 29 |
| 26 |  MIN | 26 | 0.42 | 0.50 | 28 | 15 |
| 27 |  TEN | 28 | 0.41 | 0.49 | 22 | 26 |
| 28 |  PHI | 27 | 0.41 | 0.50 | 23 | 23 |
| 29 |  OAK | 29 | 0.39 | 0.50 | 20 | 28 |
| 30 |  KC | 31 | 0.35 | 0.51 | 29 | 30 |
| 31 |  ARI | 30 | 0.34 | 0.53 | 32 | 5 |
| 32 |  JAC | 32 | 0.29 | 0.49 | 31 | 32 |
| TEAM | OPASS | ORUNSR% | OINT% | OFUM% | DPASS | DRUNSR% | DINT% | PENRATE |
| ARI | 4.5 | 33 | 3.4 | 1.8 | 6.0 | 56 | 4.8 | 0.44 |
| ATL | 7.0 | 36 | 2.7 | 0.6 | 6.6 | 53 | 3.7 | 0.21 |
| BAL | 6.4 | 40 | 2.0 | 0.6 | 6.3 | 56 | 2.7 | 0.50 |
| BUF | 6.0 | 46 | 3.4 | 1.9 | 6.1 | 55 | 2.4 | 0.45 |
| CAR | 7.2 | 43 | 2.5 | 1.3 | 6.1 | 57 | 2.0 | 0.40 |
| CHI | 5.6 | 36 | 3.7 | 1.1 | 5.4 | 57 | 4.5 | 0.41 |
| CIN | 6.5 | 42 | 3.1 | 1.0 | 5.7 | 55 | 2.1 | 0.40 |
| CLE | 5.9 | 39 | 3.2 | 0.6 | 5.9 | 58 | 3.0 | 0.48 |
| DAL | 6.7 | 38 | 3.0 | 1.2 | 6.6 | 59 | 1.5 | 0.46 |
| DEN | 7.3 | 45 | 2.1 | 1.8 | 5.5 | 59 | 3.2 | 0.37 |
| DET | 6.5 | 41 | 2.0 | 1.4 | 6.0 | 56 | 2.2 | 0.47 |
| GB | 6.4 | 41 | 1.8 | 0.7 | 5.8 | 55 | 3.1 | 0.45 |
| HOU | 6.8 | 42 | 2.4 | 0.3 | 5.8 | 61 | 3.0 | 0.39 |
| IND | 6.3 | 44 | 3.4 | 1.2 | 6.7 | 59 | 1.8 | 0.38 |
| JAC | 5.2 | 37 | 2.4 | 1.5 | 7.0 | 53 | 2.2 | 0.46 |
| KC | 5.6 | 42 | 4.2 | 2.3 | 7.3 | 58 | 1.9 | 0.35 |
| MIA | 6.0 | 39 | 2.9 | 1.8 | 6.1 | 61 | 1.8 | 0.38 |
| MIN | 5.2 | 42 | 3.0 | 1.5 | 6.0 | 58 | 1.9 | 0.37 |
| NE | 7.2 | 48 | 1.0 | 0.6 | 7.0 | 54 | 3.1 | 0.38 |
| NO | 6.9 | 43 | 3.4 | 0.8 | 7.5 | 55 | 2.1 | 0.44 |
| NYG | 7.0 | 43 | 2.8 | 0.8 | 7.1 | 51 | 4.6 | 0.31 |
| NYJ | 5.6 | 41 | 3.2 | 2.3 | 5.7 | 54 | 2.6 | 0.36 |
| OAK | 6.4 | 35 | 2.6 | 1.4 | 7.2 | 59 | 2.0 | 0.44 |
| PHI | 5.7 | 46 | 2.4 | 2.5 | 6.9 | 59 | 1.7 | 0.39 |
| PIT | 6.2 | 34 | 2.3 | 2.1 | 5.1 | 56 | 1.7 | 0.51 |
| SD | 5.8 | 37 | 3.2 | 1.3 | 6.2 | 60 | 2.5 | 0.37 |
| SF | 6.7 | 48 | 1.7 | 0.9 | 5.0 | 65 | 2.2 | 0.50 |
| SEA | 6.7 | 44 | 2.6 | 1.2 | 5.4 | 55 | 3.2 | 0.46 |
| STL | 5.9 | 42 | 2.5 | 0.9 | 6.0 | 59 | 2.9 | 0.46 |
| TB | 7.0 | 41 | 1.9 | 0.7 | 7.5 | 62 | 3.3 | 0.43 |
| TEN | 5.9 | 40 | 3.0 | 2.0 | 6.8 | 57 | 2.6 | 0.39 |
| WAS | 7.3 | 47 | 1.7 | 0.9 | 7.0 | 56 | 2.9 | 0.54 |
| Avg | 6.3 | 41 | 2.7 | 1.3 | 6.3 | 57 | 2.7 | 0.42 |
How does this model handle change in GWP over time? Are changes expected to be normally distributed (what parameters?)? Is it handled as a Markov process? Or are all games considered interchangeable, with time-series order ignored? If it's that last one, it might be causing problems for the NE numbers, whose defense is, I think it is safe to say, no longer one of the two worst in the NFL (though it certainly was earlier in the year).
I haven't looked at the methodology in a bit, but if the 49ers have had a tougher schedule than the Broncos and are ranked higher on offense and defense, shouldn't they be ahead of the Broncos? Or do you want to have a lower OGWP rating?
@Alex:
Penalty rate is probably the difference.
Brian, how would you define "best" vs. "most efficient"? If Carolina isn't the 3rd best team, who is and what's the difference? Intangibles?
The Panthers are efficient at losing, apparently.
I've seen this mentioned in a couple of other places, but that was not the first time the Ravens lost two in a row in the Flacco-Harbaugh era.
They lost two in a row twice in 2009.
@Anonymous: The discrepancy in efficiency and record may be due in large part to luck. CAR has had lots of close losses that could have gone either way.
Another possibility Brian has mentioned is coaching. Rivera was previously with SD, which showed a similar discrepancy between efficiency and record.
re: Coaching
Once again NE & BB are clearly outperforming expected win%
(11 straight years). Of course coaching is the difference.
@bigmouth: I understand the difference between most efficient and best record. I was wondering about the distinction between most efficient and best team, as Brian described in the article.
This guy(Brian) must be a Carolina homer. 4-8 is the third best(efficient) team. Carolina played a real team once (the NYG) and got blown out. This team lost to KC and allowed Brady Quinn (thats right the guy Cleveland rejected in favor of Weeden) to have almost a perfect passing day. Their defense collapses almost every week in the 4th quarter. That's not a very efficient time to have your defense collapse.
Real team finish the opponent off. Carolina should not be in the top 20 let alone the top 5.
His efficiency ratings are of HIS DEFINITION. I could say the team with the prettiest uniform is the most efficient at dressing, and therefore is the "most efficient".
If efficiency is not analogous with best, then what is the point?
Although pass yds per attempt(offensive and defensive) has been somewhat successful in predicting a winners in the past, its win rate is way below 66%. And Brian's win probabilities before the game starts is below 60% for the year. Not very predictive. Simply using vegas odds will get you about 66% Straight Up.
So if efficiency is not "the best" and is not "who is gonna win this week" Then why even calculate it?
@JonJ
Brian is definitely not a Carolina homer, that's pretty funny.
I'm curious to know how you calculated that Brian's win probabilities are below 60% or how you can claim vegas odds will get you 66%?
I'm not sure what's the best way to do this but it would be interesting to see how well Brian's predicted game win probabilities correlate to actual win %. For example, in all games that his model predicted 70% win probability did the predicted winner win 70% of the time?
I've used Brian's Eficiancy Ratings as a guide and I'm 2nd overall in my confidence pool.
I have snarked about Carolina's position in this ranking in weeks past but this week went a good way toward vindicating that. I'm surprised the comments do not seem to reflect that big win.
I will tell you what, though: there is not a single team I'd be more afraid to face right now than the Pats. They are a juggernaut, and I don't think this model is fully appreciating that fact. We've been cautioned in the past to pay more attention to the GWP number than how a team lies in the rankings (since there can be a cluster of teams quite close to each other); but by this yardstick I think NE is even more underrated. They are as far behind the Broncos in GWP (0.13) as they are ahead of the Browns and Chargers, and that's just...no.
Another way to look at the list is to see where there are gaps to sort of separate different "classes" of teams. By this measure, the Broncos and 49ers are in a class of their own (and being from different conferences, this should be the most likely SB matchup). There are no other major discontinuities until you get down to the dregs, the three worst teams. But there is quite a scrum of similarly-rated teams starting with the #7 through the #19 Jets. And I would agree that those are all decent but not great teams that are similar in ability to each other.
Finally, I would like to comment on the Vikings. I am a homer, no question about it. I do not have a shred of objectivity when it comes to Minnesota. But I would just put it to fans of other teams: if your team had a must-win game to get into the playoffs, would you really rather they play the Vikings than teams like the Bills, Dolphins, or Jets? C'mon.
Alan,
I'll try to explain how Minnesota is rate so low. but first...
First I'd like to say sorry to Brian for my rather surly post yesterday. I'm a cardinals fan and am still steaming over their meltdown on Sunday.
Secondly, I'd like to give props to Brian and his website.
I do not agree with his rankings or his weekly predictions, but I do really like his articles. Personally, I think the problems in his stat analysis lies in how he weighs the different statistical parameters. Maybe introducing some other statistical parameters (red zone efficiency, third down efficiency, sack percentage) and weighing these parameter less would sand off the rough edges in his statistical power rankings and improve his straight up win percentage.
As I understand it, Brian's efficiency is based on (I believe) the eight statistical parameters below. If this is incorrect please correct my interpretation.
1) Offensive pass yds per attempt(OPass)(adjusted for sack loss)
2) Offensive run success rate(not sure how this calculated)(ORunSR%)
3) Offensive interception rate(the number of picks per 100 pass attempts)(OInT%).
4) Offensive fumbles rate (the number of fumbles, regardless of recovery per 100 run attempts)
5)Defensive Pass yds per attempt (adjusted for sack loss)
6)Defensive run success rate(again not sure how this is calculate)(DRunSR%)
7)Defensive interception rate(the number of picks per 100 pass attempts)(OInT%).
8)Penalty rate (again not sure how this is calculated).
Now Minnesota has the 2nd worst OPass(yds per pass attempt in the NFL). I don't believe Brian made up this stat as it is very close to one you can get the ESPN website. Minnesota has a low defensive interception rate. And an above average offensive interception rate. An above average fumble rate.
So it really come down to the same problems Minnesota has had every year before and after 2009. Bad defensive secondary(Winfield is made of glass and is out 4-10 games a year) and a really sub-par Quarterback. Let's face it Ponder sucks. Until Minnesota improves in those areas significantly..they will always be below average.
Just so you know I've liked Minnesota since 76.
I'd say the jury is still out on Ponder, but he definitely needs to improve.
Winfield "made of glass"? He's consistently rated as one of the best tackling cornerbacks in the league, and showed it on Sunday. The two rookies they drafted have improved the secondary significantly.
Bottom line though is that they are 6-1 at home, beat the Bears and Niners convincingly, and nearly beat the Packers at Lambeau. And of course they have the league's leading rusher despite their opponents orienting their defenses around stopping Peterson. I'm not saying all that combined adds up to their being a top ten team, but they should be somewhere in the middle of the pack, the teens--not down at 26th.
Winfield is a great cornerback. He is just injured a lot.