The equation here is pretty simple. Take a solid defense. Add Peyton Manning. You get a team that's very hard to beat.
Click on the table 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 | 0.76 | 0.52 | 1 | 2 |
2 | HOU | 3 | 0.66 | 0.49 | 7 | 3 |
3 | SF | 2 | 0.66 | 0.50 | 3 | 5 |
4 | SEA | 5 | 0.62 | 0.53 | 16 | 4 |
5 | CAR | 4 | 0.60 | 0.56 | 18 | 7 |
6 | GB | 9 | 0.56 | 0.51 | 9 | 12 |
7 | ATL | 6 | 0.56 | 0.50 | 17 | 21 |
8 | CHI | 11 | 0.56 | 0.51 | 28 | 1 |
9 | NYG | 7 | 0.55 | 0.52 | 13 | 20 |
10 | DAL | 8 | 0.55 | 0.54 | 8 | 16 |
11 | DET | 10 | 0.55 | 0.49 | 5 | 17 |
12 | STL | 15 | 0.53 | 0.54 | 14 | 14 |
13 | TB | 16 | 0.52 | 0.47 | 4 | 26 |
14 | CIN | 19 | 0.52 | 0.48 | 15 | 19 |
15 | NE | 12 | 0.51 | 0.50 | 2 | 30 |
16 | PIT | 14 | 0.49 | 0.48 | 22 | 8 |
17 | BAL | 22 | 0.48 | 0.48 | 12 | 23 |
18 | WAS | 18 | 0.48 | 0.52 | 6 | 24 |
19 | MIA | 13 | 0.47 | 0.47 | 21 | 18 |
20 | PHI | 20 | 0.47 | 0.49 | 23 | 13 |
21 | SD | 25 | 0.46 | 0.46 | 25 | 15 |
22 | MIN | 23 | 0.46 | 0.49 | 27 | 9 |
23 | NYJ | 17 | 0.45 | 0.52 | 26 | 11 |
24 | NO | 28 | 0.44 | 0.52 | 10 | 31 |
25 | OAK | 21 | 0.42 | 0.48 | 24 | 22 |
26 | CLE | 27 | 0.42 | 0.48 | 29 | 10 |
27 | IND | 24 | 0.42 | 0.43 | 11 | 32 |
28 | ARI | 26 | 0.42 | 0.52 | 32 | 6 |
29 | BUF | 29 | 0.42 | 0.48 | 19 | 29 |
30 | TEN | 30 | 0.41 | 0.50 | 20 | 28 |
31 | KC | 32 | 0.31 | 0.47 | 31 | 25 |
32 | JAC | 31 | 0.29 | 0.51 | 30 | 27 |
TEAM | OPASS | ORUNSR% | OINT% | OFUM% | DPASS | DRUNSR% | DINT% | PENRATE |
ARI | 5.0 | 35 | 2.3 | 1.3 | 5.7 | 58 | 3.2 | 0.46 |
ATL | 7.1 | 35 | 2.0 | 0.2 | 6.8 | 53 | 3.8 | 0.23 |
BAL | 6.8 | 41 | 2.2 | 0.5 | 6.6 | 56 | 3.0 | 0.51 |
BUF | 6.2 | 48 | 3.7 | 2.3 | 6.8 | 51 | 2.0 | 0.41 |
CAR | 6.8 | 39 | 3.6 | 1.7 | 6.2 | 58 | 1.9 | 0.36 |
CHI | 5.6 | 36 | 3.6 | 1.3 | 5.2 | 62 | 5.5 | 0.39 |
CIN | 6.7 | 42 | 3.5 | 1.3 | 6.2 | 53 | 2.3 | 0.40 |
CLE | 5.7 | 38 | 3.6 | 0.7 | 6.2 | 56 | 3.0 | 0.44 |
DAL | 7.0 | 39 | 3.7 | 1.3 | 6.7 | 60 | 1.5 | 0.46 |
DEN | 7.7 | 44 | 1.8 | 2.5 | 5.4 | 58 | 3.0 | 0.40 |
DET | 6.6 | 42 | 2.0 | 1.2 | 5.9 | 59 | 1.6 | 0.50 |
GB | 6.3 | 41 | 1.5 | 0.6 | 5.8 | 55 | 2.9 | 0.43 |
HOU | 6.7 | 40 | 2.5 | 0.2 | 5.1 | 58 | 3.4 | 0.38 |
IND | 6.5 | 43 | 2.5 | 1.2 | 6.5 | 58 | 1.3 | 0.45 |
JAC | 4.9 | 38 | 2.2 | 1.8 | 6.9 | 55 | 1.6 | 0.50 |
KC | 5.6 | 43 | 4.9 | 3.1 | 7.5 | 61 | 2.5 | 0.40 |
MIA | 6.2 | 38 | 3.0 | 1.8 | 6.2 | 62 | 2.1 | 0.36 |
MIN | 5.6 | 40 | 2.5 | 1.5 | 5.9 | 58 | 1.4 | 0.37 |
NE | 6.8 | 48 | 0.8 | 0.7 | 7.3 | 54 | 3.0 | 0.37 |
NO | 7.0 | 37 | 2.4 | 0.7 | 7.7 | 55 | 1.5 | 0.40 |
NYG | 6.8 | 41 | 3.0 | 1.0 | 7.0 | 53 | 5.0 | 0.27 |
NYJ | 5.5 | 40 | 3.0 | 1.9 | 6.2 | 55 | 2.5 | 0.44 |
OAK | 6.6 | 31 | 2.4 | 1.6 | 7.0 | 64 | 1.9 | 0.44 |
PHI | 5.9 | 46 | 2.9 | 2.2 | 6.3 | 61 | 2.3 | 0.45 |
PIT | 6.4 | 34 | 1.2 | 1.0 | 5.3 | 49 | 1.8 | 0.55 |
SD | 6.4 | 39 | 4.0 | 1.5 | 6.5 | 61 | 2.6 | 0.37 |
SF | 6.5 | 52 | 2.1 | 0.9 | 5.4 | 64 | 1.9 | 0.48 |
SEA | 6.4 | 44 | 3.1 | 1.4 | 5.2 | 55 | 2.6 | 0.43 |
STL | 6.2 | 42 | 2.7 | 0.9 | 6.1 | 55 | 2.6 | 0.51 |
TB | 7.6 | 40 | 1.8 | 0.7 | 7.6 | 63 | 4.1 | 0.44 |
TEN | 5.9 | 41 | 2.0 | 2.0 | 7.1 | 57 | 2.5 | 0.40 |
WAS | 6.7 | 49 | 1.8 | 0.8 | 7.4 | 58 | 2.8 | 0.56 |
Avg | 6.4 | 41 | 2.6 | 1.3 | 6.4 | 57 | 2.6 | 0.42 |
"Add Peyton Manning. You get a team that's very hard to beat."
Particularly when you are subtracting Tim Tebow.
Recently when watching the Vikings I was wondering how good they would be with P. Manning instead of Ponder. My guess - as good as any team in the league - just like Denver.
Really surprised Carolina didn't drop heavily after an awful performance against Denver. Their offense and defense was bad!
I knew my Bills were bad, but wow. This list justifies why I watch about 1-2 quarters every Sunday, then leave to do something else.
"Take a solid defense. Add Peyton Manning. You get a team that's very hard to beat."
San Francisco was thinking the same thing not so long ago.
Why not sort run and pass ratings by rounding next decimal instead of alphabetical?
Take a below average defense, and a below average offense, and you get the 7th ranked team, ahead of a slew of teams with better ranks in both offense and defense. Buh?
I don't understand how IND rankes 32nd in defense. Their D-Pass and D-Rush are each just slightly below average, while their one major weakness is D-int%, which is heavily regressed according to what Brian has said in past articles. What am I missing here?
That 7th ranked team doesn't commit penalties. That has a considerable effect.
I should mention that penalty stats are not tallied by offense and defense, but by own team and opposing team, so it only goes to overall team rating and not offense or defense ratings.
That makes sense. Surprised the effect is so large, though, especially given their very average opp GWP. Is penalty rate sustainable?
Everything in the GWP is weighted by how sustainable it is, and penalties are surprisingly consistent.
Aside from Carolina, it seems like this table actually conforms with peoples' intuition this year.
Pshaw, these stats don't account for "fire in the belly". If Coach Pagano hadn't been diagnosed with leukemia, these Colts would be 3-6.
Seriously though, how do you explain the Colts outperforming their rank? Is it luck, and easy schedule, or a mix of the two?
Mr. Burke,
First I'd like to congratulate you for calling Denver the best team 5 weeks ago and keeping the Broncos in your top 3 all year...good job.
However, I performed a cursory analysis on your team efficiency ratings (using 8 weeks, week 4 through 11). It seems every team in the NFL moves an average 3 spots (2.95) every week. This is a lot...this means with constant direction a team that is 30th can be 6th in 8weeks and vice versa.
The only consistent team is Carolina which you insist is a top 5 or 10 team despite a 2-7 record and being blown out twice.
I don't need to tell you how off you have been on Tampa Bay, who started at 23 then moved down to 31 despite losing against 3 top ten teams(your ratings, not mine) by less than 4 pts. average. Now TB is 13.
Please explain to me, are your ratings simply academic or should a 5th rated team (at home) beat a 13th rated team?
If your ratings were predictive and not simply explanatory, then wouldn't Carolina would be 1st in the NFC south instead of last?
The algorithm I use,shows much more stability. I simply use objective statistics,(i.e. no single play subjective analysis)(e.g. red zone+third down efficiency, passer and defensive passer rating, sacks, turnovers, Yard per Carry, etc.) and pair up the teams at a neutral site to decide which is better that week.
I do have a very relevant question...what steps did you take to get your web site up and achieve a real following? I feel my algorithm is as predictive as any our there, so if you don't mind helping a future competitor, please help.
Again good job with the Broncos they were 2-3 in week 5 and you had them in the 3rd spot.
Mr. Burke,
The Anonymous in the last post has a name and an email address...
Jon Jackson
iamjonjackson@yahoo.com
iamjonjackson@cox.net
Any info would be helpful
Thanks
I am no Brian but I can give you some input on your thoughts.
I believe the people on this site are not for the prediction model. It has definitely helped to build a fan base as it has been performed extremely well. But people like me love the analysis done here. We are here because we love football. Basically, we are interested what makes teams win, especially in the NFL.
"... I performed a cursory analysis on your team efficiency ratings (using 8 weeks, week 4 through 11). It seems every team in the NFL moves an average 3 spots (2.95) every week. This is a lot...this means with constant direction a team that is 30th can be 6th in 8weeks and vice versa."
You can not predict the direction the teams move to. Injuries, for example, are impossible to predict. You probably know the math but it cought my interest! The chances of A team rising 8 weeks in a row is less than 1% (0.5 * 0.5 * 0.5 * 0.5 * 0.5 * 0.5 * 0.5 * 0.5 = 0.004). Though, there are 32 teams, so the chances that any of the 32 NFL teams rises for 8 straight weeks are about 88% ((1 - 0.004) ^ 32 = 0.88). So if we count that there are 2 chances for 8 week streaks (16 week regular season), we can roughly say that every four seasons we could expect a team to rise eight straight weeks.
So yeah. I don't think it is a fault in the model. Teams are constantly changing and the low sample size of 16 weeks doesn't help at all. But to me that is the beauty of the NFL season. Every game counts.
"Please explain to me, are your ratings simply academic or should a 5th rated team (at home) beat a 13th rated team?"
I don't understand what you mean by "academic" but a 5th ranked team should beat a 13th ranked team at home or a neutral site. I'm going to cite the "Prediction Methodology" (http://www.advancednflstats.com/2007/09/nfl-win-prediction-methodology.html) page by Brian:
"The model is based on team efficiency stats which include:
Offensive pass efficiency, including sack yardage
Defensive pass efficency, including sack yardage
Offensive run efficiency
Defensive run efficiency
Offensive interception rate
Defensive interception rate
Offensive fumble rate
Penalty rate (penalty yards per play)
Home field is also included in the model. These factors were selected because they are most predictive of future performance, and not necessarily because they explain past performance.
...
Touchdowns, or red zone performance, or third down success rates are not used in the model because I believe those things are the results of passing and running ability etc. To include them in a model intended for prediction would guarantee it is severely "overfit." In other words, it would capture and explain the unique qualities of past events at the expense of predictive power."
"If your ratings were predictive and not simply explanatory, then wouldn't Carolina would be 1st in the NFC south instead of last?"
I think you have misunderstood a little what predictive and explanatory mean. Brian's ranking are predictive. If they were explanatory then Carolina would be ranked much lower. But as it is, we are not interested in the past. We care who wins tomorrow!
Get a blog running and post your logic and the picks your model make. My suggestion: don't make it a black box. We are not competing against each other. If the source is open, people can help and contribute!
Jonathan, it's not just Carolina but also Baltimore and to some extent St. Louis that skew from intuition. I love a lot of the analysis on this site and think coaches are way too risk averse, but I think this model is buggy.
Sampo,
Nice response.
"a 5th ranked team should beat a 13th ranked team at home or a neutral site. I'm going to cite the "Prediction Methodology" (http://www.advancednflstats.com/2007/09/nfl-win-prediction-methodology.html) page by Brian:"
If Carolina is indeed a top 5 team, they should beat TB at home tomorrow. No weather effects, and both teams have some key injuries.
So I'm use your Pick on Carolina this week, for one the leagues I'm in. I trust your analysis. My system says TB should win by 3 pts but I need another underdog pick. I think you may be correct.
Carolina's schedule is fairly easy the rest of the way out
TB #13 home
Phil #20 away
KC #31 away
Atlanta #7 home
SD #21 away
Oakland #25 home
NO #24 away
So they should win what probably 6 of the 7 and end up 8-8 at the end of the year?
I haven't had a chance to review your methodology, but I will on my day off.