## Efficiency Rankings - Week 7

Recently, I've received a handful of inquiries about how these rankings work. With teams like the 2-4 Chargers and 1-5 Cowboys at the top of the rankings, that's understandable. I thought it might be helpful to explain some things in plain language.

These are not 'power ratings' or my personal estimations of team strength. They're not necessarily the teams I think will make the playoffs. They are simply rankings in order of team efficiency.

I found that efficiency, more than anything else, correlates with, and is predictive of, winning. So the model focuses on net passing and running efficiency on both sides of the ball. It also considers turnover rates, such as interception per attempt and fumble per play. Penalty yards per snap is also included. All of these factors are weighted according to how well they mathematically correlate with team wins.

What makes this model unique is that it distinguishes between factors that are explanatory (telling us why teams won past games) and factors that are truly predictive (telling us why teams will win future games). For example, turnovers and turnover differential account for a very large part of why teams won previous games. But turnovers are extremely random from week to week, and teams are not consistent from one part of the season to the next. They're obviously not entirely random. (Peyton Manning will never have a 4% interception rate--3% is league average), but a very large part of the variance in interception rate can be accounted for by randomness. In other words, Eli Manning might have a 3.9% interception rate so far this year, but it's far more likely that his future games will be much closer to the league average of 3% than they will be to his current 3.9%. So factors like interception rate, which are relatively random, get heavily discounted in this model, especially early in the season.

Lastly, there is an adjustment for previous opponent strength. Once all the teams' ratings are calculated, I average up the ratings of each teams' previous opponents. Teams with tougher opponents get their rating bumped up and vice-versa.

Let's take Atlanta as an an example. They're a fairly average team statistically: 5.8 net YPA (slightly below league avg), 4.1 YPC rushing (avg), 6.4 defensive YPA (slightly worse than avg), and 4.3 YPC on defense (slightly worse than avg). One of the major reasons they are winning, however, is their turnover differential. They are tied for 4th in the NFL with a +6 giveaway/takeaway. This is primarily due to a whopping 5.6% defensive interception rate. There is virtually no way that pace can continue. History tells us their future turnover differential is far more likely to be closer to league-average than where it is now.

So statistically, the Falcons are average or slightly below, but they've been playing relatively weak opponents. They've faced PIT, ARI, NO, SF, CLV, and PHI. Of those teams, only 2 are contenders--PIT and PHI. Even NO has been surprisingly weak so far. Ultimately, you have an average but lucky performance for the Falcons, but they managed it against some pretty weak teams. So they would drop from league-average, about 17th or 18th, to where they are at 25th. They are a missed FG and a freakish (but incredible) forced fumble away from being 2-4 instead of 4-2 against an easy schedule.

Here are this week's rankings. (Click on the headers to sort.)

 RANK TEAM LAST WK GWP Opp GWP O RANK D RANK 1 SD 1 0.78 0.38 1 11 2 NYG 2 0.76 0.54 5 1 3 IND 7 0.70 0.58 7 16 4 PIT 4 0.69 0.44 11 5 5 GB 6 0.68 0.52 4 12 6 KC 3 0.66 0.58 16 4 7 DAL 5 0.65 0.59 3 14 8 HOU 11 0.63 0.60 2 27 9 PHI 12 0.63 0.46 8 10 10 CHI 8 0.62 0.56 17 7 11 WAS 9 0.59 0.60 12 17 12 TEN 13 0.58 0.53 13 2 13 MIA 10 0.58 0.49 14 15 14 BAL 14 0.54 0.49 18 8 15 SEA 17 0.53 0.52 19 9 16 MIN 16 0.52 0.52 26 3 17 NO 23 0.49 0.36 9 25 18 NE 15 0.49 0.46 6 29 19 DEN 19 0.49 0.53 10 24 20 NYJ 18 0.47 0.49 23 13 21 CIN 20 0.42 0.41 25 21 22 SF 22 0.41 0.49 24 18 23 DET 28 0.40 0.59 15 22 24 CLE 25 0.37 0.50 20 28 25 ATL 24 0.37 0.47 22 23 26 CAR 27 0.33 0.52 32 6 27 JAC 26 0.33 0.58 27 30 28 TB 21 0.32 0.46 21 32 29 STL 31 0.31 0.46 30 19 30 BUF 29 0.29 0.51 29 31 31 OAK 30 0.25 0.49 28 26 32 ARI 32 0.20 0.44 31 20

Here are each team's efficiency stats.

 TEAM OPASS ORUN OINT% OFUM% DPASS DRUN DINT% PENRATE ARI 4.3 4.5 4.5 1.5 6.4 4.3 3.9 0.54 ATL 5.8 4.1 1.8 0.3 6.9 4.2 5.6 0.37 BAL 6.3 3.5 2.9 0.6 5.4 4.3 1.6 0.44 BUF 4.8 4.8 2.9 1.0 7.0 4.8 0.7 0.31 CAR 4.1 3.9 6.0 3.3 5.5 3.8 5.7 0.43 CHI 5.7 4.0 4.6 0.0 5.5 3.5 3.0 0.35 CIN 6.2 3.7 3.1 1.5 5.8 4.4 4.1 0.42 CLE 6.0 3.8 4.1 2.4 7.3 4.0 2.2 0.45 DAL 7.1 4.0 3.4 0.0 6.1 4.2 1.5 0.67 DEN 7.2 2.7 1.2 1.3 6.7 4.4 2.7 0.53 DET 5.7 3.5 3.1 1.0 6.4 4.9 3.6 0.55 GB 6.8 4.6 3.5 0.4 5.3 4.5 3.2 0.48 HOU 6.7 5.2 2.5 0.3 7.5 4.2 1.7 0.36 IND 7.3 3.7 0.8 1.2 6.0 4.8 2.6 0.32 JAC 5.4 4.3 5.3 1.0 7.9 4.3 2.1 0.39 KC 5.9 4.9 2.2 0.4 6.2 3.7 1.6 0.28 MIA 6.2 3.9 3.4 1.1 6.0 4.1 2.6 0.24 MIN 5.4 4.5 4.7 1.7 5.5 3.8 2.4 0.49 NE 6.6 4.4 2.4 0.0 6.9 4.0 3.7 0.39 NO 6.8 4.0 2.6 1.3 6.0 3.9 2.2 0.36 NYG 6.5 4.7 3.9 1.6 5.0 3.5 3.8 0.46 NYJ 5.6 4.9 1.1 1.3 6.1 3.5 2.3 0.55 OAK 5.2 4.2 4.1 0.7 6.4 5.0 1.2 0.60 PHI 6.7 5.0 1.0 0.6 5.3 4.2 4.4 0.55 PIT 6.8 4.1 4.6 1.3 5.8 2.7 4.3 0.32 SD 7.9 4.3 2.3 1.6 4.9 3.7 3.9 0.35 SF 5.8 4.0 4.0 1.1 6.2 3.9 3.3 0.55 SEA 5.9 3.6 3.6 0.0 6.3 2.9 2.3 0.41 STL 5.1 3.6 3.4 0.0 5.8 4.4 2.3 0.43 TB 5.9 3.6 1.7 0.8 6.8 5.3 6.6 0.30 TEN 6.1 4.4 2.1 1.4 5.4 4.0 4.3 0.59 WAS 6.4 4.1 2.3 0.7 6.5 4.7 1.5 0.35 Avg 6.1 4.1 3.1 1.0 6.1 4.1 3.0 0.43

### 21 Responses to “Efficiency Rankings - Week 7”

1. makewayhomer says:

have you tried to convert these to a point spread model? or a moneyline model? how many NFL points is worth an efficiency point? that should be your litmus test for how predictive your model is - if you can't beat moneylines you're not adding too much, but if you can...well you are doing great

2. Andy says:

I slightly disagree with you, makeway. The intent of this model is to see how well we can mathematically predict winners (while using simple metrics). This model has been predicting winners at a very high rate for about four years (~69%, i think that beats Vegas).
The model doesn't actually try to predict points, so i don't think that should be the ultimate measure of how good this model is.

3. James says:

I can only sort the columns in descending order, and not ascending order.

4. makewayhomer says:

Andy, I'm not sure you do disagree with me. though "picking winners" is a bit simplistic - what I'm saying is that the Vegas moneyline can easily be converted to say "team X has y% chance of beating team Z" which I think brian has done with these ratings. I think a comparison of those 2 should be the litmus test

5. makewayhomer says:

although I should say that we can't fully expect these rankings to compete, based on the fact they exclude special teams completely. that, perhaps more than anything else, if why teams like SD and DAL and GB appear so high here despite their poor records.

now, those teams have had a a lot of weird things go on which you can't expect to continue. but not 100% of it is random variance. some of it is skill related - unless you think 100% of special teams is unpredictable randomness.

6. Brian Burke says:

Regarding special teams. They are not luck, obviously. It takes lots of skill and dedicated practice to execute ST well. If you put me out there on the Ravens punt squad, there would be one blocked punt on every other attempt.

However...6'1" 185 lb guys who run 6-second 40 times are not out there in the NFL. The guys play ST are relatively close in ability. They're close enough that it takes many, many repetitions before we would have any idea which teams' ST squads are any better than the others. There simply aren't enough iterations of punts, kickoffs, etc in a season to get any idea of which teams are actually any better than the others.

Further, even if we could tell them apart, there are so few examples of each type of ST event in each game that a team's cumulative advantage is going to be very, very small in any one given game. The variance in ST outcomes for squads of all levels (the noise) swamps whatever true talent/ability advantages (the signal) any team would have.

7. Chase says:

While I can't say I agree that the Jets will be the 20th best team going forward, I will admit that I'm a believer in what Brian's doing here. So I'm nervous.

8. Brian Burke says:

My gut, and faith in other people's assessments, are telling me that Jets are much better than this. But for whatever reason their pass defense is surprisingly pedestrian. They excel at running the ball, but (in part thanks to you, Chase), I have to say that's a flaw in the model, not with the Jets. Going by SR, they look much better.

I also think their offensive passing in game 1 vs. BAL was unrepresentative of their "true" game. And now that S.Holmes is on board and playing well, I expect their numbers to further improve.

60-yard pass interference penalties help too!

I'm planning to post updated SR rankings tomorrow. This time they'll be adjusted for opponent. The Jets (and Ravens, and Pats) look much better.

9. Anonymous says:

wow... I found this site trying to discover who stalls in the red zone the most to help me determine who a good kicker might be... low and behold what you've got here is well beyond that - can't say I understand it all... but I'm going to bookmark it to come back and try to figure it out... peace... will

10. makewayhomer says:

Brian,

the absence of a healthy Revis over a large % of the season prob. has a lot to do with the Jets Pass D.

11. Anonymous says:

Pardon me if I missed this on your site but how often does the better team in terms of efficiency actually win?

As well, how often does the Vegas favourite win?

Also curious as to how often the team with the superior pt differential wins.

My guess is all the figures are fairly close if taken over a number of years.

12. James says:

Brian, in response to some of your more recent posts are you considering somehow incorporating SR into your efficiency rankings?

I suppose they wouldn't be pure "efficiency" rankings then, but possibly under some other title?

13. Brian Burke says:

Even w/o Revis, don't you think the Jets should be better than avg?

James-No. I'm still kicking around various models that would incorporate SR.

14. coldbikemessenger says:

I forgot how you calculated the penalty rate?

Can you go over that again or provide a link?
thx

15. Brian Burke says:

It's just penalty yards per play.

16. Michael L says:

makeway, you can compare directly to Vegas -- besides the spread, books also offer odds for winning straight up. THAT bet is called the moneyline (you have your terminology confused). No need to use the spread at all.

Brian, besides unpredictable big plays on ST, I'd think the biggest contribution by ST would be to field position. KO return seems fairly stable (I'd lean towards using Opponent Start after KO rather than return yards), and each team probably averages 5-6 per game. You start generating run and pass efficiency as early as after 2-3 games, which is maybe 40-60 plays each. Obviously the variance is greater in KR, but you'd see that many reps by 7-8 games. If run/pass #s are meaningful by then, KR #s would be eventually in the season, no?

Also KR obviously only tells part of the field position story, yet punting provides less obviously useful (i.e. not heavily situationally-biased) metrics. Perhaps touchback rate, inside 20 rate, or I20/TB ratio could provide meaningful regression coefficients?

17. Anonymous says:

I think that part of the explanation for the Jets rating is the fact that special teams is not included. Jets Special teams have been excellent, pinning the opponent inside the 20 and giving the Jets good field position. I wonder if your model could somehow include team and opponent starting position to partially include special teams in the analysis.

18. Anonymous says:

The Jets have the best record in the NFL at 5-1, but ranked 20th for "efficiency?" The #1 "efficient" team is the Chargers, who are floundering at 2-4 and tied for last place in their division. With all due respect, this is just as meaningful as passer rating, in which Tony Romo has the 3rd highest career rating of all time.

19. Anonymous says:

No, it's not much ado about nothing. Special teams have had an unusually significant impact on game outcomes this year, nowhere more so than in San Diego, who is currently on pace to have the worst special teams in the last 15-20 years by a healthy margin. But when a team is killing it on offense and defense game after game, as the Chargers are, it is highly likely to show up in future win totals. Likewise, with the Jets, the primary issue is that they have been maintaining an unsustainable turnover margin. It's not that they can't be a good team going forward, but they weren't going to continue to have a 0% interception rate, and they are likely going to fumble (and not recover) more than they have to this point. It's something to be aware of. Or not, if you feel like being dismissive. I'm a Jets fan, but I also like to look at the numbers objectively, both here and at FO, and there are clear tells as to where the team has gotten away with some unacceptable play (most notably 3rd down pass defense). They can either fix that going forward or they can lose more games; those are the only options.

20. Anonymous says:

dude, every team that the jets beat accept the God awful Bills are ranked higher than the Jets? And you think this is an accurate determinate. As far as ST, the Jets don't have a return for a TD so I would think their impact is rather consistent and likely to continue since it isn't skewered by alot of long return TDs.

21. Anonymous says:

Is there any information on this site that shows how accurate your predictions are?