armchairgm
all sports, all you
+ Add Friends
You are not logged-in.
Sign Up - Log In
Main Page
Sports
Write
Articles
Hot Links
Images
Meet People
Fun
Explore
MLB - NFL - NBA - NHL - College Basketball - College Football - Soccer - Nascar - Other
Article - Locker Room Discussion
All Articles - New Articles - Today's Articles
Submit a Link - Approve Links
Picture Game - Ratings - Polls - Pick Game - Quiz Game - Spring Silliness
Random Page - Random Image - Random Fan
Edit
Page history Discuss pageWhat links here

A Method for Calculating Power Rankings (and projecting final standings)

15
Vote

by Derekcs

Team quality is not exactly a transitive property (i.e. A is better than B, B is better than C, does not necessarily mean A is better than C.  C might just match up really well with A).  So how would you decide who is the best team in the NFL?  My idea was simply this: if you pitted each team against every other team in a home and away game, who would be expected to win the most games?  Obviously, actual results can vary widely, and don't necessarily reflect the entirety of a team's quality.  But based on the stats, we can get some idea of who's more likely to win.  With logistic regression, we can produce the probability that a team is going to win a given game.  The power ranking is simply the expected win totals (the sums of those probabilities for each team) normalized for a 16-game season.

The inputs used for this regression were rushing yards per carry, passing yards per attempt, sack rates, third down conversion rates, interception and fumble rates on offense and defense.

The power rankings as of week 2 of the 2007 season are:

Rank Team Expected Wins

  1. NE     12.7821
  2. DEN     12.3057
  3. PIT     11.9186
  4. IND     11.2294
  5. HOU     11.0758
  6. DAL     10.9747
  7. TB     9.6725
  8. CAR     9.4341
  9. DET     9.1868
  10. WAS     9.1101
  11. ARI     9.0378
  12. MIN     9.0016
  13. JAX     8.8980
  14. PHI     8.7440
  15. SEA     8.4853
  16. BAL     8.2007
  17. TEN     7.4589
  18. CLE     7.3766
  19. GB     7.2951
  20. CIN     6.8305
  21. KC     6.8023
  22. CHI     6.7402
  23. STL     6.7007
  24. NYG     6.2975
  25. SF     6.2020
  26. MIA     6.0230
  27. SD     5.9910
  28. ATL     5.5020
  29. OAK     5.1785
  30. NYJ     4.9122
  31. BUF     4.2329
  32. NO     2.3994

A couple things to point out here.  First, KC is ranked higher than the Bears because Devin Hester's talents are not included in the model.  Otherwise, Chicago has not really been any better than KC.  Second, passing is valued much more highly by the model than rushing, and offense is valued somewhat more than defense by the model.  So you might not like Detroit over Minnesota, but as I've pointed out here and here, good pass offenses are more likely to take a team to the postseason, and the strength of Minnesota's defense has been against the run.

Here's another moment where I point out how eerie Jon Kitna's prediction of the Lions winning 10 games is.  Using the probabilities generated by the model, I looked at each team's remaining schedule and projected their final win totals based on the number of games they are considered more likely to win.  Detroit and Minnesota are projected to lead the NFC North with 10 wins, with Chicago bringing up the rear (5 wins).

Two caveats about these projections:

1. SMALL SAMPLE SIZE.  I don't think San Diego is a 3-13 team.  They haven't performed all that well in the first two weeks, however.  (Opponent adjustments will help once the sample size increases by a week or two).

2. THIS MODEL DOES NOT CONSIDER ALL POSSIBLE FACTORS.  No, no one expects the Saints to go winless or the Pats to go undefeated.  A good proportion of the games are considered close calls, but in the case of the Pats, they're all in their favor.  The reverse is true for the Saints.  Luck and special teams and injuries and other factors not considered by the model will swing many games the other way.

AFC East BUF    2 MIA    5 NE    16 NYJ    2

AFC North BAL    9 CIN    6 CLE    8 PIT    14

AFC South HOU    15 IND    14 JAX    8 TEN    6

AFC West DEN    14 KC    7 OAK    3 SD    3

NFC East DAL    14 NYG    3 PHI    9 WAS    11

NFC North CHI    5 DET    10 GB    8 MIN    10

NFC South ATL    4 CAR    9 NO    0 TB    12

NFC West ARI    11 STL    4 SF    3 SEA    11


Enable Comment Auto-Refresher
Taytay 24All-American
786 days ago
Score 0+-
I think you put in an admirable amount of work to get useless data. Denver at 2? Pitt 3? Houston 5? And on it goes...
Permalink | Reply
DNLLegend
786 days ago
Score 1+-
The way Houston is playing right now, they should be up there. The problem is not the result, but that the data set is too small.
Permalink
Taytay 24All-American
786 days ago
Score 0+-
Houston is playing well (I stole their defense with my last pick of the fantasy draft), but I'm not ready to say they are the fifth best team in the NFL. And I know the data set is small. I'm just thinking this formula should be applied at maybe midseason.
Permalink
WrmjrRed-Shirting
786 days ago
Score 0+-
Not sure I understand your numbers exactly, but maybe it's the layout. Denver is #2, but you have them third in wins (after NE and Ind). I agree with TayTay that Den seems too high, but most early season projections look out of whack due to small sample size.
Permalink | Reply
DerekcsSoccer Kid
785 days ago
Score 0+-
Denver's high ranking is largely a result of their pass defense yielding 2.84 yards per pass play. They were facing Buffalo and Oakland, however. So once I start opponent adjustments in week 3, Denver's obviously going to be lower. Plus there will be regression to the mean. The second table is using the projected probabilities of winning for actual games to predict each team's final win totals. The first table uses the projected probabilities of winning all possible games.
Permalink
PeanMajor Leaguer
786 days ago
Score 1+-
I think the Giants prediction is 6.2975 wins off
Permalink | Reply
Taytay 24All-American
786 days ago
Score 2+-
Me, too. Much too high.
Permalink
CheezerAll-Star
786 days ago
Score 2+-
When you say you include fumbles in your calculations, is it fumbles lost or just fumbles.

Fumbles are predictable = Certain players tend to fumble more than others Fumbles lost are not predictable = Once the ball is on the ground, it becomes random as to which side will recover it.

The importance of running as it relates to wins comes in the second half. The better teams usually have the lead and are running more to use up the clock.
Permalink | Reply
DerekcsSoccer Kid
785 days ago
Score 0+-
Fumble rate = *all* fumbles / (completions plus running plays)

There can be fumbles on defense and special teams, but the little bit of statistical noise is acceptable.

Rushing yards/game has a higher correlation to wins than passing yards/game for the reason you stated.

And three is 50% better than two. "There's an old saying in Chicago. Once is happenstance. Twice is coincidence. And three times is enemy action."
Permalink
False ProphetAll-Star
786 days ago
Score 2+-
The problem is sample size. You'll need at least another week before you can get a goot power ranking (which is why mine won't start till after week 3)
Permalink | Reply
JuTMSY4Legend
786 days ago
Score 3+-
best cover for laziness...ever!
Permalink
Taytay 24All-American
786 days ago
Score 1+-
Three is that much better than two?
Permalink
RawbeezeitzMajor Leaguer
786 days ago
Score 0+-
How do I determine who the best team in the NFL is? I call it...the Super Bowl. It's a few months from now.
Permalink | Reply
Add your Comment
ArmchairGM welcomes all comments. If you don't want to be anonymous, Register or Login. It's free


Retrieved from "http://armchairgm.wikia.com/A_Method_for_Calculating_Power_Rankings_%28and_projecting_final_standings%29"

This page was last modified 05:24, 20 September 2007. Content is available under the GFDL.

Categories: Opinions | Opinions by User Derekcs | September 20, 2007 | NFL Opinions

Contribute

ArmchairGM's pages can be edited.
Is this page incomplete? Is there anything wrong?
Change it!

Edit this page Discuss this page Page history

Recent contributors to this page

The following people recently contributed to this article.

Embed this on your site

Main Page About Special Pages Help Terms of Use Advertise