Professor Frink's Power Rankings

by tom slick on Oct 14, 2009 6:42 PM EDT Comment 22 comments

For several years, my nerdy alter ego has maintained a spreadsheet that assigns overall ratings to each NFL team, based on a simple statistical analysis of offensive and defensive points scored & allowed. Surprisingly, it's historically been a pretty good predictor of victory, once you get 5 or 6 weeks into the season.

Weekly results and details after the jump...

Here's how the math works. For each team, calculate Points For per Game and Points Against per Game. Then, fit the league's offensive & defensive data onto a couple of Bell Curves (or normal distributions, if you prefer), a very common analysis technique. Depending where each team lies on the curve, this yields a value between 0 and 1 for each team's O and D. For each team, add the offensive and defensive values, then multiply by 50 to put the ratings on a 0-to-100 scale. Statistics 101 stuff, nothing fancy.

Not perfect (what power rankings are?) but I like the simplicity, and it's stood the test of time fairly well IMO. Without further ado, for your amusement...

Rank	Rating		W-L-T	PF/GM	PA/GM
1	91.2	Giants	5-0-0	30.2	14.2
2	90.4	Saints	4-0-0	36.0	16.5
3	86.7	Colts	5-0-0	27.4	14.2
4	83.4	Vikings	5-0-0	31.2	18.0
5	79.5	Falcons	3-1-0	25.5	15.8
6	73.3	Ravens	3-2-0	27.6	19.4
7	71.0	Eagles	3-1-0	31.8	21.5
8	71.0	Seahawks	2-3-0	23.0	16.4
9	70.0	Bears	3-1-0	26.3	19.5
10	69.7	Broncos	5-0-0	19.8	8.6
11	64.9	Cowboys	3-2-0	24.4	19.6
12	59.7	Steelers	3-2-0	22.6	19.6
13	59.2	Jets	3-2-0	20.2	17.6
14	59.1	Patriots	3-2-0	20.8	18.2
15	59.1	49ers	3-2-0	22.4	19.6
16	58.0	Bengals	4-1-0	20.2	18.0
17	55.2	Packers	2-2-0	26.0	23.3
18	53.0	Dolphins	2-3-0	22.4	21.2
19	48.7	Redskins	2-3-0	14.6	16.4
20	46.1	Chargers	2-2-0	25.3	25.5
21	45.4	Cardinals	2-2-0	21.3	22.3
22	44.4	Texans	2-3-0	23.0	24.0
23	29.2	Jaguars	2-3-0	19.4	25.4
24	26.3	Bills	1-4-0	15.4	23.2
25	23.2	Lions	1-4-0	20.6	32.4
26	17.2	Chiefs	0-5-0	16.8	27.6
27	16.9	Titans	0-5-0	16.8	27.8
28	16.8	Browns	1-4-0	11.0	24.2
29	15.5	Panthers	1-3-0	14.3	26.0
30	10.6	Raiders	1-4-0	9.8	26.0
31	10.4	Buccaneers	0-5-0	13.6	28.0
32	3.6	Rams	0-5-0	6.8	29.2

The full spreadsheet can be downloaded here.

If any of youse guys want to see future editions of this, please comment or do a Rec. I don't mind the few minutes' work posting this weekly, provided there are people who care to read it. Cheers!

This FanPost was written by one of The Falcoholic's talented readers. It does not necessarily reflect the views of The Falcoholic.

9 recs | Comment 22 comments

Story-email Email Printer Print

Comments

I like it

I was absolutely floored to see that we have allowed the second fewest points in the league. Talk about bending and then not breaking, eh?

by Dave the Falconer on Oct 14, 2009 7:31 PM EDT reply actions 0 recs

Falcons got the mojo on D thus far

Under 16 points per game (4th place actually) shocks me too, though it’s early. I would never have thought that a team allowing 4.9 yards per carry (28th place!) could do that.

by tom slick on Oct 14, 2009 10:43 PM EDT up reply actions 0 recs

Fourth fewest...

…but who’s counting.

"This is America, if we can’t self-righteously look down on others and blame them for our faults, the commies win."-----Cormican on Bleeding Green Nation w/r/t fans overreacting to the Eagles signing of Michael Vick

by Jesse28 on Oct 15, 2009 1:44 PM EDT up reply actions 0 recs

Could've sworn second

But I’m going to take your word for it.

by Dave the Falconer on Oct 15, 2009 4:09 PM EDT up reply actions 0 recs

1. Broncos – 8.6 PA/G
2. Colts – 14.2
2. Giants – 14.2
4. Falcons – 15.8

Although the Pats game skews it, we are currently 24th in scoring offense. In the end, I’m not sure that will change much and I’m not sure it will matter, so long as the defense continues to gel/grow/get more experienced. We are currently maintaining a +9.8 pt diff, good for 7th best, which I hope will grow some ensuring a couple of key wins later in the season.

by Jesse28 on Oct 16, 2009 8:42 AM EDT up reply actions 0 recs

Oh Tom

If you would be so kind, would you mind adding a +/- Differential column to the end of your spreadsheet. You don’t really have to cause I can add it each time you post the spreadsheet, but I thought it might something interesting for others to see when you post.

Just a thought.

by Jesse28 on Oct 16, 2009 8:44 AM EDT up reply actions 0 recs

Perhaps if there's room...

Interesting idea, but after I add the columns for the offensive and defensive ratings and totals, I’m not sure how much room will be left. I was planning to split those off to the right, to keep the chart readable and uncluttered. Personally, when I look at a chart that’s too cluttered, my eyes go all Marty Feldman on me. I will give it a try and see how it looks. Thanks for the feedback!

by tom slick on Oct 16, 2009 12:07 PM EDT up reply actions 0 recs

If it's too cluttered

I’ll put it on the spreadsheet so you can d/l it.

by tom slick on Oct 16, 2009 12:16 PM EDT up reply actions 0 recs

Meh

No worries. Like I said, I simply added it to the one I DL’d and I can just post the results in the comments the next time you post it. So long as you make it DL-able each week, that is.

by Jesse28 on Oct 19, 2009 9:30 AM EDT up reply actions 0 recs

This is awesome

by thisistherevolt on Oct 14, 2009 7:38 PM EDT reply actions 0 recs

This is great.

I don’t usually like stats but this is a good simple way of indicating rankings throughout the league. I wouldn’t mind having this as a weekly addition.

New Orleans Hornets: The most inconsistent team in the NBA.

by Grayson on Oct 14, 2009 11:20 PM EDT reply actions 0 recs

Thanks

Nice to hear that from someone who’s not a big stats person. Simple is good, all right.

by tom slick on Oct 15, 2009 3:18 PM EDT up reply actions 0 recs

No worries.

I think people can place stats any which way without considering their meaning or context. At the same time people can jibber-jabber on about things but have no evidence to back up what their saying. This can be a helpful section to get a gage on how teams are going as a whole.

New Orleans Hornets: The most inconsistent team in the NBA.

by Grayson on Oct 16, 2009 1:03 AM EDT up reply actions 0 recs

Love it

Surprised to see the Seahawks that high (I guess a 41-0 win helps). I knew the Bengals were playing close games, so I expected them to be relatively low. And wow, the Rams aren’t very good, are they?

Very similar to a Pythagorean Win Expectency, but I like how you adjust it to a 1-100 scale for more of a “Power Poll” look. Great work.

by orang3b on Oct 15, 2009 12:11 PM EDT reply actions 0 recs

Damn Seahawks!

That’s the problem with algorithms, there’s always a turd in the punch bowl somewhere. Thanks.

by tom slick on Oct 15, 2009 3:02 PM EDT up reply actions 0 recs

Agreed, Rec'd, and more preeze!

Point differentials tend to work quite well for most sports teams, regardless of sports. They are always a decent indicator of a teams’ true identity. I’d definitely be interested in seeing this each week.

by Jesse28 on Oct 15, 2009 1:47 PM EDT reply actions 0 recs

Funny how the simplest methods

can work as well as they do. On my old Commodore 64, I used to crunch huge amounts of data using linear regression formulas and they didn’t work any better than this simple formula does.

I might be reading you wrong, but this isn’t based on point differential. Offense and defense are each assigned a value between 0 and 1 depending where they lie in the normal distribution curve. Maybe I’ll add that data to the next summary to make it clearer. Thanks!

by tom slick on Oct 15, 2009 3:15 PM EDT up reply actions 0 recs

Yeah,

Add that next time – because I misunderstood what this was, too (Pythag comment)…

by orang3b on Oct 15, 2009 3:31 PM EDT up reply actions 0 recs

Appreciate the feedback

I was trying to not drone on, and didn’t explain it thoroughly. If you open the spreadsheet and select the “Ratings” worksheet, you can see all the numbers step by step.

by tom slick on Oct 15, 2009 4:32 PM EDT up reply actions 0 recs

Differentials

No, not directly, but because you are combining where each teams O and D places based on points for and points against, it is similar to a basic point differential evaluation, or in the least it allows you to see the differentials. I understand that your rating’s weren’t based off of differentials, but it allowed me to view them so that’s why I mentioned it.

Then again, I don’t get to use this very often and I’m a littly rusty, so it very well could be that I’m just confused.

by Jesse28 on Oct 16, 2009 8:54 AM EDT up reply actions 0 recs