Done this a couple times on the blog, and figured, with money in front of me in Vegas, I might want to do it again, looking at the current MGM over/unders.
There has been created what is called a "Pythagorean Theorem" of the NFL.
The exponent has been modified to 2.37, but it seems to at least work to some extent.
Basically, what you do is you create a fraction: You take the number of points scored, raised to the power of 2.37, and divide that by the sum of that numbers (points scored ^ 2.37) plus the number of points allowed raised to that same power.
You'll get a fraction between 0 and 1. Multiply that fraction by 16, and subtract that number from the actual win total.
If you get a positive number, the team overperformed. A negative one means the team underperformed.
The key part of the theorem is if you get any result that involves a value of 2 or more, it indicates the team SHOULD (but does not always) make a move in that particular direction toward their Pythagorean the next year. (An underperformer will improve, an overachiever will drop.)
So, I took the 2014 numbers and put them into a spreadsheet, and this is what I got:
Some other observations first... Usually, I see defensive stats based on yardage. Yardage doesn't (always) score points, especially in a rigged league. I use more the points allowed, even if the points are scored defensively or on special teams.
The five teams with the fewest points allowed in 2014:
5) Arizona - 299 (11-5, eliminated in Wild Card Round): They were the team who became the second team screwed by the "losing record but winning division gets you a home game", but this was one of the larger reasons Arizona surprised, even with all their injury issues.
4) Buffalo - 289 (9-7, out of the playoffs): One of the reasons a lot of people are taking a look at Buffalo this year is that defense. Another is Rex Ryan.
3) Detroit - 282 (11-5, screwed in the Wild Card Round): Don't think I need to go through this again, but the big names are now gone in Detroit on the defensive side.
2) Kansas City - 281 (9-7, out of the playoffs)
1) Seattle - 254 (12-4, rigged into the Super Bowl, rigged out of winning it) Three years in a row now that Seattle is #1 in points allowed.
The five teams with the most points scored:
5) Dallas - 467 (12-4, screwed in the Divisional Round) And no, that is not Chris Christie's weight, or the number of
thousands millions in his political slush fund.
4) New England - 468 (12-4) I'm not even dignifying something even the NCAA would revoke!!!
3) Philadelphia - 474 (10-6, out of the playoffs) But giving up 400 shows Chip Kelly only still appears to know one side of the ball.
2) Denver - 482 (12-4, out in the Divisional Round)
1) Green Bay - 486 (12-4, laid down in the NFC title game for the 1-2-3 and the J-O-B) WW-NFL, people.
--
Anyhow, here's what I found applying the NFL Pythagorean Theorem:
1) Every divisional winner overperformed.
Yep, all 8 of 'em. The two closest were Seattle and New England, so the Pythagorean had the Super Bowl right. All 6 of the others were about or over a game overperforming.
2) Only one of the 12 playoff teams underperformed: Baltimore.
By about 3/4 of a game, their 10 wins were an underperform. Why? Only 302 given up on defense, 6th least in the league.
3) If you wish to apply this theorem, watch for improvements by Tampa Bay, the Giants, and Tennessee.
Those were the three biggest underperformers, but none got over about a value of a game and a half. I do see (for ulterior reasons) Tampa taking a nice jump this season.
4) On the other side of the coin, the theorem actually does call for a big drop by Arizona.
Arizona is the only team which actually made the 2 cutoff up or down. (Cincinnati and Detroit got close, Dallas a bit less so.) But the fact that Arizona went 11-5 with a season differential of only 11 points forbodes for them, except one has to wonder what that NFC West is going to look like this year.
I have another almost two days in Vegas yet, so I'm going to take a look at a few more numbers and might have more on the numbers if something jumps out at me.