"Elo" ratings in NFL football

All Things Chess
1

.
http : // fivethirtyeight .com/ datalab/week-17-nfl-elo-ratings-and-playoff-odds-special-seahawks-edition/

fivethirtyeight.com/datalab/week … s-edition/

“538”
NFL Football, USA

[size=150]Week 17 NFL Elo Ratings And Playoff Odds: Special Seahawks Edition[/size]

by Nate Silver
2014/Dec/23 06:01am

[ GeneMi:
I was unaware that anyone is using “Elo” ratings for the NFL, or outside of chess.

I notice that the skill range in the NFL is faaaar smaller than in chess (not surprising). In chess the Elo range is about 800 - 2850. In the NFL the range is 1262 - 1755.
]
.

2

Elo-like ratings are also used in table tennis, and have been for a few decades.

This came about because somebody from Washington State (perhaps Robert Karch?) who was a member of both USCF and USTTA “imported” the chess system into table tennis.

Bill Smythe

3

The system can be applied to all competition imho :mrgreen:

4

The Sagarin ratings in USA Today have an “Elo Chess” component for all of the major sports.

5

This seems to base the “Elo ratings” somewhat on margin of victory. There is no way to do that in chess.

Alex Relyea

6

Elo ratings have been modified for team sports for a while now. Such modifications can include margin of victory in individual games (as Mr. Relyea noted, that is not reasonably possible in chess).

Elo ratings were even a part of the Bowl Championship Series calculation (because the Sagarin rankings have an Elo component).

7

Elo is also used to compare fighters in mixed martial arts.

http://www.mma-elo.com

8

The NFL would be more like the FIDE subset of ratings than the whole batch.

9

They offer a more “nuts and bolts” explanation on a linked page on their site.

10

I think “sandbagging” should be taken into account. The at one time 0-16 Lions rated higher than the Ravens? I can only imagine this system used in the NBA for the 76ers. I bet they would have an even bigger “rating jump” in a couple of years than the Lions did this year. Teams change more than individuals do.

11

A most improved list for the year would be an interesting addition to Chess Life.

12

Then teams should have a higher K-factor, I would think.

Perhaps a team’s K-factor at the start of the season should depend on how many personnel changes there have been since the previous season.

Bill Smythe