Sunday, 12 October 2008

American Football Elo

JPG recently asked me how I apply Elo ratings to American Football.

I find this easier than real football, for two reasons. One, the elimination of the draw as a possible outcome, and two, I use the ratings to directly determine a points differential between the two teams.

For example, today sees the New Orleans Saints at home to the Oakland Raiders. I have the Saints with a bonus for home advantage rated at 25 points, and the Raiders at 16. The handicap for this game has Saints -7.5 available at 2.16. To me this is value.


Anonymous said...

Cassini, thanks for the info on this mate. Your quite right, once youve got your ratings into a points expectancy, its relatively a lot more straight forward to apply to betting markets then Soccer (mainly due to the draw.)

However, the problem Im having is the first part. Firstly, what adjustment do you use for home advantage? Traditional soccer odds compilers have used a weighting of 1.4 (or 40%) for the home team whilst ELO adds on 100 points to the home teams rating to calculate the win expectancy.

After this, how do you convert your win expectancy into a points scored expectancy? Unless you're using an ELO approach Im not aware of. I also dont know how to adjust a teams rating after a match relating to the size of their win or loss.

Any help on this would be greatly appreciated! - JPG

Cassini said...

JPG - Home advantage is simply 3 points. Nothing sophisticated there. I think that with systems, it is often too easy to over complicate things. As for adjusting the ratings after each game, I use a unique sliding scale adjustment based on the margin of defeat / victory, and I don't count overtime. I can e-mail you the precise numbers I use if you would like, but you are probably just interested in the method and can throw your own numbers in as you see fit.