Friday, May 25, 2007

A Bit More on Scariness (Including Last Night's Game)

Here’s a look at how much of a change you might see on a game by game basis.

New York’s win raises their scariness rating by .1 point, which meant that they swap positions with Dallas. In this case, I think my system underrates the scariness of the Red Bulls with JPA but as he plays more games, and they get results, it gain accuracy.

Team new rating last week
New York 2.04 1.94
Dallas 1.98 1.98

Chicago’s loss hurt them a lot more. The dropped .54 points, which means they slipped below LA, Chivas USA, Columbus, and Houston. They’re now only .06 points scarier than Real Salt Lake—which makes me feel a bit better about my most recent picks

Team new rating last week
Chicago 0.48 1.02
Salt Lake 0.42 0.42
There are a several things that should still be worked into this to create a better scariness rating:
  • Goal differential (but on a per game or a last 5 basis?).
  • Opponents scariness (for teams that win or draw)
  • Fatigue (when a team has played an extra game already this week (e.g., both RSL and the Fire will be playing their second game of the week on Sunday, so they should both be considered ‘less scary’ than normal)

Are there other things that ought to factor in here? Does the rating make sense so far?

No comments: