Below are my first actual ARPI rankings for the year, based on all games played through Sunday, September 16. Ordinarily, I vet my ratings and rankings against the NCAA's before publishing them, to be sure either I or the NCAA do not have data errors. I also check, for the first set of ratings, to see whether the NCAA staff has made slight changes in the ARPI bonus and penalty adjustments to properly calibrate them to reflect the Women's Soccer Committee's intentions. I have not done that this week, as the NCAA's published ratings and rankings, purportedly covering games through September 16, actually as missing a significant number of games already played. They're working on that problem, but rather than wait I'm going ahead with my publications for this week.
Regarding the "accuracy" of these rankings and the ratings on which they're based, as of this stage of the season last year, the ratings (adjusted for home field advantage) were consistent with game results 63.8% of the time and were within a "half result" of being consistent 19.7% of the time, for a total correct or with a half 83.5% of the time. (Being with a "half result" means the ratings indicated a tie and the game was a win/loss or the ratings indicated a win/loss and the game was a tie.) These numbers will improve week by week as more games results get into the data base.
No comments:
Post a Comment