Charlie Shub wrote:
> The January 3 NCAA news reports on changes to how RPI will be used as
> a tool for the 2005 basketball championships.
>
> Changes were based on a 4 year study that shows home teams win about
> 2/3 of all games
>
> They will still use the 25, 50, and 25 percentages, but now
> road victories weigh 1.4 and road losses weigh 0.6
> home losses weigh 1.4 and home victories weigh 0.6
>
> The article also stated that the new weightings replace the added
> bonuses or penalties for road performance and schedule strength

I wonder how carefully they've adjusted for strength?

Doing some rather simplistic calculations whenever I run KRACH, I
usually find that home team winning percent is somewhere between 0.03
and 0.07 better than the relative ratings of the teams indicate it
should be.  It can vary between those numbers over the course of the
season, although the early-season numbers are bound to be less accurate.
(I'm talking about hockey here, I don't run KRACH on basketball.)

Nor is this universal, one season a couple of years ago, I was running
KRACH on women's results, and I found that home teams were winning less
often than the relative ratings indicated that they should.  That is,
home ice was a disadvantage.