--- In collegerpi@yahoogroups.com, Jon Scott <jonpscott@y...> wrote:
> If you were to correlate the NCAA seeds (from the #1 seeds down to
theat-large bids) against the RPI, you will find that while there is
a general correlation, the accuracy is only so-so. You can do the
same analysis for polls (by including the 'also receiving votes')
and other mathematical models and will find varying correlations,
some worse but many better than the RPI.

My reply:
Jerry [Palm, of course] has run these numbers and finds the polls to
be the most accurate predictor of seeding.

> But what is really interesting to me is that if you take an
average of these disparate ways of ranking teams and then correlate
it against the actual NCAA seeds, you will find a very good
correlation.

Inevitably true, none will predict exactly, you just have to figure
out the weighting (which will vary year-to-year, of course).

> The Georgia example is one which points to the power of the RPI,
but there are certainly other examples where a team made the at-
large field with a less than stellar RPI rating. As I said, on
average, it's obvious that there are quite a number of factors which
go into the NCAA's decision making, whether people want to recognize
it or not.

We were just talking on the other list about 2002 Oregon: RPI 33 but
2 seed and went to the elite eight. Conversely Gonzaga who people
thought got screwed last year.

:o) ccf