Bowlsby on the New RPI, circa Dec 2004

I dug up this gem from an e-mail I received from former Selection Committee Chairman and current Stanford AD Bob Bowlsby over two years ago:

"Our committee has come to resolution on our formula after consultation with some of the leading mathematicians in the nation and I believe we will not be discussing our formulas again anytime soon."

Yikes! What does that say about the state of mathematics in the U.S.? I've shown that the new RPI is consistently about 5% worse than the old RPI, which is pretty astounding considering that the old RPI wasn't that good to begin with. Obviously, they wouldn't have made the change unless they thought it was an improvement over the old system, right?

The Committee members were and still are delusional about the accuracy of the new RPI. In a recent NCAA teleconference, current Selection Committee chair Gary Walters made the following comment:

"I think the important issue related to the RPI is that we now take into account where games are played. That's very, very important."

Well, he's right about taking location into account. Not accurately. Not in a fashion that adds value, but taken into account nonetheless.

In the same interview, Walters goes on to talk about the focus on non-conference play:

"Having said that, we're also looking at the strength of non conference schedule and performance against that non conference schedule. Obviously that has an impact not only on the quantitative data and RPI of the team, but also gives an impression as to whether or not teams can compete against the teams that are in the top 50 in the country."

I agree with the logic, but the thing he doesn't understand is just how much worse the RPI gets when you starting breaking it down into components like non-conference, etc. Teams that can be helped out by inflated non-conf numbers include:

Arizona (again) (NC RPI 2 vs NC Colton 15, NC RPI SOS 1 vs NC Colton SOS 15)
Southern Illinois (13 vs 38, 42 vs 48)
Gonzaga (45 vs 113, 9 vs 12)
Davidson (33 vs 119, 75 vs 104)
Memphis (17 vs 68, 22 vs 50)
Drexel (5 vs 73, 5 vs 22)
Old Dominion (51 vs 109, 33 vs 71)
Butler (14 vs 62, 24 vs 67)
Appalachian St (7 vs 107, 4 vs 4)
BYU (69 vs 93, 126 vs 205)
UNLV (12 vs 35, 30 vs 95)
Texas Tech (43 vs 57, 52 vs 142)
Xavier (49 vs 97, 49 vs 138)
Arkansas (22 vs 28, 29 vs 107)
Kentucky (6 vs 19, 2 vs 28)
Illinois (29 vs 41, 37 vs 172)
Tennessee (again) (8 vs 24, 7 vs 69)

A lot of bubble teams on this list. This is a list you rather be on if you're a bubble team. I'm concerned about teams like Drexel and Old Dominion perhaps squeaking in on the strength of some inflated numbers. The Colton Index has them 60th and 61st overall respectively, or in other words barely in the bubble conversation, as opposed to RPI's of 38 & 40. That, combined with the inflated non-conference numbers above really helps their chances. And why more programs don't copy Arizona and Tennessee's non-conference scheduling strategy is beyond me.

I fully expect that the brackets will be nearly as bad as last year. I don't see any Utah State's or Air Force's on the horizon, but you never know. I chastised Lunardi for even having Utah State listed as a bubble team in his projections last year, and they ended up making it as an at-large. Of course, I was basing my stance on the fact that Utah State's profile was barely ahead of Northwestern (12-15) and DePaul (13-15) and behind 21 other schools that were more deserving of at-large bids, but somehow Utah State pulled it off.


Wegoblogger #31 © 2011 | Designed by Bingo Cash, in collaboration with Modern Warfare 3, VPS Hosting and Compare Web Hosting