Dance Card

[Bill Marsh]

Why all the talk about "old" bs "new" RPI. AFAIK, the weighted RPI has been in use for 10 years. Have I missed some new change?

[stever20]

because the dance card inexplicably still uses the old RPI and with a team like a St John's- that makes a huge difference.
Old RPI 30
Current RPI 38

Evaluating St John's at RPI 38 is a lot different than at RPI 30. Here both would get in. But if it got to say 45 for old and 53 for current- that's a different animal completely.

[stever20]

there are others that are like St Johns- even worse
NC State
old RPI 25
current RPI 40

Miami
old RPI 57- because of this in dance card as 2nd team left out
current RPI 67

Purdue-
old RPI 50- because of this 12 teams in after them
current RPI 59

on the other end-
Texas A&M
old RPI- 38- because of this 4th team in
current RPI 31

[paulxu]

Dance Card's use of the old RPI is not "inexplicable."
They use it because it is one piece of their unique formula (that contains a lot of other pieces) that continues to work and produce good results.
They missed one last year, North Carolina State.
They were perfect in 2013.
They missed one in 2012,

[Bill Marsh]

because the dance card inexplicably still uses the old RPI and with a team like a St John's- that makes a huge difference.
Old RPI 30
Current RPI 38

Evaluating St John's at RPI 38 is a lot different than at RPI 30. Here both would get in. But if it got to say 45 for old and 53 for current- that's a different animal completely.

Thanks. I'm not familiar with Dance Card.

[stever20]

California last year 56
NC State last year 51

current formula
nc State 55
cal 63

So while maybe not saying NC State should have gotten in, but any idiot would know a team with a 63 RPI has pretty much a zero shot of getting in at large..

So I'm sorry- but them continuing to use the old RPI does cause problems. It makes absolutely no sense that they don't change the RPI component from the old formula to the new formula- you would think that it would make the rating much stronger.

It's just tough to take anything seriously that uses a component that isn't what is used to make the decisions.

I mean a poster-child right now is Dayton... In the old formula that Dance Card uses, they have the #51 RPI. Because of that they have Dayton at #56- or the 10th team out. Dayton's real RPI is #35- with as a result they are #36 in Bracket Matrix right now. 16 spots in the RPI is just a HUGE gap.
(for those wondering why Dayton is so different)
15-0 at home
4-5 on the road
2-1 neutral
So right now only have 4 RPI losses, to go along with 16.6 wins. 16.6-4 makes an own winning percentage of .806....

[stever20]

I was curious so I actually emailed the guy who does the site...

Steve,

First: many thanks for your interest!

In short: although it’s odd to find it this way, the old version of the RPI is more strongly related to whether a team gets a bid than the current formula. In fact, once you know several other pieces of information as well, it’s not even close. So we’ve just stuck with the old one in our analytics.

Obviously, the committee could change this year and align more closely with the current version, but they certainly haven’t up to this point.

Jay

Conceptually I think the one thing that you could say about it is the committee takes a look at the teams overall record a lot more than we think they do. That's about the only thing that I could think of. Any thoughts?

[paulxu]

So I'm sorry- but them continuing to use the old RPI does cause problems. It makes absolutely no sense that they don't change the RPI component from the old formula to the new formula- you would think that it would make the rating much stronger.
It's just tough to take anything seriously that uses a component that isn't what is used to make the decisions.

I do not know exactly how they calculate their dance card. I believe the actual stuff is downloadable as a PDF, but you have to pay for it.

But I'm assuming they have a rather detailed formula, especially given their credentials:

B. Jay Coleman, Ph.D. is the Richard deR. Kip Professor of Operations Management and Quantitative Methods in the Coggin College of Business at the University of North Florida. He earned his Ph.D. in Industrial Management from Clemson University.

and

Mike DuMond of Economists Incorporated, and Allen Lynch of Mercer University.

I also don't know if their card is "if the selection was made today" or predictive of what they think will happen (and that changes daily).
But I do know that they have adjusted their formula for committee bias, successfully starting in 2012, and that they only missed 2 teams in 3 years...a fairly good track record.
You may not like their use of the "old" RPI calculation, but they aren't using it like the committee. They're using it in a formula.
A formula that has produced fairly good results...serious or otherwise.

[Omaha1]

Creighton AD Bruce Rasmussen is on the NCAA committee and he did a radio interview yesterday locally. He said he doesn't look at RPI, but does look at BPI, KenPom, and something else (KPI?) generated by some guy in Michigan.

[stever20]

Creighton AD Bruce Rasmussen is on the NCAA committee and he did a radio interview yesterday locally. He said he doesn't look at RPI, but does look at BPI, KenPom, and something else (KPI?) generated by some guy in Michigan.

The question is going to be is he in the minority or the majority.