rpi past its prime...?

[johngateswhiteley]

http://insider.espn.go.com/espn/blog/index?entryID=3225609&name=bilas_jay&action=login&appRedirect=http%3a%2f%2finsider.espn.go.com%2fesp n%2fblog%2findex%3fentryID%3d3225609%26name%3dbila s_jay

RPI is past its prime
posted: Friday, February 1, 2008 | Print Entry

Blog items for Friday:

Numbers game: It's starting already … we've now begun to simply count wins against the RPI rated teams as the best evidence of how good a team is. Are we not smarter than that? The RPI is unnecessary and past its prime. If you actually watch all of these teams play (and the committee says it is doing that), then why do we need the RPI? It presents a certain perception, and that perception is not always correct. I believe that the selection of teams has become formulaic, and I'm not crazy about it. If you think that Team A is better than Team B, then Team A should go into the field first. You don't need to count wins against RPI Top 100 or RPI top 50 competition. The only objective participants in the NCAA Tournament are the automatic qualifiers (the teams that win their conference titles). Everyone else is in a beauty contest. But, based upon RPI numbers, we are all assuming that Drake is in the field -- barring a collapse. But, if Purdue has played the same schedule as Drake and was 16-1, do you think anyone would be saying that the Boilermakers would be in? I doubt it. I think we would be questioning the schedule and whether Purdue had beaten anyone. Watch Drake play, and decide whether the Bulldogs are among the best 34 teams after the automatic qualifiers. That is plenty good enough. I think we need a new measure, and we need to get away from the RPI. It doesn't work, and it never did. It is a crutch, and we can do better without it.

[Extra Stout]

I don't need no stinking RPI. My own eyes tell me that all 12 ACC teams should make the NCAA Tournament.

[Ed Helicopter Jones]

I like how Jay offers so many alternatives.

There actually have been some people who've tried to improve the "accuracy" of the RPI by building margin of victory, among other variables, into their models. There are some pretty decent computer rankings out there.

I think you need to use common sense, but at the same time, you need some sort of ranking that won't build in too much bias. Without referring to the RPI, or some similar formula(s) to rank teams you'd have an NCAA tournament with 55 teams from either coastline and all the at-large bids going to major conferences....oh wait that's what we've got now.




Never mind.

[MajorMike]

And the BCS sucks too but I don't see ether going anywhere anytime soon.

[RonMexico]

And the BCS sucks too but I don't see ether going anywhere anytime soon.

Wait, wait, wait.... the guy who began the RPI obsession on this forum now says it "sucks"?

Are you bi-polar or do you just think RPI is worthless when your team sucks balls and is now 0-7 on the road this year?

[MajorMike]

Ah, ronnie; I can't even read what he says and yet it is certain that his obsession with me continues. I bet if you did a count, that over 85% of your posts iin the college forum are directly aimed at and posted directly after me. Such a sad little man.

[RonMexico]

Again - the guy who will post about OSU equestrian teams and A&M's RPI but disappears when his team loses in Stillwater calls me "obsessed."

I believe I labeled you as the obsessed first and then you blocked me. The irony is incredible.

[K-State Spur]

The RPI is fine so long as it is used as just one of many tools and not an end-all-be-all forumula like the BCS.

[MajorMike]

If you think about it, the RPI is pretty spot on. When the highest ever to be left out of an at-large bid is 34-ish and the lowest ever at-large was 66-ish, tells you that something is working. There are times when some teams fare better because of odd reasons, but it comes down to wins and SOS for the most part. Tenn will end up in the top 5 or so in RPI even tho we all know they won't get a top 4 seed. Drake will be in the top 10, however I doubt they get a top 2 seed.

Right now, the 1-seeds and fairly set if the teams play out like they should. Memphis and the ACC winner will fight for the overall #1 seed. KU looks to get one, and the remaining one will be between UNC/UCLA/Geo. The #2 seeds look to be Tenn, the 2 of the 3 mentioned before that don't get a 1, and probably the B10/1 winner (Wisc or Indy). There is an instance where the RPI does not work. Indy could very well win the B10/1 and have around a 20 RPI, but they will still get a #2 seed because they are the last winner of major conf. The only ailibi for a conf getting a 2nd in before them is UNC and possibly UT (if they win out) and/or UConn (2nd in BE who will get 8 in most likely.

After you seed the top 4, which is the important part because it tells you which high seed get chosen for advantageous sites, there is an overall 1-65 seeding of every team in the show. For instance, the worst #1 seed gets the best #2 seed. And so on. Seeding is partly RPI and partly conf finish. If KSU, BU and atm all pretty much win out with no bad losses, KState will have a worse RPI that BU and atm but will most likely be seeded ahead of them, especially if they finish 14-2 in the conf and beat UT at home.

People have to realize that RPI is a great predicting tool, not the end all means to seed the tourney (no matter what lies jgw tries to say I said). The RPI is actually much better than the BCS, because it has a human backup.