Advertisement

Monday, March 9, 2015

Keegan

Column: RIP RPI; long live TVM

Advertisement

The RPI, which some in college basketball like to use as an indication of a team or league’s strength, stands for Really Poor Indicator.

So it matters not what the Big 12 or any other league is ranked in it. The truth is, nobody ever has come up with a worthwhile indicator of a conference’s relative strength.

So everybody guesses.

“I think this year, nationally, the perception is that it’s great,” Kansas University basketball coach Bill Self said of the Big 12. “I think the perception is it’s not a top-heavy league, it’s a parity league, which I think is fair. The difficulty in our league is that there are only 10 teams. The reality in some leagues, you may have 14 or 15 teams, so you have five or six, really, really, really tremendous teams at the top, but you could have four or five teams that maybe aren’t as strong toward the bottom.”

No system addresses that disparity of league sizes.

Until now.

It’s called Tournament Victories per Member, TVM for short. Don’t feel bad if you haven’t heard of it. I just invented it. Abandon all acronyms, save TVM.

The formula is as simple as its inventor: After the Final Four, divide the number of NCAA Tournament victories for a conference by the number of members in that league.

Looking back at the seasons since Self has coached at Kansas, the Big 12, with .651 NCAA Tournament victories per conference member, ranks fourth among six power conferences, behind the Big Ten (.742), Big East (.706) and ACC (.679) and ahead of the SEC (.588) and Pac-12 (.526).

The Big East has lost key members, so it’s reasonable to believe the Big 12 has moved ahead of it into third among Div. I basketball conferences, at least for as long as the leagues remain so aligned.

In Self’s 11 completed seasons, the Big 12 had the highest TVM in 2008 and 2012, both times with one tournament victory per member, 12 in 2008, 10 in 2012. The Big 12’s worst performances came in 2006 (next-to-last) and 2013 (last).

NCAA Tournament victories per conference members from 2005 through 2014

1 - Big Ten .742;

2 - Big East .706;

3 - ACC .679;

4 - Big 12 .651;

5- SEC .588;

6 - Pac-12 .526.

Best conference performances in past 12 years, based on NCAA Tournament victories per member school

1 - 2004 ACC 1.66;

2 - 2013 Big Ten 1.17;

3t - 2005 ACC 1.09;

3t - 2005 Big Ten 1.09;

5 - 2006 SEC 1.08;

6 - 2009 Big East 1.06;

7t - 2008 Big 12 1.00;

7t - 2012 Big 12 1.00;

7t - 2007 Pac-12 1.00;

10t - 2012 Big 12 .92;

10t - 2009 Big 12 .92.

Comments

Suzi Marshall 5 years, 2 months ago

I'll be very interest to see how the Big 12 does this season because form the looks of several mock brackets Kansas, OU, ISU, Baylor should do very well. WVA and UT have the personnel to create some early upsets.

Joe Ross 5 years, 2 months ago

Im not saying the RPI is wonderful, but at least it's not a lagging indicator like the one Tom introduces us to. The TVM gives us a picture of the strength of a conference last year. Personnel change from year to year. Coaches may as well (not only with respect to what schools they coach for, but also in terms of offensive and defensive strategy from year to year, some of which is based on the personnel they have). Players mature. Playing sites change (not so much in the Big 12 but in larger leagues you may get one team on your floor one season and play them on theirs the next). Lineups may change. Kansas, for example, was a different team last year than it was this year. The year before that it was different than a year ago. From year to year, there is even movement among conferences as a whole in terms of how their strength is viewed by analysts and coaches. The conference is the sum of its component parts, and if all of its teams change annually then theres an evolution in the league itself. That variability greatly compromises any indicator which is a lagging statistic, especially one that lags from season to season. It could be said, I suppose, that the RPI and BPI do the same within every academic year, since they take into account games at the beginning of the season even on Selection Sunday. I grant that much is bound to change for a team from the beginning to the end of the year, but because every team experiences that phenomenon you can factor it out. Besides, at least you're dealing with the same personnel, so youre not comparing apples and oranges. No. A TVM statistic from the year before may be helpful in giving some retrospective concreteness to conference by conference evaluation; however it is not a tool that can give you a snapshot in real time. And at the end of the day, you want an indicator that can do the latter, because they are used in establishing seeding based on the current season's performance. Specifically, RPI and BPI may help to answer the question of "how many teams should (this particular conference) get in?"

Benz Junque 5 years, 2 months ago

Since he said "The formula is as simple as its inventor: After the Final Four, divide the number of NCAA Tournament victories for a conference by the number of members in that league." I would assume this means that it is based on teams that made the tourney THIS year, not last year.

Joe Ross 5 years, 2 months ago

You may be right. But if this is the case then the TVM, unlike the RPI, has no predictive value. And since the TVM and RPI were being compared, I would fail to grasp the point of it. Now one could argue, I suppose, that the only thing being appreciated is a retrospective look at the strength of a conference; in which case I would not argue that the indicator could prove useful. But again, the use of the RPI and BPI, Kenpom, and other indicators is in comparing teams and conferences so that their placement in the March tournament field might be more accurately judged. It would make no sense in that case to compare the TVM and RPI. What are your thoughts?

Greg Woolen 5 years, 2 months ago

Just curious how "TVM" has anything to do with a conference's relative strength or how much parity there is in a particular league in a particular year? Historically, it tells you how well a conference has fared in the tournament but that isn't what HCBS was even talking about when quoted in this article.

“I think this year, nationally, the perception is that it’s great,” Kansas University basketball coach Bill Self said of the Big 12. “I think the perception is it’s not a top-heavy league, it’s a parity league, which I think is fair. The difficulty in our league is that there are only 10 teams. The reality in some leagues, you may have 14 or 15 teams, so you have five or six, really, really, really tremendous teams at the top, but you could have four or five teams that maybe aren’t as strong toward the bottom.”

So how do we use TVM exactly to either support or refute his claims regarding THIS YEARS Big XII conference?

Robert Brock 5 years, 2 months ago

I can see the Big 12 faring poorly in the Dance due to one factor: no bigs.

Cameron Cederlind 5 years, 2 months ago

So let me get this straight, the years that KU had 6 NCAA victories (2008) and 5 NCAA victories (2012) were the years the big 12 had the best TVM? Take away KUs victories in the post season and I'm guessing the big 12 is dead last most seasons.

Keith Gellar 5 years, 2 months ago

you needed this article to figure that out?

Cameron Cederlind 5 years, 2 months ago

I guess my point was that the Big 12 (outside of KU) hasn't done anything in the tournament in over a decade.

Tom Jones 5 years, 2 months ago

So we should judge a team's strength based on the strength of their conference in previous years?

Got it.

JJ West Er House 5 years, 2 months ago

Keegan's conclusion and analysis, per usual, is incomplete.

He asserts that this is a superior ranking system but fails to take into account the fact that it's (a) a lagging indicator and (b) subject to outliers, giving disproportionate weight to tournament wins and thus rewarding top-heavy conferences where the same two or three teams make deep runs to offset their middling-team first-round exits.

If the question is "Which league is the best today?", his system doesn't (and can't) answer that. But if the question is "Which league has been the best in recent years?", he STILL comes to an incomplete conclusion -- and, I'd argue, the wrong one -- based on his data.

The way Keegan looks at it, the conferences with the highest aggregate TVM are the "best" conferences -- and, if all you look at is "NCAA Tournament victories per conference members from 2005 through 2014", he posits that the Big 12 is 3rd or 4th behind the ACC and Big 10 at least.

But, you can parse the data differently -- and if you look at the second column "Best conference performances in past 12 years, based on NCAA Tournament victories per member school", the Big 12 makes 4 appearances. It was the best conference 4 years out of 12. The ACC only makes 2 appearances. So too does the Big 10. The Big 12 is the best conference every 3d year. The ACC and Big 10 might be second or third best more years than the Big 12 the years they don't win it . . . . but close only counts in horseshoes.

TL/DR: Keegan is using the numbers only partially in context.

Benz Junque 5 years, 2 months ago

This entire article is ridiculous. For the sake of accuracy, the Horizon Leaguie must be thee greatest conference ever since they had a TVM of 5.0 in both 2010 and 2011 after their one team, Butler, won five games each year before losing in the championship game.

Basically, all Keegan is doing is trying to argue that tourney performance determines how good a conference is, not regular season performance and then creating new stats to support his argument. Pretty typical.

Aaron Paisley 5 years, 2 months ago

You just failed at applying Keegan's formula. It's not wins per team in the tournament, it's wins per total teams in the conference. Butler's 5 wins would be divided by however many teams were in the Horizon league those two seasons.

Let's say the Big 12 wins 12 games this tournament, that would give the Big 12 a 1.2 rating because you divide the 12 wins by the 10 teams total in the Big 12.

I agree it's flawed like RPI in determining who the best league is though.

Benz Junque 5 years, 2 months ago

That makes it even more worthless. And who knows where he got his "stats". The 2004 ACC had 9 teams in it, 6 of which made the tourney. Those six teams, Duke, NC State, Wake Forest, Ga Tech, Carolina and Maryland went a combined 14-6, which the last I checked 14 divided by 9 = 1.56, not 1.66. If the only one I checked is already wrong then I doubt the accuracy of any of it.

Commenting has been disabled for this item.