December 22, 2004
RPI tinkering

Ken Pomeroy says that the NCAA is fixing to tinker with the RPI formula for ranking basketball teams. ESPN has the story.


Road wins will carry more weight in the Ratings Percentage Index after unanimous approval by the NCAA championships committee last month.

Iowa athletic director Bob Bowlsby, chair of the NCAA Tournament selection committee, said 70 percent of home teams win during the regular season, making it imperative to factor that into the power rating system.

"Winning on the road is going to be more valuable than a win by the home team," Bowlsby said. Bowlsby declined to specify how much of an impact road wins would have in the overall RPI formula.

[...]

"Go to the RAC (at Rutgers) and see how tough it is to win there," Connecticut coach Jim Calhoun said. "Rutgers is dangerous at Rutgers, but they don't have a great reputation (nationally). We played Rice last year down there in a tough game.

"It's going to help the teams like Rice, that comes here this year because if those teams are good enough to beat us here, then they're good enough to go to the NCAA Tournament, and they should get extra credit for that," Calhoun said.


Sadly, Rice lost at Connecticut by an 81-72 margin. They play at Syracuse today, though, so they'll have another shot at that extra credit Calhoun mentions.

Bowlsby reiterated that the selection committee uses the RPI as one of many factors and doesn't solely pick or seed teams on the numbers. He said eliminating the RPI wasn't discussed.

Some coaches think it should be -- and not just those whose teams land on the NCAA bubble.

"We shouldn't even talk about the RPI because it's not important anymore, it really isn't," Duke coach Mike Krzyzewski said.

"We're going to see that the RPI will be like those eight-track tapes," Krzyzewski said. "It served its purpose. You shouldn't use the RPI to make scheduling moves. You should just play good teams. This isn't a formula game. It's a tournament game. We determine a champion. We finished in the top of the RPI in three of the last six years and have only one national championship. If it were football (with the BCS), we'd have a couple more."


Yeah, but how do you know who the good teams really are without some kind of objective system? I mean, Texas A&M is 8-0 right now against a schedule so loaded with creampuffs even Bill Snyder would be embarrassed. And accordingly, they're ranked #222 by RPI. If you didn't have RPI, you'd have to look at the opponents they've played to understand their record so far, and once you've done that, you're 75% of the way to using the RPI formula anyway. What would you use in its place, Coach K?

Anyway. Ken shows that by tinkering with only a part of the RPI formula and not all of it, you will get some unintended consequences. Read his analysis for more.

By the way, as long as we're speaking of formulas used to determine postseason selections, I note that the Associated Press poll will no longer be used in the BCS rankings.


Rather than use another poll in the formula, a BCS official said Tuesday night that the BCS commissioners are considering appointing a blue-ribbon committee of athletic directors and other executives to name the teams that will play in the national championship game.

The concept, although foreign to college football, is not foreign to the NCAA as a whole. Many NCAA postseason playoff participants are named by committees, including March Madness.


Boy, that ought to quell all those championship game complaints. Hold onto your big foam fingers, it's gonna be a bumpy ride.

Posted by Charles Kuffner on December 22, 2004 to Other sports | TrackBack
Comments

The problem with NCAA Basketball changing the RPI system is that there are simply too many teams in Division I to allow for any other method. I don't have a problem with providing a higher value for road wins if the home team wins a large percentage of its games in their building. If a team is 4-21, however, a road win should not provide any meaning whatsoever to a visiting 25-0 team.

" Hold onto your big foam fingers, it's gonna be a bumpy ride."

Of course, in New York, we don't have to buy fingers. We gladly give them to you for free. :-)

Posted by: William Hughes on December 22, 2004 10:11 AM

I just found out about the change in the way the NCAA calculates the W-L percentage for the RPI values.

The teams adversely affected by this will be teams that dominate in their building, regardless of that team's road record. My team, Kentucky has historically won 90% of its Rupp arena games and a similar % for 25 years in Memorial Coliseum.

The explanation offered by Iowa athletic director Bob Bowlsby, chair of the NCAA Tournament selection committee, that 70 percent of home teams win during the regular season, making it imperative to factor that into the power rating system concerns me. Should Kentucky only receive credit for 0.1 win for each home victory because it wins 90% of its home games?

I would love to know the rationale the committee used to select a factor of 1.4/0.6.

On another related, but tangential issue, I have recently been looking at the RPI SOS values to compare the season performances of teams. Has anyone else looked at the SOS values, and do you think it is an appropriate measure?

Glad I found this site.

Posted by: Richard Cheeks on December 24, 2004 10:08 AM

Richard, the rationale of the committee using the factor of 1.4/0.6 is merely a function of the fact that they have ascertained that there have been 70% home wins over the past 4 years. For years they have used the same weight for home and road games (1.0/1.0), thus giving each 50% of the RPI equation. If you take 1.4 and divide into 2.00 (which is 1.4 + .6) you get the win % of all home teams for the past 4 years (70%)as gleaned by the committee. For example if the committee determined that home teams had won 80% of all of their home games, this factor would have been 1.6/.4.

Is this a step in the right direction? Probably is for now, at least it somewhat makes sense. Also the middle of the road teams from the power conferences seem to be hurt by this since most schedule their non conference games at home against mediocre opposition. By scheduling inferior opponents not only improves their own W-L record but also improves the other teams RPI's in their conference from the schedule strength factor.

The inherent problem with the good mid major teams is that they can't gain much in the RPI once they start playing conference games. Their conference opponents by playing non conference games on the road and losing lower their own W-L record thus affecting the better mid major schools that play them.

A mid major school in a mediocre conference would basically have to sweep the card to have any chance of attaining a high RPI rating.

Enjoy March Madness!!!

Posted by: Randy Rueger on February 20, 2005 10:38 PM

Although some could quibble about the loading, the new formula will hopefully encourage those middle teams in power conferences to go on the road to play some of the mid-majors. While the mid-majors are hurt (or at least not helped) by their conference play, the real hurt has come from being forced to go on the road for ANY games involving a major conference - they simply won't come to our house (my Salukis have only lost 1 or 2 games at home over the past 3 years!). Southern Illinois is 0-4 against RPI teams with a ranking of 1-50, and ALL of those have been road games.

With the changes to the number of preseason tournaments we can attend, some years it's impossible to find anything but a road game against a power conference team. So, hallelujah for the RPI change. Although I don't think SIUC is really a 13th ranked team, at least that RPI should make it easier to get in without the annual worries when we fail to win our conference tournament.

Posted by: Mike Young on February 21, 2005 10:24 PM