Zips RPI Questioned

[Captain Kangaroo]

For the record...I agree with the guy. Our RPI seems too high for our record and caliber of opponents.

************************************************************************

Fuzzy math shrouds NCAA hoops

By JOHN KNOX

Published , March 09, 2005, 06:00:01 AM EDT

It's that time of year, when a sports fan's fancy turns to ... math.

Mathematics helps determine which teams get invited to the Big Dance, the NCAA men's basketball tournament. Here's how it works: Unless your team gets an automatic slot (generally by winning a conference tournament), you need an at-large bid.

The NCAA handles at-large bids the way colleges do admissions -- the insiders repeatedly reassure applicants that they aren't reduced to a number.

But in the end, it's all about the numbers.

With college admissions, the all-important number is the SAT. The NCAA selection committee uses the Ratings Percentage Index, or RPI. A low RPI, and you're in the tournament as an at-large team. A high RPI, and you're not.

Like the SAT, the calculation of the RPI is a very ad-hoc, unjustified and unjustifiable process.

The RPI formula is, according to the Web site (www.collegerpi.com):

-- 25 percent team winning percentage

-- 50 percent opponents' average winning percentage

-- 25 percent opponents' opponents' average winning percentage.

For the 2004-05 season, the formula was changed to give more weight to road wins vs. home wins.

I'm blowing the whistle on this silly RPI formula.

First of all, the RPI decouples a team's record from its performance in a game and from its opponent. If your team wins by 30 points on the road at Duke, it has the same impact in terms of

winning percentage as a 1-point win at lowly Savannah State.

Second, who decided that a road win is worth 40 percent more than a neutral-court win and should count 133 percent more than a road loss?

That's excessive, given that statistics from other more sophisticated ratings schemes (see below) indicate that home court advantage is about four points.

In effect, the RPI paradoxically rewards losing on the road and winning at home. Proof from Tuesday's computer polls:

Pittsburgh is 20-7 in the powerful Big East Conference and is a Top 25 team according to the human polls.

But the Panthers are only No. 41 in the RPI. Why? Pittsburgh played only nine road games (but went 6-3) and earned one neutral-court win. Their actual winning percentage is .740, but by the RPI formula it's reduced to .699.

The Panthers' home losses to probable Big Dancers such as UConn, West Virginia and Georgetown are killing them in the RPI. One spot ahead of Pittsburgh in the RPI at No. 40 is Akron.

The Akron Zips? They're 17-9 and they play in the solid Mid-America Conference. But how is that better than 20-7 in the Big East? Easy, if you're the RPI: Akron played 13 road games (they went 5-8), half of their schedule. The road losses are minimized by the RPI. Meanwhile, the Zips' road wins are magnified even when they're against dogs like Marshall and Duquesne.

And so Akron's actual winning percentage of .667 is boosted to .705 by the RPI formula, just ahead of Pittsburgh's.

Winning percentage is only 25 percent of the overall RPI formula, true. But because the other 75 percent is based on other teams' winning percentages calculated by the same lousy formula, even a few fouled-up ratings contaminate the whole scheme from top to bottom.

There should be a better way to quantify the quality of a basketball team.

There is.

More sophisticated algorithms have been devised that know when you beat a tough team on the road or lose to a crummy opponent at home, and which use advanced statistical methods to estimate a team's quality.

One of the pioneers of this approach is Jeff Sagarin, an MIT graduate whose ratings are found in USA Today.

In Sagarin's latest ratings, Pittsburgh is No. 27, Akron is No. 74. That makes a lot more sense to me.

But on Selection Sunday, remember that dumb math may be behind some of the poor decisions of the NCAA selection committee

[KingZip]

You are 100% right - who has this team beaten?

Going 11-7 in the MAC is better for Akron U but is no great shakes for national recognition.

[Dr Z]

How the selection committee works