Championship Week 2002

Message Board

ESPN Auctions
Friday, March 1
RPI doesn't tell whole story

By Jay Bilas
Special to

As the regular season winds down and Selection Sunday approaches, few terms of art in college basketball are as ubiquitous as "RPI" -- the short form usage for Ratings Percentage Index.

A team's RPI ranking tends to connote that team's worth in the eyes of the NCAA Tournament selection committee. It conjures up images of a team's I.Q., heart rate, Blue Book Value, cholesterol level, SAT score, growth potential, employment history, and inkblot test results all rolled up into one convenient, and easy to use number. The RPI ranking of a team is taken by some to be the gospel, the holy number that anoints the true contenders and awards them with precious bids delivered by the wise elders in Indianapolis. The RPI is thought to scientifically and sometimes magically separate those upon whom divine light shines and the unwashed masses of Division I basketball.

We should all be thankful for the RPI. Take, for example, the clear differences between contenders like Alabama and Oklahoma, who are both vying for a No. 1 seed in the NCAA Tournament.

The Sooners have a No. 5 RPI ranking, just ahead of 'Bama at No. 6. But, the telling formula of the RPI reveals the clear separation between the two teams: Oklahoma's RPI is 0.6491 to Alabama's 0.6427.

What could be clearer, or more revealing?

Or, consider Kansas' RPI is 0.0021 better than that of Maryland. Need we even play on from here?

I say, what is the RPI, anyway?

It must be terribly a complicated mathematical formula, worthy of a Nobel Prize. The RPI surely must take into account variables such as shooting percentages on the road and at home; offensive effectiveness against man and zone; and operating efficiencies in full court and halfcourt games.

Clearly, the RPI must factor in free throw numbers in the last five minutes of a ballgame, victory margins against common opponents, and performance levels attained with certain officiating crews. Tournament selection would demand no less than a formula that takes into account every reliable factor, and provides us with a scientific reading that sorts out these teams in a coherent fashion so that we are able to easily discern between those that are in and those that are out.

No such luck with the RPI, which is a stunningly simple formula.

A team's RPI is measured from the strength of a team's schedule, and how that team fares against that schedule. Because Kansas is currently the RPI's No. 1 ranked team, we will demonstrate with the Jayhawks.

One half of the formula is the average win percentage of the Jayhawk's opponents, one quarter of the formula is Kansas' win percentage, and one quarter is the average win percentage of all of the opponents of Kansas' opponents. That's it.

Now, the NCAA has some brownie points they add to this formula in order to claim that their RPI is some "double secret" version of the RPI. A good conspiracy theorist would surmise that this secret formula can only be for purposes of plausible deniability, or a good "CYA" for the committee, so it can say with a straight face that it had different information than we do, so could reasonably reach a different conclusion.

The RPI is not magic, but simply a good starting point to evaluate teams, and serves as a good indicator of where a team stands relative to its competitors. However, the RPI does not indicate how good a team is relative to other rated teams. The RPI is just a tool, and a useful one, but should not be relied upon to judge teams. Only a full review of a team and the details of its season should be counted upon to tell us just how good a team is, and whether a team should be invited to the dance, or asked to stay home.

The RPI does not take into account important factors such as whether a game was played on the road or at home; the margin of victory or defeat; or the circumstances surrounding each game. The RPI does not factor in the true strength of an opponent at the time the game was played, including injuries that may have affected the outcome of a game or games.

For example, Michigan State suffered five losses without leading scorers Marcus Taylor and/or Adam Ballinger in the lineup. Most of those five losses were all close, and either of those players would have made a significant difference in the outcome. Certainly, Michigan State's record would be different for those five games, as would the records of some of the opponents Michigan State lost to during that stretch, and the true quality of those losses should be factored in.

Another example would be the deflated value of a win over a team that was flying high at the time the game was played, then suffered a subsequent collapse. A game against any opponent counts the same in the RPI at the end of the season, without regard to how well the team was playing at the time of the game. Beating Iowa in December, when the Hawkeyes were playing top 10-caliber basketball, clearly has more meaning than beating the Hawkeyes in their mid-February tailspin. But the RPI does not take that into account.

Conversely, wins can be inflated as well. A win over Utah early in the season was a relatively easy task because the Utes were reeling at the time, yet by the end of the season, the RPI treats such a win as if it were a war.

Perhaps most important, the RPI does not give any credit for playing and beating teams on the road. Any person who has ever played or coached college basketball can tell you unequivocally that winning on the road is infinitely more difficult than winning at home. Gonzaga played 16 of its first 28 games on the road, and Syracuse played 16 of its first 28 games at home. Assuming the opponents were exactly the same, the RPI of both teams would be the same. That is not right.

While the tournament selection committee will fall all over themselves saying that the RPI is not as important in the selection process as people tend to believe, it can be quite helpful as a predictor of inclusion into the NCAA Tournament, and as a predictor of seed. By and large, if a team is ranked in the RPI top 40, it will receive an at-large bid. The best RPI ranking ever left out of the tournament was No. 33 Oklahoma in 1994. If a team is outside of the top 50 of the final RPI rankings, chances are it will be heading to the NIT. The lowest RPI ranking ever to make the NCAA Tournament as an at-large team was No. 74 New Mexico in 1999, then No. 66 Minnesota in 1995.

Oh, and if you're looking for this year's version of the 2001 Georgia Bulldogs, the team that got an at-large bid with a 16-14 record, there isn't one this season. Georgia played nine games against RPI top 25 teams -- winning four of those games -- and had eight wins against the RPI top 50. No bubble team in 2002 can match those numbers.

 More from ESPN...
Who's on the bubble?
Who's in, who's out? ...

Katz: Bubble Busters
Teams who could pop the most ...

 ESPN Tools
Email story
Most sent
Print story