Ides of March: RPI, Subjectivity, and the Selection Committee

I like the Rating Percentage Index. I think RPI is a useful tool as it minds its own business mathematically portraying how college basketball teams stack up against each other. It has even been tweaked recently to reward away wins more than home wins, while doing a better job to slightly disregard away losses and make home losses hurt more. But how useful is RPI in both determining and predicting what teams will be allowed to play for a national championship come March?

Over the last few seasons, when it comes to selecting the field of 65, the selection committee has shown that subjectivity reigns over the math. Is that a bad thing? That depends on who you talk to. The committee certainly has to be subjective to differentiate between teams from tougher conferences and weaker conferences, along with comparing quality wins and atrocious losses. However, some of the decisions from the selection committee have been so subjective that it appears the RPI is about as useful to them as knowing Latin.

The 2006 NCAA Tournament featured four teams from the Missouri Valley Conference, two of which advanced to the Sweet 16. The mid-majors have arrived! Four teams came out of the MVC that year because the selection committee had absolutely no choice than to take three at large teams. The RPI of the four teams was 25, 27, 29, and 33.

A year later a team from the MVC had an RPI of 36 and a strength of schedule (SOS) of 21. The MVC was ranked sixth in conference RPI, ahead of even the Big 12. Those raw numbers would give any team, especially a team coming off a Sweet 16 appearance the year before, a good shot at making the tournament. No dice. Meet the Bradley University Braves, who with an RPI of 36 was actually the second best ranked RPI team left out by the selection committee. Is there a reasonable explanation for this? Yes. Let’s take a look at the three best RPI teams left out in 2007:

Air Force (22-8) –RPI 30, SOS 80
Bradley (21-12) – RPI 36, SOS 21
Missouri St. (21-11) – RPI 37, SOS 42

Not terrible right? Now their respective records against the RPI Top 50:

1-4
2-8
3-5

The teams didn’t cut it against top-tier competition, and they only have themselves to blame for that right? Now lets look at another team that made it into the tournament via an at-large bid:

Stanford (18-13), RPI 65, SOS 32

Okay, the Cardinal had a decent, at best, RPI and a SOS similar to the three snubbed mid-majors. Now let’s take a peek at how Stanford’s record against the RPI Top 50 set them apart from the top teams that didn’t make the tournament:

4-8

See? Wait a minute. 4-8 against the RPI Top 50? Is that really more impressive than the unimpressive performances by the mid-majors? Yes, the Pac-10 is a better conference, but to get into the tournament with less than 19 wins, more than 12 losses, and an RPI over 60 without being a conference champion (and with a 4-8 record against the top 50) is well…a little shocking. But wait, did Stanford finish strong? Maybe the committee took that into consideration. Nope. Stanford finished the season 1-4, the lone win coming over an Arizona St. team that finished 7-22. Remember Air Force? Now take a look at this score from a game in Palo Alto from that same 2007 season:

Air Force 79 Stanford 45.

That pretty much speaks for itself. That’s just being done in by both the math and subjective committee member opinions who overlooked that score. Now I don’t mean to pick on Stanford, I have nothing against them. So let’s take a look at a few more examples, this time from 2008, of what I call committee over-subjectivity (records against RPI Top 50 in second parentheses):

In
Kentucky (18-13) –RPI 57, SOS 19 (4-6)
Oregon (18-13) – RPI 58, SOS 37 (4-9)

Out
Dayton (21-10) – RPI 32, SOS 33 (4-4)
Illinois St. (23-9) – RPI 33, SOS 71 (2-5)


The trend seems to be that snubbed teams are from smaller conferences and are overtaken by big name schools from better conferences, but certainly not much better resumes. But what about when power conference teams are compared? Two of the following three teams made the tournament in 2008:

Ole Miss (21-10) – RPI 48, SOS 65 (5-3)
Kansas St. (19-11) – RPI 50, SOS 34 (3-6)
Villanova (20-12) –RPI 51, SOS 47 (4-7)


If you guessed, remembered, or looked up that Ole Miss was the odd team out, you’re right. The difference? Ole Miss went 7-9 in the SEC, Kansas St. went 10-6 in the Big 12 and Villanova went 9-9 in the Big East. So it’s not so much what your RPI is or how well you do against top tier RPI teams, but in this case almost certainly how well you do in your league. And for the record, I don’t think a team that is under .500 in conference play should be in the tournament (of course the Big East this season will probably have me backtracking on this statement). This is an example of good committee subjectivity, or at least logical subjectivity.


Now, selection committee subjectivity isn’t just a game of who’s in and who’s out. It also comes into play with seeding. Time for some more hands-on fun. You’re on the selection committee and the committee has decided that these two teams are in the tournament. It’s now your turn to seed them (again with the Top 50 records):

26-6, RPI 7, SOS 30, (8-4)
20-10, RPI 55, SOS 49 (7-4)

If you think both of these teams should be #4 seeds, you too have the ability to be as subjective as the selection committee. Meet Southern Illinois (RPI 7) and Virginia (RPI 55) from 2007.

Subjectivity isn’t bad. Subjectivity can be useful. It can certainly differentiate between the likes of Ole Miss, Kansas St., and Villanova and it can help decide between teams with similar resumes.

Sometimes, though, subjectivity from the committee gets to be a bit much, like when well known names like Kentucky and Stanford are sneaking in the tournament when more deserving but lesser known schools have to go host an opening round game in the NIT. When subjectivity puts teams separated by 48 spots in the RPI ranking as equals, it means that those taking the RPI rankings as a guide really don’t take the RPI rankings seriously. If this trend continues, unfortunately, for the sake of staying on the same page as the selection committee, it may get to the point where neither can we.

Keep all this in mind as the college basketball season progresses. It’s not flawless, but I’m a fan of the RPI system. If you’re using the rankings to predict at-large bids, even as part of a bigger team resume, take them with a huge grain of salt; because the selection committee, after all, has been unpredictable.

This Week's Notable Performers

Luke Harangody (Notre Dame) – Harangody led the Irish with 25 points (including a 3 pointer), 16 rebounds, and 6 assists against Syracuse on Sunday. Unfortunately for him, Kyle McAlarney was the only other Irish player to show up in the 19 points loss.
Syracuse – Four players had 14 or more points in an all around 93-74 dismantling of Notre Dame on Saturday. Conversely, only two players reached double digits in their 18 point loss to Pittsburgh on Monday night.
Ryan Tillema (Wisconsin - Green Bay) – The Phoenix's leading scorer performed well in two bigs wins over Horizon Conference foes, scoring 24 points in a two point win over Wisconsin-Milwaukee, and then putting up 19 points and 10 rebounds against Cleveland State.
Lawrence Westbrook (Minnesota) – Westbrook scored 29 points, including a game tying three to send the game to overtime, as Minnesota came back from a 9 point deficit with three minutes left in regulation to earn a win at Wisconsin last Thursday.

Related Posts by Subject