March 12, 2007 (Selection Notes)

With the announcement of NCAA tournament selections, it's time for everyone's favorite game: second-guessing the tournament committee. Let's take a look at assembling a field, and see how the actual tournament compares. For starters, we have 31 conference champions, all of whom are given automatic bids. This leaves 34 openings.

Next up, we'll apply the rule that all teams with RPI rankings in the top 32 will be taken. This has been true of every tournament we have examined. Here we find our first discrepancy: Air Force had an RPI of #30 and was omitted. I'm not sure what to make of this. Air Force had a solid (23-8) record, and went 10-6 in the Mountain West. However, their #56 "Dolphin ranking" certainly indicates weaknesses in the schedule, something the committee may have seen as well. A more cynical explanation would be to point out that the committee didn't give more than two berths to any non-BCS conference, and that Air Force would have been the third selection from the MWC. At any rate, taking the RPI top-32 teams other than Air Force, we've added 22 more teams, leaving just 12 openings.

The next set of teams to push in by "historical precedent" would be teams from the big nine conferences (BCS plus the A10, C-USA, and MWC) that finished in the top 50 of the RPI and won 20 or more games. Seven teams meet these criteria, and six were selected -- all but Syracuse. Syracuse would have been the seventh Big East team selected, which doesn't seem like too many for a 16-team conference. Nor were they bad by Dolphin Ranking standards, coming in at #45. The only likely reason for this omission is that they were indeed right on the borderline, with an RPI ranking of exactly #50, so let's drop them into the "rest of the bubble" category.

The first team to pick out of the bubble is Virginia, whick ranked #55 in the RPI but finished second in the ACC. The committee generally tries to go in order of conference finish, so Virginia had to be taken.

Searching the teams still available for those ranked in the RPI top 68, and with a conference record within a game of .500, there are 25 teams to pick from for the final five spots. The best three, according to the Dolphin rankings, were all from the ACC. To end up with the same set of five the committee selected requires some gymnastics:

First, I have to eliminate any teams from non-BCS conferences that would be the third selection from that conference. This is the only consistent criteria with which I can eliminate Missouri State (RPI #36, Dolphin #40), Bradley (RPI #38, Dolphin #54), Air Force (RPI #30, Dolphin #56), Drexel (RPI #44, Dolphin #58), and Utah State (RPI #45, Dolphin #80). This also eliminates UMass and SDSU, bringing us to 18 teams for 5 berths.

Next, I'll look through the RPI rankings of the remaining teams, and select Arkansas (#35) and Old Dominion (#39), the two best by a wide margin. Now we're down to 16 teams for 3 berths.

Next, I have to look at the total tally per conference already selected: 6 ACC, 5 SEC, 6 Big 10, 6 Big East, 3 Big 12, and 5 Pac-10, if looking in order of conference strength. Let's restrict this to six teams per conference, plus one for the ACC. With the bias against teams under .500 in conference, I'll drop Clemson and FSU, keeping Georgia Tech in the running. Doing the same for the SEC, we'll drop Alabama (for its 7-9 record) and pick Mississippi from the three 8-8 teams for having the highest RPI (as well as the highest ranking in my system). This also eliminates Big East teams West Virginia, Syracuse, and Depaul, as well as Big Ten team Michigan. This leaves seven teams competing for three spots.

Of the remaining teams, Georgia Tech has the highest RPI (#52), as well as by far the highest Dolphin ranking (#30), so we'll take it. The next two-highest ranked teams in both systems are Kansas State and Texas Tech, either of which would become the fourth Big Twelve team selected. From these two, the committee went with Texas Tech, despite having a worse conference record and having been eliminated from the conference tournament by Kansas State in a 66-45 rout. Texas Tech's only edge was a #53 ranking in the RPI, versus #56 for Kansas State. With these two selections, we have one spot and five teams.

Dropping the clear-cut worst of these teams (Akron), this leaves four teams up for consideration:

22-11 10-6  47  56 Kansas State
20-12  8-8  59  58 Mississippi
25-7  15-3  60  62 Appalachian State
18-12 10-8  63  65 Stanford

Kansas State has the best ranking all-around and would be just the fifth teams selected from a 12-team conference, so the obvious pick, right? Nope. OK, what about Mississippi? They won 20 games, would be the sixth team from the second-best conference (Stanford would be the sixth team from the weakest BCS conference), and again have the ranking edges all the way around. Again, nope.

OK, so between Appalachian State and Stanford, we have two teams in the 60's in the rankings, but Stanford would be the Vegas favorite in a matchup and the committee has been unkind to the midmajors, so we'll finally make the committee's selection with Stanford. Still, the question remains how any person -- let alone a committee -- could rationally conclude that Stanford is more deserving of a tournament bid than Kansas State or Mississippi.

The glaring issue this year was in the treatment of non-BCS conferences. (I won't call them midmajor, since the Mountain West, A10, and Conference USA aren't always classified as midmajors.) Between #30 and #68 in the RPI rankings, there were 23 teams from BCS conferences and 10 from non-BCS conferences. Just over half (12) of the BCS teams were selected, including RPI #53 Texas Tech and RPI #65 Stanford. And, only two of the non-BCS teams were selected -- two teams in the 30's of the RPI. Looking at it differently, the three highest-ranked teams left out, and five of the top six, were non-BCS teams. Using another ranking system, such as the Dolphin rankings, in place of the RPI makes the discrepancy less dramatic but doesn't make this go away.

The other big mystery was the gift to Stanford, which with an RPI of 65, was ten points worse than the next-worst bubble selection (Virginia).

Return to ratings main page

Note: if you use any of the facts, equations, or mathematical principles introduced here, you must give me credit.

copyright ©2007 Andrew Dolphin