03.152011 NCAA Men’s Basketball Tournament: SportsMeasures vs. the NCAA
I wonder if the NCAA Selection Committee knows something about estimating ability of athletic teams that I don’t? Don’t get me wrong, I’m not insulting the committee of men who gathered in a hotel conference room last weekend to select the 68 teams for this year’s Men’s Basketball Tournament. However, I am asserting my knowledge about objective, scientific measurement and how it is applied to rank ordering sports teams.
The Field of 68 has left me scratching my head for a few different reasons. Let’s start with the most pressing issue that comes up each Sunday night after the field is announced: who barely got in and who was left on the outside looking in through a knot hole in the wooden fence.
In SportsMeasures’ Field of 68, the last four teams in should have been New Mexico (21-12, No. 47), Colorado (21-13, No. 48), Penn St (19-14, No. 49), and Georgia (21-11, No. 50). These last four teams in would have been seeded twelfth in the new “First Four” format.
According to SportsMeasures the first four out should have been UCLA (22-10, #51), UAB (22-8, #52), Tennessee (19-14, #56) and USC (19-14, #61).
If you have visited SportsMeasures website, SportsMeasures.com, you will know that winning percentage is meaningless to our data analysis methodology. The actual game results drive the analysis. Whereas RPI (ratings percentage index) is 25% team winning percentage, 50% opponents’ average winning percentage, and 25% opponents’ opponents’ average winning percentage and uses only their opponents and their opponents’ opponents, SportsMeasures uses the whole data matrix to construct their rankings. RPI has a very limited degree of interaction between teams and is unreliable.
Using the NCAA’s methodology, it is difficult to argue with the inclusion of UCLA and UAB, after all, they are only .40 points (SportsMeasures) behind Georgia for the last spot in the tourney. OK, so those two teams are out of the argument.
Even including Tennessee isn’t too much of a stretch since only Washington State and Nebraska were both skipped over. So, I give the selection committee a pass on three of the teams they included. Isn’t that nice of me?
After that though, I have a tough time thinking that even their methods would show that the other three teams merit inclusion in the tournament. Southern Cal (19-14, No. 61), UTEP (25-9, No. 72) and VCU (23-11, #76) are just too far afield to even consider. On top of this, the NCAA gave USC and VCU eleven seeds, Tennessee a nine and UCLA a seven! Was that in honor of John Wooden or something? In SportsMeasures’ tournament (if they got in) UCLA would have had a thirteen seed.
OK, I think I made my point on that subject.
Before I dive into the next topic, I feel the need to re-assert that SportsMeasures does NOT care about won-loss record. Also, the measures, rankings, seeds, etc., are not the opinion of any person. The measures are a direct result of the games so, these rankings, since they are scientifically derived, are more “factual” than any other method available. I publish the error of measurement as an admission that these measures are NOT an absolute declaration of a teams’ ability and any teams’ chance to win any game are the reason the games are played (see Northern Iowa vs. Kansas in the 2010 tournament). If these measures were absolute and the game results were driven by these measures, there would be no point of playing the games. These measures show which team is better given how they have played the games on their schedule – nothing more, nothing less.
OK, that said…how would you like to be Wisconsin or Kansas St? Kansas St should be a seven seed and the NCAA gave them a five. They are playing a twelve seed in Utah St, which should be a four! Wisconsin is seeded properly, but they’re playing Belmont, which the NCAA gave a thirteen seed and should be a five. I am very eager to watch these two games. I have already picked USU and Belmont to win their first round games. I have to believe that both teams will be rather irked at their seeds and will come out with collective chips on their shoulders.
Those two games are the two most grossly mis-seeded of the tourney. There are a few others that come to mind also. I already mentioned UCLA, the recipient of a gift seven seed, which shouldn’t even be in the tournament!
Another game of particular interest is the Xavier (pronounced ‘zavier’ not x-zavier) -Marquette match-up. Xavier is mis-seeded higher by three spots and Marquette is also out of place, three spots lower. So, while the NCAA has this game as a six-eleven game, in reality it is a nine-eight game. Xavier is really a nine and Marquette, an eight.
Missouri is mis-seeded by four spots. They are an eleven and should be a seven. Mizzou is playing Cincinnati, which is seeded correctly at six. In SportsMeasures’ rankings these two teams are only separated by five spots. This should be a very competitive game.
UNLV received a seed three spots lower than they should have and given how poorly (underachieved, given their vast talent pool) Illinois has played this year, UNLV should enjoy a 15-point victory.
Florida also received a late Christmas present with their two seed. They should be a five. Since they play a fifteen seed in the opening round it shouldn’t matter at all, especially since they are playing UC-Santa Barbara.
Butler must’ve gotten the ‘championship game appearance’ exception by getting a seed four spots higher than it should have. So, instead of playing a five seed in the opening round, they will play Old Dominion, a nine seed. This, too, should be a nice game.
Vanderbilt and Richmond combine for a seeding difference of five spots. Vandy should be an eight and Richmond should be a ten. Instead the NCAA made this matchup a five-twelve game. It will be much more competitive than many games like those usually are. However, if my memory is any good, twelve seeds have had excellent success against five seeds in years past. Perhaps, it is the NCAA’s method that is the root cause of this anomaly. A five seed is among the top 20 teams in the country and a twelve is among only the top 48 – a nice team, but hardly one that should have the success against a top 20 team at the rate the 12s have had over the years.
Michigan and Tennessee is an interesting game as well. Blue has been upgraded by the NCAA by three spots and Tennessee was given an unearned invitation – which would translate to approximately a 14-seed. So, overall, there is really only about three seed spots between them or 13 ranking positions according to SportsMeasures. This should still be a competitive game.
Washington and Georgia also have a combined difference of four spots. Instead of being a five-12 game (SportsMeasures), the NCAA made it a seven-ten game. Washington, playing very well of late, should win this game rather easily. Remember, Georgia should have been the last team in the tournament.
Those are all the games where the NCAA and SportsMeasures disagree the most. I encourage you to watch these games with particular interest. All other games and teams had discrepancies of only one or two seeds. That amount is surely within an allowable gray area of any measurement methodology.
Enjoy the games – I know I will!