Bracket math (long but intriguing!)

I thought I’d do a comparison between Joe Lunardi’s (ESPN.com Bracketologist) and Jerry Palm’s (CBSsportsline.com) Selection Sunday morning brackets and the actual bracket. Initially I just wanted to find out how accurate they were, especially making predictions within hours of when the actual bracket was released, when most all of the data was finalized, but it went WAY BEYOND that once I got in to the numbers a little bit. First off, both bracketologists had 63 of the 65 teams in the field correct. The teams they missed on were the same: Arizona (who was given the final at-large bid, theoretically) and Mississippi State (who won an automatic bid); and the teams they had in their places were also the same: Creighton and St. Mary’s.
I also looked at how accurately the experts predicted how the teams were seeded. I only looked at seeds 1-12, since 13-16 only made the field because of automatic bids, where at least some 12s were at-large teams. I don’t expect the experts to put much time and effort into projecting differences between teams that are likely to lose in the first round anyway. Jerry Palm predicted teams with their correct seed 59% of the time. Joe Lunardi got it right 43% of the time. When comparing conference-by-conference, Palm predicted more accurately than, or at least as accurately as, Lunardi in every single conference. ACC: Palm 57%, Lunardi 57%. Big 10: Palm 57%, Lunardi 43%. Big 12: Palm 83%, Lunardi 50%. Big East: Palm 71%, Lunardi 57%. Mid Majors: Palm 54%, Lunardi 38%. Pac 10: Palm 60%, Lunardi 40%. SEC: Palm 50%, Lunardi 0%.
Both experts did the worst at projecting seeding for the Mid Majors and the SEC. They tended to overproject teams from those two groups, meaning the "experts" gave them more generous seeds than the committee. This wasn’t always the case, especially in the cases of Utah and Siena, where projections were under by as much as 3 bids, but in the aggregate they thought the Mid Majors/SEC teams deserved higher bids than what the committee awarded them. The median projection for Mid Majors put them one seed below what the experts said, where the average was only 1/4 of a seed (because Siena and Utah received such higher seeds than projected). Jerry Palm also overprojected the Big 10, where Lunardi only gave too much credit to the two aforementioned groups. The committee awarded more generous seeds to the ACC, Big East, Big 12, and Pac 10 than the experts gave. The difference was around a half a seed per team. In short, BCS conference teams received more favorable seeds than Mid Majors. I believe that the criteria used to judge Mid Majors is inconsistent/biased, and these experts can’t quite grasp exactly how it is judged behind closed doors, probably because they assume it isn't biased when it likely is.
Now the committee usually has several major factors to determine the quality of at-large teams, and how they are prioritized varies by committee member. [I believe it is also prioritized by how it benefits BCS teams.] The biggest is the RPI. Which at-large bids make sense based on RPI? Well, the 4 lowest RPI teams to get at-large bids: Wisconsin 48, Maryland 52, Boston College 57, and Arizona 59. Notice any common traits among those teams? Now, here are teams with better RPIs than ALL 4 of those teams that did not make the field: San Diego State 27, Creighton 41, Illinois State 45, UAB 46, and St. Mary’s 47. How about common traits among those schools? There are also several other similarly-traited schools with RPIs better than Arizona. So, RPI certainly cannot explain these at-large bids.
Usually the second most discussed by the committee (though not this year) is performance in the last 10 games. So let’s look at these 4 at-large teams again: Wisconsin, 7-3; Maryland, 5-5; Boston College, 5-5; and Arizona 5-5. Compare that to the 5 left out of the field: San Diego State, 6-4; Creighton, 9-1; Illinois State, 5-5; UAB, 7-3; and St. Mary’s 7-3. It becomes obvious why the committee chose to ignore this one this year: because it couldn’t justify anybody but Wisconsin.
Next is quality (top 50 RPI) wins: please note; this is a number, not a percentage. I won’t go into all the details of everybody but Arizona was 6-10 against the RPI top 50. St. Mary’s was 2-3. So, total number is 6-2, in favor of Arizona. Here is the first valid argument they have for Arizona. Of course, St. Mary’s actually won a higher percentage of their top 50 games, but that’s not as important, apparently. Really, I think Arizona proved they DIDN’T belong in their 16 tries. St. Mary’s didn’t prove it one way or the other in their 5 games. I would also like to point out that Wisconsin lost to EVERY NCAA tournament team on their schedule except for Michigan. They did win a few games against Big Ten teams that made the tournament (at home), but they also lost to each of those teams (on the road). They lost all of their non-conference games to NCAA Tournament teams, home, road, or neutral.
Sometimes they discuss how teams do as a result of injuries, and if a player was out but will be back for the tournament. That wasn’t discussed this year, because it favored St. Mary’s over the other schools. Other times we hear about conference or preseason tournament performance, because that is the most similar environment to the NCAA tournament. We didn’t hear about that this year either: other than Maryland which lost to eventual ACC champion Duke in the semifinals, the final at-large bids went to teams that exited early from their conference tournaments where SDSU, St. Mary’s, and Illinois State all fell in conference championship games: SDSU to a 5-seed in a game that came down to the final possession and St. Mary’s to a 4-seed when their best player was only in his second game back from injury. Another one is road and neutral site record: another bad one for these 4. Wisconsin was 6-8, Maryland was 5-9, Boston College was 7-7, and Arizona was 5-10 (with only ONE of those wins coming against a team in the top 150 in the RPI). Compare that to the records of SDSU (9-7), Creighton (10-4), Illinois State (10-7), UAB (8-10), and St. Mary’s (12-5). The committee ignored that one too, for obvious reason.
What we did hear about was strength of schedule (which to some degree you have no control over, especially as a middle-of-the-road mid-major who can’t get big teams to play you and don’t have tough conference schedules-see St. Mary's). We did hear about how games in November and December matter just as much as games in February and March. Why is that? Isn’t one of the great things about the NCAA tournament how it rewards improvement over the course of the season? Not when Arizona beat SDSU, Gonzaga, and Kansas in December, but hasn’t done anything in February and March. Maryland defeated Michigan State in November, and won at Michigan in December, and, besides beating UNC in a miraculous comeback at home, they haven’t done much since. St. Mary’s lost at UTEP in November. SDSU, Illinois State, and Creighton all had “embarrassing” losses in November/December. So this year (and probably only for this year) November and December has to be just as important as February and March in order to justify inclusion of certain teams from certain conferences.
In summary, the committee has several factors it looks at, so that it can find at least one factor that allows the BCS schools in. It’s as simple as that! This year they had to reach way back and pull out a new one, never been used before. It is wrong and unfair, and I am glad that Dick Vitale called them out on it. I am calling them out on it too. Selection Committee: explain yourself! I am not saying that San Diego State, Creighton, and St. Mary’s belonged in the tournament over Wisconsin, Maryland, and Arizona, but I’m not not saying it either. Certainly there are as many arguments for the former as there are for the latter. The fact that none of the former three made it and all of the latter three did is enough reason for me to scratch my head, do way too much research, put way too much time into it, and post a REALLY LONG blog article about it. Thanks for reading. I’ll probably post something about my sleeper picks late Wednesday night (I won’t be home until about 9pm Mountain Time), in case you want to make any last minute changes to your bracket (like you picked one of my sleeper teams to win and you decide you’d better not agree with me), check late Wednesday night or early Thursday morning. Most brackets must be submitted by noon Eastern on Thursday.