A few notes as we quickly approach 6pm.
- I called an audible today and made a change that I think will improve the model's performance in both at-large selection and seeding. I've written a few times that the model frequently has problems overestimating mid-major teams with good profiles. Usually the "eye test" shows that these teams have built up their profiles by beating up teams near the middle and bottom of the pack. To account for this, today I added the number of games against teams ranked 101+ in RPI (decent but not very good teams) and games against teams ranked 201+ in RPI (not good teams). I like what I see, so I'm going to cross my fingers and leave it in. This mostly affects teams like Utah State (for seeding) and Creighton and St. Mary's.
- There are five teams fighting for the #1 seeds. North Carolina, Pitt, Louisville, UConn, and Memphis all have 15.81 or more seed points, meaning the seed model considers each of their resumes #1 seed quality. Of course, there are only four slots, so Memphis is (barely) the odd one out. The Tigers are 0.04 seed points behind UConn, so I wouldn't be surprised to see either Memphis or UConn heading to Phoenix as the #1 seed.
There are several teams in the mix for the last few at-large spots.
Arizona - 6 RPI Top 50 wins, but #156 Oregon State was their best road win. They beat #27 Gonzaga on a "neutral" court in Phoenix.
San Diego State - Above .500 road record, but only 2 RPI Top 50 wins and their best road wins are two over #66 UNLV.
Creighton - The Bluejays also had only 2 RPI Top 50 wins, but did win 8 of 12 on the road, including three in the Top 100.
Maryland - Three RPI Top 25 wins (including UNC and Michigan State), but only 2-6 in true road games with the best win at #111 NC State. They also lost at home to #130 Morgan State. The choice between the Terps and teams like Creighton and San Diego State comes down to signature wins against good but not great body of work.
Florida - Went 16-1 in Gainesville and beat Pac-10 regular season champ RPI #14 Washington in Kansas City, but went 2-7 in true road games, with the best win at #95 Vanderbilt. Unfortunately, Washington was their only win over a Top 50 team (the average RPI of teams they defeated was #175). The Gators are probably on the wrong side of the bubble.
Penn State - Despite some very good road wins in the Big 10, a weak non-conference will probably haunt the Nittany Lions.
New Mexico - They'd have a much stronger case if they had gone further in the MWC tourney. Only one RPI Top 25 and three RPI Top 50 wins will hurt, along with the home loss to #165 Central Florida.
We'll learn a lot if the committee goes with the BCS schools over St. Mary's, SDSU, and Creighton.
- Big Ten final - The model currently has Purdue as the last #4 seed and Ohio State as the second #6. With the game concluding only 30 minutes before the bracket is unveiled, my guess is that the committee may have slots for the winner and loser of this game. If Purdue wins, we should be fine but if Ohio State wins we may miss the seeds on both. At least it's not like last year with the Big XII final between Texas and Kansas potentially determining a #1 seed.
- As I did last year, I'm going to take a crack at placing the top 4 seed lines in pods and regions.
- I may take one last look at the coaches poll before 6:00, but I don't want to overthink it.