Saturday, February 25, 2012

Bracketology: Analyzing the RPI

As we tick down the days until Selection Sunday, every ESPN broadcast will be inundated with Joe Lunardi's "Last 4 In" and "First 4 Out." While ESPN broadcasters take Joe's predictions as bible fact, Joe actually is pretty average among the 60+ bracketologists tracked by The Bracket Project. Last year, Joe's final bracket had a Paymon score of 310, his highest score in the 6 years the Bracket Project has existed. (For the record, "Paymon Score" was developed to "score" bracket predictions. You get 3 points for every correct team in the field, 2 points for every team correctly seeded, and 1 point for a team seeded within 1 of the seed where you predicted them.) In their "All Time" standings, Joe ranks 27th out of 44 bracketologists. And he gets paid a boatload of money by ESPN? Yikes.

As those of you know who have been reading my blog for a while, you'll know that I consider myself an amateur bracketologist. My Paymon Score last year was 312. Marginally better than "Joey Brackets"? Damn right.

I got 65 of 68 teams correct, 32 correct seeds, and 53 within 1 seed. Not bad for an amateur. 

This year, in the interest of greater transparency, the NCAA has been releasing the actual NCAA RPI and Team Sheets that the committee uses when selecting the field of 68.

Even though it isn't the greatest tool of analysis, the Committee still heavily relies on the RPI when ranking teams. I compiled some data over the last 4 tournaments on how RPI relates to tournament bids. Every year, there is a mid-major school that has a high RPI rating, a low strength of schedule, and winds up playing in the NIT. Using the RPI as a comparison metric, here are some interesting findings. Remember, this data is just for the last 4 tournaments.

Every school with a Top 30 RPI has made the field. Of the 120 teams with Top 30 RPIs, 40 have received automatic bids as conference champions and 80 have received At-Large Bids. Surprisingly, there have been the same number of Major and Mid-Major conference champions with Top 30 RPIs (20 each). However, among the At-Large group, there have been 64 teams from Major conferences and 16 teams from Mid-Major conferences. Regardless of conference size, all 80 of these teams have gone dancing.

Now remember, this is not saying that the Top 30 in the RPI will be the teams seeded on the higher lines in the bracket. This is just saying that they made the tournament. After the 30 line is where you start to see the data diverge between Major conference schools and Mid-Major schools. Of the 19 Major Conference schools ranked between 31 and 40, all 19 made the tournament (17 of which were At-Large bids). Of the 21 Mid-Major schools, only 12 made the dance, only 6 of which were At-Large bids. If you take the 6 Mid-Major conference champions out of the equation, only 40% (6 of 15) At-Large eligible Mid-Major teams with RPIs between 31 and 40 actually got tournament bids.

The 2009 Dayton Flyers were the highest ranked team to not get an NCAA Tournament bid over the last 4 years (RPI 32). They went on to beat North Carolina in the NIT Championship.

If we consider the Top 30 RPI line to be a "loose" delineation of where the bubble begins, especially for mid-major squads, there are some teams that should be doing some sweating over the last week of the regular season and into conference tournament week.

Colorado State, Creighton, New Mexico, Harvard, St Mary's, San Diego State, Murray State, Middle Tennessee State, and Long Beach State have RPIs that have been hovering in the 30s this year. As of right now, Colorado St and Creighton are under the 30 line but are by no means "safe." One bad loss and they're back on the wrong side of the line.

The delineation grows even wider between Majors and Mid-Majors as the RPI numbers go up. For teams with RPIs between 41 and 50, only 2 major conference teams were not given At-Large bids. 90% (19 of 21) Major conference teams with RPIs in the 40s went dancing. Conversely, the data was not nearly as promising for Mid-Major schools. Only 3 Mid-Major schools with an RPI in the 40s got bids while 10 were left out in the cold. Only 23% of At-Large eligible Mid-Major schools got tourny bids, while 90% of their Major conference counterparts were invited to the Big Dance.

Obviously, this is further proof that the RPI is not the be-all-end-all of the decisions made by the Tournament Committee. Factors like strength of schedule, quality of victories, and other positives and negatives in a team's body of work play into the decision. But since all of the team sheets the Committee uses work within the framework of the RPI, it is a good baseline data set to use.

There is a distinct divide in the data between teams with RPIs in the 40s and RPIs in the 50s. No mid-major school with an RPI over 50 has made the tournament in the last 4 years. For major conference schools, the probability decreases dramatically. For teams with an RPI between 51 and 60, 11 of 23 (48%) got At-Large bids. Only 2 of 18 teams with RPIs between 61 and 70 got At-Large bids, and only 1 team with an RPI over 70 (USC last year) has gone dancing.

If we take the mark of a Top 50 RPI as a delineator of where the "bubble" starts for Major conference teams, there are a number of schools that have some work to do to improve their resumes: Northwestern, Miami, Oregon, South Florida, Washington, Purdue, Texas, Mississippi State, Ole Miss, LSU, Arizona.

As we wind down the last week of the season, I'll be posting my first bracket prediction at the end of the regular season, along with Conference Tournament previews for all 30 Conference Tournaments. I've been doing something of a nightly "Bubble Watch" on Twitter as well for those that need that bracket fix. If you love college basketball as much as I do, this is one of the best times of the year. Get ready for an exciting two weeks leading up to Selection Sunday!

No comments: