A look at conference tournaments using the RPI to seed

The SEC made it official on Thursday that they would no longer have two divisions. One of the side effects of that decision is that the SEC Tournament will now be seeded differently. In the past, the SEC Tournament seeding was difficult to figure out. The top two teams from each division got a bye, with the third/fourth place finisher playing the sixth/fifth place finisher from the other division.

It was complicated, annoying to try and figure out, and, frankly, a bit unfair for the teams in the more powerful East division.

Apparently, Billy Donovan wants the SEC to continue to be the unique conference tournament, as he proposed seeding the teams based on their RPI. The RPI is an extremely flawed metric, but there is no need to rehash those issues here; we've been over it too many times.

Perhaps the biggest issue, however, is that the RPI is a constantly moving number. One of the factors included in the RPI is a team's opponents' opponents winning percentage. The outcome of a game that a given team doesn't participate in -- that one of their opponents doesn't participate in -- can affect a team's conference tournament seeding. When is the cutoff?

That is a minor issue, however. The bigger issue is that the RPI doesn't take into effect conference record or head-to-head matchups. In other words, the potential is there for a team with a better league record and a sweep of the two head-to-head games to get stuck behind that opponent in the tournament seeding.

If you scroll down in this post, you'll see the spreadsheets for what each of the BCS conferences would have looked like this past season if their tournaments were seeded by RPI. The first column is what each team was actually seeded in their respective conference tournaments, and the numbers in the last column are the team's league record. The RPI is based off of the values on Sunday, March 6th, as pulled from StatSheet and CollegeRPI.com.

Its not hard to see some of the issues. For starters, ACC champion North Carolina would have been seeded behind second place finisher Duke. Georgetown, who finished eighth in the Big East at 10-8, would have received the three seed while Louisville, who earned the three seed in a three-way tiebreaker with St. John's and Syracuse, would have been the seven seed.

Those are just the most glaring issues, but they are far from the only problems with using the RPI. Kansas State would be seeded above Texas A&M despite losing to the Aggies in their only head-to-head battle. Michigan was in a four-way tie with Michigan State, Illinois, and Penn State at 9-9 in the Big Ten. They went 4-1 against those three opponents, earning the four seed in the Big Ten tournament. But based on RPI, Michigan would have been seventh, or fourth out of those four.

Ironically enough, the only league where using the RPI makes sense is in the SEC, where the top five seeds would have been from the East division. But that was before the changes to the league's schedule.

The bottom line is that the RPI factors in the entire season, including non-conference play. Determining the league champion and the conference tournament seeding is based strictly on performance against league opponents.

ACC:



Big 12:



Big East:



Big Ten:



Pac-10:



SEC: