WTF is the RPI Bracket?
The RPI Bracket is, quite simply, a bracket for the NCAA Men's Basketball Tournament that is designed to take a lot of the "guess work" out of the process, placing an increased emphasis on computer analysis and statistical formulas, in particular our proprietary metric, RPIBracket.com's Tournament Index, or T.I.
Why do we need it?
It is our opinion at RPIBracket.com that much of the bracketology discussions that start to heat up around this time of year are a bunch of hooey. In fact, we are absolutely sure of it. While it is very interesting to debate which teams should be in and which teams should not, we have developed a statistically relevant formula, based on several statitically relevant data points, that will determine the best teams in college basketball at any given time, based on performance, not based on conference affiliation, revenue generating potential or juicy matchups.
We orginally started looking at the RPI as our key metric, but even though the RPI is perfect at what it does, every single time, it is not perfect on its own. So, we have blended the RPI with what we feel are the other most statistically relevant pieces of information to come up with our Tournament Index, which is explained below. What the T.I. does is provide an objective quantification of a team's worthiness to take part in the tournament compared to every other team in Division I. Its complexity ensures that the numbers are an accurate representation of what teams should be in at any given time, and its simplicity allow us to make a very easy argument to a team that does not make the tournament: "Sorry, but you came up just short in the T.I. Perhaps you should have demonstrated your worthiness to take part in the tournament by beating more of the teams you played, or alternatively, losing to less of them. That loss to the team with an RPI of 187 didn't help, either."
How does it work?
In the RPI Bracket, there is basically only one rule -- T.I. is king, and we don't mean the rapper T.I., although if you can't get your head to bobbing whenever you hear "Ride Wit Me" then you should be checked for a pulse. Whatever the Tournament Index says is the law...if it says that a perennial powerhouse is the last team out, and a second bid for a traditional one-bid conference is the last team in, then so be it...if a perennial power conference only ends up with 3 teams in the bracket, then so be it...the NCAA will just have to miss out on the revenue that the powerhouse teams and conferences can provide. Our algorithm that spits out the T.I. incorporates the following data into its analysis.
- Team RPI. This is the standard RPI that everyone else uses as a starting point for the discussion about every team.
- Conference RPI. Some might see this as "power conference bias", but that is too bad. There is a reason why those conferences are good. We have made every effort to not turn this part of our formula into the crazy, money-based antics of the selection committee, but we cannot discount entirely the fact that the teams in the best conferences consistently play the best teams, and should be rewarded for doing so.
- Strength of Schedule. We want to reward teams for playing in early season tournaments and penalize teams for playing cupcakes. In the RPI, a win is a win is a win, which is not always the best way to reward and penalize teams.
- Top 50 and Beyond. Wins against the Top 50 count for a positive credit. Losses against teams in the top 51-100 count the same amount against a team. Losses to teams in the Top 101+ count 1.5 times as much against a team. We do not care about wins against the Top 51+ because tournament teams should win those games...and we don't count losses against the Top 50, since tournament teams playing each other should play close games.
- Recent Performance. Our algorithm rewards teams that are rolling along and hitting their stride in February and March, and penalizes teams that loaded up on cheap wins in November and December, but are struggling in the final weeks before the tournament.
Once we run the data for every team through the algorithm, our bracket builder can start to do its job. Here are the basic steps it takes to make this happen:
- Initial Sort. Sort all of the teams according to the Tournament Index (this is the final number that our algorithm spits out).
- Automatic Bids. Put all 31 automatic bids on the list of tournament teams.
- At-Large Bids. Start to move down the Tournament Index list. If you get to a team that is not on the list of tournament teams, add them in there.
- Final Sort. Once you get to 65 teams, make sure they are sorted according to the Tournament Index.
- Seeding Groups. Start to move down the list of tournament teams and make 15 groups of 4 teams and 1 group of 5 teams. These are the seeding groups. Teams 1-4 are the #1 seeds, teams 5-8 are the #2 seeds, etc.
- Regional Brackets. Here is the procedure for making the regional brackets. It is the same method used by the selection committee, and it is commonly referred to as the "S Pattern".
- The #1 overall seed will be in a bracket with the winner of the play-in game, or the worst #16 seed (#64 or #65 overall), the best #15 seed (#57 overall), the worst #14 seed (#56 overall), the best #13 seed (#49 overall), etc.
- The #2 overall seed will be in a bracket with the second worst #16 seed (#63 overall), the second best #15 seed (#58 overall), the second worst #14 seed (#55 overall), the second best #13 seed (#50 overall), etc.
- Assign Regions. Once each 16-team regional bracket is formed, you assign it to a particular region. First, put the #1 overall seed in the region where the regional final is closest to its campus. Next, assign the #2 overall seed to the region of the remaining 3 regions where the regional final is closest to its campus. Do the same for the #3 overall seed and then assign the #4 overall seed to the last remaining region. We DO NOT care about location for the first 2 rounds...if that means that North Carolina does not get to play in the "Tar Heel Invitational", a.k.a. the Greensboro Subregion, then so be it.
- Clean-Up. The final step is to make sure that we are meeting all the "rules". I won't bore you with the details, but here are basics:
- Teams from the same conference must be evenly distributed among the 4 brackets. For example, if you have a conference with 7 teams in tournament, they must have 2 teams in 3 of the regions and 1 team in the 4th region.
- Teams from the same conference cannot meet until the Sweet Sixteen. That is, teams from the same conference cannot be in the same subregion, or "pod."
- Teams seeded 1-5 in each region should be protected from having to play a "road game", or any opponent playing in close proximity to their campus, within the first round.
- No team can be seeded in the region it is hosting. For the 2010 bracket, the hosting teams are Utah (West Region), Syracuse (East Region), MVC (Midwest Region) and Houston/Rice (South Region).