 |
|
| A10 BY THE NUMBERS Recent Articles 02/11/10: 2/11 A10 Postseason Odds Report 01/09/10: 1/9 A10 Postseason Odds Report 01/05/09: The Effect of the Unbalanced Schedule on the 2010 A10 Conference Season 12/30/09: A10 Postseason Odds Report 12/23/09: The Effect of Melquan Bolding's Absence on Duquesne 12/22/09: A10 Postseason Odds Report 12/18/09: A10 Postseason Odds Report 12/11/09: Quantifying the Rebounding Problems of St. Joe's Basketball Links NIT-ology Ken Pomeroy Mid-Majority Yoco Hoops Bracket Board College Hoops Net A10 Forum Storming the Floor College Chalktalk Villanova by the Numbers Valley Ledger March Madness All Season Contact mikebscullin@gmail.com | A10 Postseason Odds Report December 18, 2009 by Mike Scullin As a fan of multiple sports, I've always very much enjoyed attempts to project the probability of postseason berths, such as those of Baseball Prospectus and Football Outsiders. It's relatively easy to develop a system for calculating the odds for these sports -- playoff spots are determined wins and losses, so by using relatively reliable win probability metrics, these sites can provide a very accurate picture of things. Postseason forecasting for college basketball is a whole different animal -- in addition to nailing down a team's likely future performance, one must also nail down how a very human committee is going to look at things. For quite some time now, numerous bracketologists have done a good job of predicting who most deserves to go to each tournament "if the season ended today," or "if the season ends how we think it's going to end." My goal is to develop a reliable way estimate the probability of postseason berths, rather than just finding the most worthy 65 NCAA teams and 32 NIT teams and calling them "in". What follows is my first attempt at accomplishing this. Explanation is below the odds. TR = target record for NCAA or NIT at large bid TRPI = target RPI rank for NCAA or NIT at large bid NCAA%/NIT%/PstSn% = percent chance of reaching the NCAA tournament, NIT tournament, and one or the other, respectively, as an at large bid. PROJECTION THROUGH GAMES OF DECEMBER 17, 2009
Expected A10 NCAA bids: 3.12 / Expected A10 NIT bids: 2.73 NCAA BID BREAKDOWN: 1% chance of 1 bid 18% chance of 2 bids 54% chance of 3 bids 22% chance of 4 bids 5% chance of 5 bids Methodology I came up with target records/rpi's that a10 teams will need to get into the NCAA tournament or NIT, respectively, in my opinion. I based my estimates of target finishes on factors such as quality wins, strength of schedule, upcoming opportunities for quality wins, etc. Teams with things in their favor such as preseason rep and abnormally high quality wins in their back pockets will probably have more record/rpi leeway, while the opposite will be true for teams without these advantages. The target records/rpi's will certainly change from week to week based on things these teams and teams they may be competing against for bids happen to do. I then computed the chances of them reaching those target records based on win probability estimates - for now, I'm using sagarin's predictor (i like kenpom's ratings better, but this is simpler to use due to rpiforecast being integrated with it). Rpiforecast does a good job of breaking down the probability of a team finishing with each possible record and RPI rating -- you can find these projections on each of the team pages here. In order to be on the conservative side, I assumed that getting exactly to their target finish gives them only a 75% chance of being selected for the NCAA or NIT. Surpassing it means they are assumed to be in. Not an exact science by any means, but I'm confident I can provide a reasonable enough approximation of reality for this to be a worthwhile exercise. Note that my opinion only factors into the determination of target RPIs/records. The likelihood of reaching those targets is based only on the hard numbers to date. |