Congratulations on taking over as the head of the Selection Committee this year. I applaud your recent decision to make the official RPI available to the public. I agree that releasing the results has a number of positive benefits for both programs and fans, greatly increasing the transparency and general interest in college basketball as a whole.
However, I cannot see the logic in continuing to use the RPI as a tool in the selection process. I believe that the RPI does a poor job with what it was designed to do -- measure a team's SOS and its success against that SOS. Additionally, last year's home-and-road adjustments have made the ratings even more misleading. I know that the RPI is just one tool that the Committee uses, but the RPI also filters down into a number of other tools that the Committee has traditionally looked at (Top 50 W-L, Top 100 W-L, etc). And even in the very limited context in which the RPI is placed, any reliance on inaccurate information will ultimately lead to bad decision making, despite the Committee's best intention.
As a basketball fan, I simply do not want that to happen. I believe that the best and most deserving teams should get in. Basketball players and coaches should know upfront that they if the perform over the course of the season, that they'll earn the opportunity to make the tournament. I sincerely hope that this is the NCAA's and Selection Committee's intention as well. Maybe the Committee can see through the flaws in the RPI enough to get it right in the end. However, I believe that relying on inaccurate information is making an already tough job even tougher. And that's unfortunate when more accurate information may already be available.
I hope that as the new Chair of the Selection Committee you'll have to opportunity to revisit every step of the selection process in order to make the game and the tournament even better. I'd be happy to discuss any ideas with you in more detail.
The RPI doesn't get a team in nor does it eliminate a team from the tournament. The Comm spends a great deal of time watching teams play which is a far beter way to evaluate teams. Thanks for your interest. CL
(Reaction: It's clear he didn't even read my e-mail. I'm sure he gets 100 'The RPI is screwing my team' e-mails per week and this is his gut response.)
I clearly understand the RPI and how it is and isn't utilized by the Committee. All I'm saying is if you are trying to measure one team's resume of wins and losses, or 'body or work', versus another's, why not use a rating tool that does exactly that? Although it was designed to help in that regard, the RPI falls well short of that objective to the point it hinders instead of help your decision making process. A more accurate rating tool would make your job a lot easier.
Hello Jim--Each Comm member decides how to use the tools that are available. I think the composite of all resources allows each member to make good decisions. I would not say that my method is any better or more accurate than one that is more influenced by the quantitative tools. If there is such a tool that is better than a composite of all of the resources that we have, I haven't seen it yet. Thanks again for your thoughts.
Reaction: Still skeptical but at least leaving the door open.
CL Smooth (Okay...I made that part up),
Obviously no computer system can capture all the nuances that occur over the course of the basketball season, but a ranking system should be able to make sense of all of the win-loss data. I've developed a rating tool that is a true 'body-of-work' index (call it the JCI). In other words, using all of the win-loss information over the entire season, it evaluates whether or not Team A's resume of wins and losses is better than Team B's. Even if it's just one resource, it would help the Committee get to their answer faster and more effectively, with a higher degree of confidence compared to the RPI. Same inputs as the RPI, more intuitive, just as easy to produce, and much better output.
Below is the current top 60. You won't find many, if any, examples where if you went through your careful analysis you'd find your final answer closer to the RPI than the JCI. In fact, for a quick-and-dirty check, compare Group A vs. Group B and Group C vs. Group D below I and think you'll reach the same conclusion. The other nice benefit is the SOS rating captures the home-road effect accurately whereas the RPI SOS ignores it (by the way, the home-road adjustments made to the RPI last year only make it less accurate than it was before).
I'm not trying to push my rating system to you. I'm just trying to show you that quantitative resources that are far superior to the RPI are out there. As the Chairman of the Selection Committee, I'm sure you are committed and probably passionate about making the tournament the best that it can be. I assume that part of that committment would include making sure that each one of the resources that is at the selection committee's disposal is the best that it possibly can be. Given this level of committment, I'm surprised that the Committee continues to use an antiquated system such as the RPI.
Thanks for the time. Jim
(Included Current Top 60)
Group A: Top 25 RPI, Non Top 25 JCI
26. UCLA 17-4, RPI: 14 JCI SOS: 88 RPI SOS: 14
31. Southern Illinois 16-4, RPI: 23 JCI SOS: 118 RPI SOS: 80
36. Creighton 14-5, RPI: 25 JCI SOS: 64 RPI SOS: 46
40. Wis-Milwaukee 14-4, RPI: 20 JCI SOS: 121 RPI SOS: 81
46. Arizona 13-7, RPI: 16 JCI SOS: 46 RPI SOS: 3
Group B: Top 25 JCI, Non Top 25 RPI
7. George Washington 16-1, RPI: 43 JCI SOS: 182 RPI SOS: 240
12. Georgetown 14-4, RPI: 29 JCI SOS: 19 RPI SOS: 74
13. North Carolina St 16-4, RPI: 30 JCI SOS: 25 RPI SOS: 67
19. West Virginia 14-4, RPI: 33 JCI SOS: 23 RPI SOS: 54
22. Bucknell 15-3, RPI: 34 JCI SOS: 122 RPI SOS: 133
Group C: Top 50 RPI, Non Top 50 JCI
52. Air Force 15-3, RPI: 49 JCI SOS: 276 RPI SOS: 186
60. George Mason 15-5, RPI: 36 JCI SOS: 185 RPI SOS: 89
61. Saint Joseph's 9-8, RPI: 42 JCI SOS: 12 RPI SOS: 11
62. Old Dominion 15-6, RPI: 39 JCI SOS: 165 RPI SOS: 66
63. UNC Wilmington 16-6, RPI: 44 JCI SOS: 153 RPI SOS: 100
Group D: Top 50 JCI, Non Top 50 RPI
39. Louisville 14-6, RPI: 67 JCI SOS: 51 RPI SOS: 59
42. Washington 16-4, RPI: 56 JCI SOS: 180 RPI SOS: 102
44. Arkansas 14-6, RPI: 58 JCI SOS: 69 RPI SOS: 70
49. Miami-Florida 12-8, RPI: 71 JCI SOS: 17 RPI SOS: 41
50. Florida St 12-5, RPI: 95 JCI SOS: 78 RPI SOS: 171
Note: According to the JCI numbers, Group B would be expected to win 76% of the games in a round robin event against Group A. Groud D would be expected to win 59% of the games against Group B. So tell how the NCAA can justify using a RPI that has the higher quality teams 21 slots worse on average than the lower quality team.
CRAIG: Jimmy to da C (okay...maybe not),
Thanks for your analysis. I'll look over the information. CL
Reaction: Who knows if he's really going to look at it or not. He was probably just telling me this so I would stop harassing him. However, the numbers speak for themselves, so hopefully he will reach the right conclusion sooner rather than later.