RPI and the effect of smart scheduling

The first official RPI was released yesterday. To be frank, I hate the RPI. The problem is that it’s not anywhere close to the best predictor of how good a team is. If you don’t believe me, then just start betting the money line in Vegas based on RPI scores. When you’re pawning your electronics to pay your bill, give me a call.

There are much better sources – Pomeroy and Sagarin are two good examples. Personally, I prefer Pomeroy. The Vegas line almost always closely mirrors the projected line by Pomeroy. That’s strong work.

The other reason I hate the RPI is that it’s so easy to manipulate. The simple version is this: play a tough schedule. When Virginia plays a team like Morgan State it hurts their RPI no matter how well they perform. The Hoos easily won that game 75-57. VMI beat Morgan State by one in overtime. According to the RPI, those two outcomes are identical.

Unfortunately, the powers that be in the ACC have not figured this out. Or if they have, they don’t care. Some schedules have been weak. And come Selection Sunday the conference will rediscover the price of that.

With the release of the RPI I compared the initial rating with the current ratings at Pomeroy. Ideally, a team which is doing a good job of manipulating the RPI will be better according to the RPI than they are according to reality (using Pomeroy as the best indicator).

Here’s how that table looks for the ACC. And it’s not a pretty site for Virginia or Clemson.

Team RPI KP Plus/Minus
Duke 1 1 0
Miami  8 23 15
NC State 15 35 20
UNC 42 40 -2
Maryland 67 48 -19
FSU 72 70 -2
Georgia Tech 123 72 -51
Virginia 133 28 -105
Boston College 140 146 6
Virginia Tech 144 147 3
Wake Forest 158 156 -2
Clemson 204 92 -112