[M T]

One of the things that bugs me about the RPI rankings is the possibility that small differences get into the ratings and for them to cause  feedback that magnifies those differences.  Specifically, my concern is if team A is ranked slightly higher than team B (of a different conference) going into conference play (that is, a small set of teams playing each other), then all the conference opponents of A get that marginal benefit by playing against A (either winning or losing), and then they further spread that benefit to other teams in the conference themselves.  So, basically, conference A is lifted by its bootstraps relative to conference B.

Let me emphasize that I don't know the RPI math enough to confirm that is the case.  I'd like to see a sensitivity analysis so I can understand better how these things work.  I do know that it seems strange to see a team's RPI go down as a result of winning a game.

Here's an interesting situation this week.  At http://www.rpiratings.com/womensrpi.php, Stanford is rated 3rd in RPI, ahead of ND & Tenn.  At http://www.rpiratings.com/womrate.php, Stanford is rated 4th, well behind ND, and barely ahead of Duke & Tenn.  When you look at the difference in the SoS in those different ratings, you can see how the little differences get magnified.  Is Ohio State's schedule 61st or 108th?  Should Kansas State be ranked 10th or 23rd?

Yes, the teams that make it to the tourney still have to win their way forward, but the difficulty of the road is set based partially on these RPI/rankings.    I'm really hoping that Stanford doesn't get seeded 4th or 5th overall.

[M T]

One of the things that bugs me about the RPI rankings is the possibility that small differences get into the ratings and for them to cause  feedback that magnifies those differences.  Specifically, my concern is if team A is ranked slightly higher than team B (of a different conference) going into conference play (that is, a small set of teams playing each other), then all the conference opponents of A get that marginal benefit by playing against A (either winning or losing), and then they further spread that benefit to other teams in the conference themselves.  So, basically, conference A is lifted by its bootstraps relative to conference B.

Let me emphasize that I don't know the RPI math enough to confirm that is the case.  I'd like to see a sensitivity analysis so I can understand better how these things work.  I do know that it seems strange to see a team's RPI go down as a result of winning a game.

Here's an interesting situation this week.  At http://www.rpiratings.com/womensrpi.php, Stanford is rated 3rd in RPI, ahead of ND & Tenn.  At http://www.rpiratings.com/womrate.php, Stanford is rated 4th, well behind ND, and barely ahead of Duke & Tenn.  When you look at the difference in the SoS in those different ratings, you can see how the little differences get magnified.  Is Ohio State's schedule 61st or 108th?  Should Kansas State be ranked 10th or 23rd?

Yes, the teams that make it to the tourney still have to win their way forward, but the difficulty of the road is set based partially on these RPI/rankings.    I'm really hoping that Stanford doesn't get seeded 4th or 5th overall.