[CTcard]

So it seems that RPI not only is early-season focused, but it tries to conclude too much about a team from too little information.    But maybe this already has been said so often that I'm revealing how late I am to this game.

Lots of problems with RPI, but I would put at the top of the list that it is at the same time both too heavily weighted towards strength of schedule (75% of the RPI) and has far too simplistic a method of determining SOS (only has second-order considerations, i.e. it counts opponents' record [50%] and opponents' opponents' record [25%]). 

Somewhere in the Boyds World archives is a substantial mathematical description of how the RPI's algorithm drive's a program's value toward the center if they play in a region with fewer teams (basically, by playing a high fraction of possible opponents, the 75% SOS component gets driven towards 0.500). Obviously that tends to hurt the RPI of better western teams while helping the RPI of bad western teams. Problem is, it's the better teams which are in contention for the NCAA tournament, and nobody cares so much which teams are in the 250-300 range.   

I am not quite sure if Boyd's ISR system has a bias that somehow works for the western teams, those teams definitely do better in that system. It's certainly not from Boyd's personal biases, he's a Mississippi State guy. 
http://www.boydsworld.com/baseball/isr/currentisr.html

[CTcard]

So it seems that RPI not only is early-season focused, but it tries to conclude too much about a team from too little information.    But maybe this already has been said so often that I'm revealing how late I am to this game.

Lots of problems with RPI, but I would put at the top of the list that it is at the same time both too heavily weighted towards strength of schedule (75% of the RPI) and has far too simplistic a method of determining SOS (only has second-order considerations, i.e. it counts opponents' record [50%] and opponents' opponents' record [25%]). 

Somewhere in the Boyds World archives is a substantial mathematical description of how the RPI's algorithm drive's a program's value toward the center if they play in a region with fewer teams (basically, by playing a high fraction of possible opponents, the 75% SOS component gets driven towards 0.500). Obviously that tends to hurt the RPI of better western teams while helping the RPI of bad western teams. Problem is, it's the better teams which are in contention for the NCAA tournament, and nobody cares so much which teams are in the 250-300 range.   

I am not quite sure if Boyd's ISR system has a bias that somehow works for the western teams, those teams definitely do better in that system. It's certainly not from Boyd's personal biases, he's a Mississippi State guy. 
http://www.boydsworld.com/baseball/isr/currentisr.html