I know that the various folks with a proprietary ranking system each have their "secret sauce", I understand that. However, some of them are so different as to cause me to wonder what is happening. For example, after last night's game against Illinois here are three different rankings:
ESPN RPI: OSU 10-5 record with a rank = 62 and an SOS of 18
ESPN BPI: OSU 10-5 record with a rank = 44 and an SOS of 21
Sagarin: OSU 10-5 record with a rank = 56 and an SOS of 194
The variation in the rankings are close enough to accept since the algorithms used have differences. However, the SOS differences suspend believe. Two are close to one another but the other is 10 times higher. I have no clue as to the real SOS as it is clearly algorithm dependent.
Can anyone here with a solid understanding of mathematical modelling explain how such a large variation in SOS can still yield similar rankings?
Thanks.