I don't follow analytics all that closely, but I've got ESPN+ so I occasionally check out Bill Connelly's SP+ rankings. This week is an example of where analytics can really get its head up its own ass sometimes.
It's always prefaced with language about how it's an efficiency ranking, forward-facing, not predictive, etc. It's meant to eliminate fluky bounces and rewards teams that line up and beat (or nearly beat) their opponents straight up. Fair enough.
So given that, I don't know how you explain the last couple of weeks.
Five days ago:
#41. Iowa (1-0)
#42. Iowa State (1-0)
This morning:
#39. Iowa (1-1)
#43. Iowa State (2-0)
Yesterday was full of fluky bounces, but they should more or less cancel out. Perhaps Iowa's blocked punts carry a lot of weight. But can there be any argument that ISU didn't beat Iowa straight up, and that it was not a fluky win?
Seems like when your ranking moves up a team that lost at home, and moves down the team that beat them - not weeks later after other results are factored in, but just hours after the game was played - you have a pretty iffy ranking system.