Homer wrote: ↑Mon May 06, 2019 11:20 amI know you're way deeper into the weeds on this stuff than I'll ever be, so maybe what I'm going to say is already obvious. But the one thing that immediately comes to mind that'd produce small differences like that is that IIRC you're supposed to drop an opponent's game(s) against the team you're rating from the calculation. In other words, if you're computing Hopkins' RPI you enter Towson as 10-4 not 11-4, UNC as 8-6 not 8-7, Maryland as 11-2 not 11-4, and so on. (And the same for opponents' opponents, so e.g. Georgetown as a Towson opponent goes in as 13-3 not 13-4.)laxreference wrote: ↑Mon May 06, 2019 11:02 am The formula is very simple. The 25% vs 50% vs 25% is well understood. It's the inputs for each team that Admin was referencing I think. For some reason, various outlets tend to have different results. I have no earthly idea why. I have been trying to figure it out since I started putting up my RPI stuff.
To be claer, the ordering of the teams in the final RPI tends to be the same, so it's not likely we are all putting out different lists, but you'd think the specific underlying inputs would be the same as well, which doesn't seem to be the case.
And no, Hampton does not count.
If some sites are doing that correctly* and others aren't, that'd account for minor discrepancies in input of the kind you're describing.
* It's also possible I have it backward and you AREN'T supposed to do that but some people do. But I'm pretty sure the way I have it here is the right one.
Years ago that was a very common error.