Re: 2017 Pairwise Predictor
Edit: Never mind. I forgot to include the effect of a Minnesota loss to Wisconsin in the WCHA final on its record vs. common opponents comparison with UMD. This would make that criterion tied, and UMD wins the comparison with RPI is the tiebreaker. I'll leave what I wrote below so that people can pick apart other aspects of my thought process, but it probably doesn't matter.
Grant, is this intended as a Pairwise Predictor, or just an RPI calculator? I ask, because if it's the former, it's returning incorrect results for some scenarios of the UMN/UMD comparison. If Minnesota beats UMD in the WCHA semifinal, they would win the individual comparison with the Bulldogs.* UMD would have a slight edge in RPI, Minnesota would have a slight edge in Record vs. Common Opponents, and the Gophers pick up a point for the head-to-head results.
In some cases, this makes a big difference. Take the scenario where:
1) Someone other than BC wins Hockey East;
2) Clarkson and St Lawrence meet in the ECAC final;
3) Wisconsin beats Minnesota in the WCHA final;
4) The CHA does whatever it wants (I think I covered all of the possibilities.)
This leaves UMD and UMN at #4 and #5 respectively in your calculator, except that I'm pretty sure it should be the other way around and the Gophers would host.
This also leads to some complicated outcomes. Take this scenario:
1) BC wins Hockey East;
2) Clarkson and St Lawrence play in the ECAC final;
3) Wisconsin beats Minnesota in the WCHA final;
4) Robert Morris and Syracuse play in the CHA final (I'm tired of running through all of those possibilities.)
Your calculator has UMD, BC, and UMN at #4, #5, and #6, in that order. But, we'd have a situation in which UMD wins the comparison vs. BC, BC wins the comparison vs. UMN, and UMN beats UMD. I seem to remember that the NCAA uses some not entirely intuitive process to resolve this sort of thing, but I can't remember what it is.
Or take this scenario:
1) Northeastern beats BC in the Hockey East final;
2) Clarkson and Cornell play in the ECAC final;
3) Wisconsin beats Minnesota in the WCHA final;
4) Robert Morris and Syracuse play in the CHA final.
This causes the calculator to give us UMD at #3, St Lawrence at #4, and UMN at #5, but we again have a round robin of actual comparison wins: UMD over StL; StL over UMN; UMN over UMD. Here, UMD and StL have almost identical RPIs, and they have about a .003 edge on UMN, so I suspect that the committee would see them as getting the #3 and #4 spots, except that travel restrictions would mean a UMN/UMD matchup, and I seem to remember precedent that the team that wins the individual comparison would host any particular matchup, thus leading to that game being in Minneapolis, despite that nominally being the #5 team hosting the #3 team and making a complete hash of the logic that made StL the #4 seed.
*This, and everything that follows, assumes that the NCAA doesn't just decide that RPI is the only thing that matters.
