There has been massive confusion over the National Collegiate women’s hockey selection process, leading to several conspiracy theories:
— the NCAA conspired to prevent an all WCHA-final and put all WCHA teams on one side of the bracket;
— the commitee chose to send Mercyhurst to struggling BU to increase Frozen Four host Mercyhurst’s chance of advancing; and
— the NCAA is too cheap to fly more than two teams for the NCAA quarterfinals.
Despite all the theories, the selections and brackets can be rationalized.
Q: Why did the committee decide to send UMD to Wisconsin and Dartmouth to Cornell? Why did the committee fail to protect the No. 1 seed?
First, recognize that the committee’s preferred rankings were (1) Wisconsin, (2) Cornell, (3) BU, (4) BC, (5) Minnesota, (6) Mercyhurst, (7) UMD, and (8) Dartmouth (more on these rankings later). Such a bracket would require four cross-country flights. The NCAA has not been willing to break the bank for so many flights since the inaugural tournament of 2005. The NCAA has repeatedly proven itself willing to fly three teams, however.
The top four seeds are locked in. The only ways to prevent flights were to (1) send Dartmouth to Cornell, BU, or BC or (2) send UMD to Wisconsin. Sending Dartmouth to Wisconsin would prevent either possibility. The committee’s hands were tied, and they sent Dartmouth to the highest possible Eastern seed, Cornell.
Who to send to Wisconsin? Given that Dartmouth was locked in to Cornell, the committee chose to “protect” Wisconsin by sending them the next-lowest ranked team, UMD, which conveniently created an additional local matchup. The NCAA’s long-proven willingness to fly three teams for the women’s quarterfinals suggests the committee would have been able to send an eastern 7th-ranked team to Wisconsin, but the rankings did not fall that way.
Q: The USCHO Women’s Division I PairWise Rankings have Mercyhurst fourth, and Minnesota and Boston College tied for fifth, with Minnesota having the edge in the RPI tiebreaker. Yet the NCAA Selection Committee appears to have ranked Boston College fourth, Minnesota fifth, and Mercyhurst sixth? What gives?
The comparisons in the PairWise Rankings are at best an approximation of how the committee sees them. The committee saw quite a few comparisons differently from the PairWise.
The women’s championship handbook includes the sentences, “When comparing two teams, the committee reserves the right to weight criteria differently based on relative team performance. For example, if there is only a tiny fraction of a difference between two teams records’ vs. common opponents, and a large difference in their results vs. teams under consideration, the committee may weight results vs. teams under consideration more heavily than common opponents.”
Translation: you can’t be sure what we’re going to do unless one team beats another in all criteria.
The PairWise ranks Mercyhurst above Minnesota. However, the commitee likely recognized that Minnesota had a superior performance against two common opponents with Mercyhurst, Wisconsin and Bemidji State, though Mercyhurst has the better common opponent win percentage in the PairWise. The committee likely gave Minnesota the comparison based on its superior record vs. teams in the RPI top 12 (6-8-2 for Minnesota vs. 1-2 for Mercyhurst) in addition to this common opponent edge.
The Pairwise gives Minnesota the comparison with Boston College. However, the committee likely choose to give greater weight to BC’s superior record vs. the RPI top 12 (8-3-1 compared to Minnesota’s
6-8-2) over Minnesota’s RPI edge of .0003.
The PairWise gives UMD the comparison with Minnesota. However, the committee likely recognized that Minnesota had the far superior RPI and that its disadvantage in common opponents and record vs. RPI top
12 was due to Minnesota’s loss to Wisconsin in the WCHA final, which UMD did not qualify for.
I believe the committee agreed with PairWise interpretation of Mercyhurst vs. BC. Mercyhurst had the edge in RPI and common opponents. BC fared relatively poorly against common opponents Maine and St. Lawrence.
Under these comparisons, Mercyhurst, BC, and Minnesota are all tied for 4th. While RPI is typically considered the tiebreaker in these situations, the RPI differences between the teams were all tiny. The commitee must have instead used BC’s superior record vs. the RPI top 12 (8-3-1 compared to Minnesota’s 6-8-2 and Mercyhurst’s 1-2) to rank the teams as BC No. 4, Minnesota No. 5, and Mercyhurst No. 6.
Q: Why did the commitee choose to take Dartmouth over North Dakota?
Here, the PairWise rankings get it right. Dartmouth had the superior RPI (.5496 vs. .5476). The teams had the same win percentage vs. teams in the RPI top 12. Dartmouth had a superior record vs. common opponents (Dartmouth beat Vermont while North Dakota went 1-0-1 against them). North Dakota beats Dartmouth in no criteria, so there’s no room for subjectivity from the committee.
