2018 NCAA Women’s Hockey Tournament Selection Could Be A Headache

One of the quirks of the top level of women’s college hockey is that the NCAA tournament isn’t a “Division I” tournament at all. It’s actually designated a “National Collegiate” tournament — a rarity in the NCAA, as almost all sponsored tournaments are broken out into Division I, Division II, and/or Division III brackets.

Women’s hockey is different. There is a separate D-III tournament, but the D-I and D-II teams technically compete in the same tournament.

See for yourself:

Up until now, this has been irrelevant. Since the NCAA first sponsored a women’s hockey tournament in the 2000-2001 season, no Division II team (or Division I independent, for that matter) has even been eligible, let alone been ranked high enough for selection.

That eligibility issue will change next season. The six independents (D-I Holy Cross, Post, & Sacred Heart, and D-II Franklin Pierce, St. Anselm, & St. Michael’s) announced last week that they have formed a temporary, one-season scheduling alliance, setting up a pseudo-conference of sorts that will see each team play each other four times — a total of 20 games each.

Coincidentally (or perhaps not coincidentally at all), the NCAA requires that teams play at least 20 games against D-I or D-II competition to be considered for at-large selection to the tournament:

This had previously been the reason that no D-II teams had ever been eligible for the NCAA tournament — the D-II teams played in a conference with D-III teams, and the D-I independents played a schedule consisting almost exclusively of D-III teams as well. Of the six teams, only Sacred Heart has played any D-I opponents outside of this group in the last several years, getting pasted by lowly Brown in each of their matchups.

This creates a pretty big problem for the NCAA.

Selection for the NCAA tournament among eligible teams is based off RPI, Head-to-head competition, and Common Opponents:

Since these teams will be, essentially, only be playing each other, there’s a two-pronged issue here. First, with no head-to-head or common opponents, RPI will be the only ranking criteria relative to the rest of the D-I programs. Second, without any “connecting” games to compare the Group of 6 with the “regular” D-I teams, RPI will be mathematically unable to determine how good these teams are relative to the rest, and will effectively treat them like they’re in an “average” conference.

That’s right — even though the eyeball test puts these teams somewhere in the middle of the D-III pack, the math will essentially assign them an average strength of schedule relative to the rest of the D-I teams and rank them accordingly.

Don’t believe me? Using the BC Interruption Pairwise Calculator, I plugged in sample results from the Group of 6 scheduling alliance, and it doesn’t take much for at least one of these teams to have a high enough RPI for tournament selection.

It’s just a one-off example, but mathematically speaking, it’s very difficult to keep all six teams outside the top 7-8 in the Pairwise without all of them having more than 4 or 5 losses. It’s not unreasonable to expect one or two teams to separate themselves from the pack, like any other conference, and any kind of stratification among these teams will likely result in at least one of them giving the NCAA selection committee quite a problem to deal with.

We reached out to the Head Coach of St. Michael’s College, Chris Donovan, to discuss whether there was any consideration of tournament eligibility when setting up the scheduling alliance for 20 games each:

I don’t think so. This was more created out of a necessity to give our teams a schedule for ‘17-’18. We just discussed playing each other three or four times, and we thought four was better since [it would have been easier to schedule] six out-of-alliance games* rather than eleven or more.

*It’s worth noting that St. Michael’s out-of-alliance games would be against D-III competition, which would not affect their record for NCAA tournament consideration.

For one of the teams to be included in NCAA tournament, I guess that is a conversation for the NCAA championship committee.

He’s right, of course. So, we did have that conversation with the chair of the NCAA championship committee, and reached out to committee chair Sarah Fraser. Fraser had this to say on the subject:

The committee reviews the selection criteria on an annual basis. While there is no guarantee that the criteria will not change, it is not the committee's intent that changes in the criteria would eliminate teams from being eligible for selections.

Nothing explicitly stated here, but this would seem to imply that these teams would indeed be eligible for selection as the current criteria is written.

Would the committee, in a scenario where one or more teams is mathematically in the top 8 of the Pairwise, select these teams for the NCAA tournament? The tournament was, after all, set up specifically to be one in which the D-II teams were eligible to participate. They couldn’t be expected to complete a standard D-I schedule.

On the other hand, this is probably not a situation that the NCAA expected to see — six teams, far below the level of competition of the rest of Division I, let alone the top eight, allying together to play the number of games required to be eligible for selection, taking advantage of some of the mathematical flaws of RPI which could potentially see one (or more!) finding their way into the NCAA tournament conversation. So it remains to be seen whether the NCAA will follow the letter of their own selection criteria, whether they will ignore these teams entirely, or whether there will be an update to the criteria next season adding a requisite number of games against “traditional” D-I teams to be eligible for selection.