If the season ended today, 2000 February 15

© 2000, Joe Schlobotnik (archives)

URL for this frameset: http://elynah.com/tbrw/tbrw.cgi?2000/pairwise.000215.shtml

Game results taken from US College Hockey Online's Division I composite schedule

You can also go through this process interactively with the up-to-date results using the "You Are The Committee" script.

With the seeding of the NCAA tournament a little more than a month a way, let's take our first peek of the season at how the tournament selection procedure might play out. There are increasingly many variables these days, but this way we can look at how some of the issues might work themselves out.

As always, the first step is to assign automatic bids to the regular season and playoff champions of the four established conferences. In keeping with the tradition of past years, let's begin our imaginary selection process by assuming the regular season titles go to the current league leaders--Wisconsin in the WCHA, Michigan in the CCHA, St. Lawrence in the ECAC, and Boston University in Hockey East--and leave the tournament champions unspecified. The at-large bids are assigned using the pairwise comparisons among the 25 teams with non-losing Division I records:

Because the criteria in use by the NCAA do a poor job of judging the overall quality of teams which play most of their games in a markedly weaker league than the others, the selection committee asserted at the start of last season its "right to evaluate each team based on the relative strength of their respective conference". In practice, this meant excluding MAAC regular season champion Quinnipiac despite their success in the pairwise comparisons. Again we find Quinnipiac, along with Niagara from the fledgling College Hockey America conference, among the top teams in the nation according to pairwise comparisons. Here is a look at the six Division I conferences' performances against one another:

It seems once again clear that the level of play in the MAAC is not up to that of the other conferences, and it should ba safe to assume that Quinnipiac will once again be excluded. The CHA situation is a little trickier. The committee mentioned "overall conference ratings percentage index" as a means to determine competetive equity, and the average RPI of the three Division I teams in CHA is actually higher than that of two of the established conferences. However, a major contribution of that is Niagara's own RPI; as the last column in the table above shows, the average RPI of their two eligible conference opponents is much lower. Nonetheless, Niagara have played competitively against a schedule which contained several strong teams; compiling a record of 7-5-1 against major conference teams. However, their ratings are also inflated by a schedule which contains eight games (all wins) against MAAC and CHA opponents. (As indicated in the table above, no MAAC team has ever beaten a member of an established conference, while the other two D1 CHA teams, Army and Air Force, have combined for a record of 1-8-1 against major conference opposition.) So the committee would seem to have a judgement call to make on the question of Niagara. I suspect they will split the difference, by allowing the Purple Eagles into the tournament on the basis of their pairwise comparisons, but using the conference strength argument to overrule those comparisons when it comes to seeding the tournament. (As detailed on our comparison of ranking systems page, alternate systems which account for strength of schedule more carefully gauge Niagara at around #15 in the nation, which would put them just on the outside edge of the tournament "bubble".)

So, once we exclude Quinnipiac, we find that New Hampshire, North Dakota, Niagara, Maine, Boston College and Colgate win pairwise comparisons with all the other teams still in the running for at-large bids. That leaves two bids to be filled from the "bubble" consisting of Minnesota, Michigan State, St. Cloud, RPI, and Ferris State. Of those five, Minnesota and Michigan State both win comparisons with three of the other four, and so would be chosen to the tournament. (Even if we were to drop RPI and FSU from the bubble, leaving the three remaining teams winning one comparison and losing the other within their "mini-bubble", Minnesota and Michigan State both have higher RPIs than SCSU.)

That gives us a field of twelve qualifiers, seven of which (if we consider Niagara to be in the East) are Eastern teams, and only five from the West. Considering the lowest-ranked Eastern team, Colgate, as honorary Westerners gives the following starting point:

(An alternative approach would be to throw Niagara in with the West; ordinarily there'd be the danger of shipping them back to the East, which the committee would presumably rather not do, but Niagara's misleading pairwise comparisons would actually come in handy here, since they'd peg them as a natural 3 seed in the West; not strong enough for a bye, but not weak enough to be shipped East.)

Next, we need to identify the top two teams in each region and assign them the two first-round byes (since we have assigned no conference tournament champions, there are no automatic byes in our hypothetical seeding). In the West, they clearly go to Wisconsin and North Dakota. It's sort of a near thing in the East, since the top three teams split their pairwise comparisons, but using RPI as a tiebreaker gives them to BU and UNH; even if this were not the case, it seems like a good guess that the committee would play the "conference strength" card to avoid giving Niagara a bye.

The next step is to switch two Eastern teams with two Western teams to obtain well-mixed regionals. By default, these are the bottom two teams in each region. However, since Minnesota are hosting the West regional, they must stay in their own region. This makes the obvious "Western" swapees Colgate (who the committee will likely want in the East anyway, as the closest participants to the regional site of Albany) and Michigan State. From the East, the comparisons tell us to ship out BC and SLU. However, this would seem like a probable time for the committee to overrule Niagara's comparisons on the basis of a weaker schedule on their part. Furthermore, shipping SLU and Niagara would leave the four Hockey East teams in the same regional, which would be problematic from the point of view of avoiding intraconference matchups. So instead, let's try sending Niagara and BC to the West.

Now, there's one potential second-round WCHA matchup in the West, but that's absolutely unavoidable with both bye teams and the host being from the same conference. In the East, the natural seeds give a first-round game between SLU and Colgate, but that can be avoided by rearranging the seeds. Maine is automatically looking at a second-round game with a fellow Hockey East member, if they stay in their own region, but there's no way around that if we insist on shipping on only two Eastern teams including Niagara. So a likely choice for tournament brackets could be

5W Michigan (C)                    6E Colgate (E)       
4W Boston Coll (H)                 3E Maine (H)         
     1W Wisconsin (W)    --+--2E New Hampshire (H) 
                           | 
     2W North Dakota (W) --+--1E Boston Univ (H)   
3W Niagara (A)                     4E Mich State (C)    
6W Minnesota (W)                   5E St Lawrence (E)  

On the other hand, if the committee is really keen to avoid intraconference matchups in the East (which may depend on which matchups have already occurred in the Hockey East playoffs), they might do what they did last year and swap and additional pair of teams, in this case Maine and Michigan, to give the following regionals:

This is also what would happen if the committee considered Niagara to be a Western team and then shipped the two Hockey East teams West. Shipping Maine and BC in place of Colgate and SLU goes against the pairwise comparisons, while helping with the goal of avoiding intraconference matchups. Another important factor is attendance. Colgate is close to the regional site in Albany, St. Lawrence less so, but still in the same conference as hosts Rensselaer. On the other hand, Maine and BC are large schools with big fan bases, although the same could also be said for UNH and BU, the two Hockey East teams already playing in that regional. At any rate, if this option is chosen, some juggling of seeds is necessary to avoid first-round intraconference matchups. Here are the resulting brackets:

5W Boston Coll (H)                 6E Colgate (E)       
4W Niagara (A)                     3E Michigan (C)      
     1W Wisconsin (W)    --+--2E New Hampshire (H) 
                           | 
     2W North Dakota (W) --+--1E Boston Univ (H)   
3W Maine (H)                       4E Mich State (C)    
6W Minnesota (W)                   5E St Lawrence (E)  

The Gory Details

If you want to have a look at why each pairwise comparison turned out the way it did, you can click on the individual comparisons in the table at the top of this article for a breakdown of criteria.