Last Modified: 2012 March 25
Joe Schlobotnik / joe@amurgsval.org
URL for this frameset: http://slack.net/~whelan/tbrw/tbrw.cgi?2000/pairwise.000306.shtml
Game results taken from US College Hockey Online's Division I composite schedule
This analysis is based on the pairwise comparisons at the time it was written. You can also go through this process interactively with the up-to-date results using the "You Are The Committee" script.
The regular seasons of the six Division I leagues have come to an end, and with two weeks of conference playoffs remaining, we now know several things about the seeding of the NCAA tournament for certain. Wisconsin, Michigan, St. Lawrence and Boston University will all take part, as regular season champions of their respective conferences. Army, Brown, Union, UMass-Lowell, Fairfield, American International, Alaska-Fairbanks, and Ohio State have all finished their seasons with losing records and will not be considered for the tournament. Holy Cross and Connecticut each have up to three games to play in the Metro-Atlantic Athletic Conference playoffs, but since even three wins will leave either team below .500, they cannot be considered for the tournament. Michigan Tech, Yale, Dartmouth, Western Michigan, Bowling Green, Miami, Nebraska-Omaha, UMass-Amherst, Denver, Minnesota-Duluth, Merrimack, and Northeastern are all far enough below .500 that they can only become teams under consideration by winning the relevant conference tournament, thereby gaining an automatic berth. Air Force currently has a record of 9-9-1, exactly at .500, but they can only become a team under consideration if they play one more Division I game to reach the minimum of 20. This means they have to defeat Findlay in the College Hockey America play-in game and thereby face Niagara in the semifinals. They then of course have to win this game to avoid falling below .500.
Up to four more bids will be given to the winners of the four major conference tournaments, and the remaining four to eight at-large bids will be given on the basis of pairwise comparisons among the teams under consideration. Here are how those comparisons look at the moment for the 26 teams who have a non-losing record in 20 or more Division I games:
Also to be taken into consideration is the relative strengths of the various conferences, to allow for cases like last season where Quinnipiac's weak schedule allowed them to win many pairwise comparisons. Looking at a table of conferences' performance agaisnt one another and the average Ratings Percentage Index within each conference:
| Conference | Avg RPI | vs HE | vs WCHA | vs CCHA | vs CHA | vs ECAC | vs MAAC | Leader | Opp RPI |
|---|---|---|---|---|---|---|---|---|---|
| Hockey East (H) | .5345 | 13-6 | 10-7 | 3-2-1 | 26-15-3 | 5-0 | NH | .5268 | |
| WCHA (W) | .5155 | 6-13 | 15-12-1 | 3-1 | 10-2-1 | 0-0 | Wi | .5040 | |
| CCHA (C) | .4905 | 7-10 | 12-15-1 | 1-2-1 | 11-10-2 | 0-0 | Mi | .4831 | |
| CHA (A) | .4904 | 2-3-1 | 1-3 | 2-1-1 | 5-8-1 | 15-5-2 | Ni | .4517 | |
| ECAC (E) | .4902 | 15-26-3 | 2-10-1 | 10-11-2 | 8-5-1 | 4-0 | SL | .4817 | |
| MAAC (M) | .4531 | 0-5 | 0-0 | 0-0 | 5-15-2 | 0-4 | Qn | .4360 |
It's easy to see that the MAAC is once again a step below the other conferences, and so that Quinnipiac will almost certainly be left out again this year. It's a little tougher to judge College Hockey America; the average RPI of the three Division I teams is just below the average of those of the 12 clubs in each of the ECAC and CCHA. However, most of that is due to Niagara's own RPI; their two Division I conference opponents have a considerally lower average RPI. With Niagara relatively close to the tournament cutoff, it's reasonable to think that a slight discounting of their pairwise comparisons would also lead to their omission from the tournament field. This analysis is also in line with more weak-schedule-proof rating systems such as KRACH, in which Niagara is 19th in the nation and Quinnipiac 43rd (out of 54).
Excluding Niagara and the various MAAC teams, seven teams in addition to the four conference regular season champions win pairwise comparisons with everyone else: Maine, New Hampshire, North Dakota, Colgate, Boston College, St. Cloud and Michigan State. If the tournament were seeded today, the remaining at-large berth would go to one of the following teams:
| MSU-Mankato (W) | 8 | .526 | CC | Mn | RP | Ck | LS | NM | Pv | Cr | ||
| CO College (W) | 8 | .536 | FS | Mn | RP | Ck | LS | NM | Pv | Cr | ||
| Ferris State (C) | 6 | .529 | Mk | RP | Ck | NM | Pv | Cr | ||||
| Minnesota (W) | 5 | .539 | FS | RP | LS | NM | Pv | |||||
| RPI (E) | 5 | .535 | Ck | LS | NM | Pv | Cr | |||||
| Clarkson (E) | 4 | .521 | Mn | LS | Pv | Cr | ||||||
| Lake Superior (C) | 3 | .512 | FS | NM | Cr | |||||||
| Northern Mich (C) | 2 | .523 | Ck | Pv | ||||||||
| Providence (H) | 2 | .515 | LS | Cr | ||||||||
| Cornell (E) | 2 | .500 | Mn | NM |
Mankato State and Colorado College both win comparisons with all but one of the other potential entrants, and Mankato wins the individual comparison with CC, so it looks like the Mavericks would get the final at-large bid in this scenario. (The only way they would not would be if the committee looked at Mankato, CC and Ferris State together; with one comparison win a piece the tie would be resolved by the Ratings Percentage Index, with CC having the highest RPI of the bunch. However, there seems to be little reason to include Ferris State and not, for example, Minnesota, in the final analysis.) This gives us the following field of 12:
| West | East | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Wisconsin (W) | 5 | .619 | ND | Mi | MS | SC | Mk | 1 | Maine (H) | 4 | .594 | SL | Cg | BU | BC | ||
| North Dakota (W) | 4 | .596 | Mi | MS | SC | Mk | 2 | New Hampshire (H) | 3 | .597 | Me | SL | BC | ||||
| Michigan (C) | 3 | .572 | MS | SC | Mk | 3 | St Lawrence (E) | 3 | .584 | Cg | BU | BC | |||||
| Mich State (C) | 2 | .553 | SC | Mk | 4 | Colgate (E) | 3 | .570 | NH | BU | BC | ||||||
| St Cloud (W) | 1 | .551 | Mk | 5 | Boston Univ (H) | 2 | .588 | NH | BC | ||||||||
| MSU-Mankato (W) | 0 | .526 | 6 | Boston Coll (H) | 0 | .579 | |||||||||||
The two best teams in each region according to pairwise comparisons normally receive first-round byes in the NCAA regionals. In the West this is Wisconsin and North Dakota. In the East, things are a little more complicated, but Maine and New Hampshire seem to end up with the two byes no matter how they're reckoned (either by giving one to Maine and chosing New Hampshire above SLU and Colgate based on ratings percentage index or by focussing on comparisons among those four teams, in which case Maine and UNH both win two comparisons to SLU and Colgate's one each). Note that if any of the regular season champions from the major conferences also win their conference tournaments, they will receive automatic byes, which means that Michigan, SLU, or BU could change things dramatically.
Next we swap two Western teams for two Eastern teams, and at the moment the pairwise comparisons make things easy by telling us to swap the third and fourth teams from the WCHA and Hockey East, giving the following regionals:
| West | East | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Wisconsin (W) | 1 | .619 | ND | 1 | New Hampshire (H) | 1 | .597 | Me | ||||
| North Dakota (W) | 0 | .596 | 2 | Maine (H) | 0 | .594 | ||||||
| Boston Univ (H) | 3 | .588 | BC | Mi | MS | 3 | St Lawrence (E) | 3 | .584 | Cg | SC | Mk |
| Boston Coll (H) | 2 | .579 | Mi | MS | 4 | Colgate (E) | 2 | .570 | SC | Mk | ||
| Michigan (C) | 1 | .572 | MS | 5 | St Cloud (W) | 1 | .551 | Mk | ||||
| Mich State (C) | 0 | .553 | 6 | MSU-Mankato (W) | 0 | .526 | ||||||
The natural seedings produce no intraconference games in the regionals at all, so the brackets are a slam-dunk:
5W Michigan (C) 6E MSU-Mankato (W)
4W Boston Coll (H) 3E St Lawrence (E)
1W Wisconsin (W) --+--2E Maine (H)
|
2W North Dakota (W) --+--1E New Hampshire (H)
3W Boston Univ (H) 4E Colgate (E)
6W Mich State (C) 5E St Cloud (W)
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.