Pairwise Rankings Analysis: Selection And Seeding

In past seasons, the selection of teams for the NCAA tournament was absolutely straightforward. Last year, however, a new wrinkle created some ambiguity in the process when the selection committee added language that allowed it to reject a team based on conference strength.

This was put into place primarily because of the insular nature of the MAAC’s schedule last year. And sure enough, Quinnipiac, the MAAC regular-season champion, finished high enough in the Pairwise Rankings (PWR) to be selected for the NCAA tournament. However, Quinnipiac was left off the list of selected teams, precisely the outcome the language was intended to allow.

Again this year, Quinnipiac finished in the top 12 teams in PWR, but there is no real question — Quinnipiac lost every game it played versus teams from the “big four” conferences, and no one expects Quinnipiac to get a bid. Next season, the MAAC will probably have an automatic berth to award, but for now, the MAAC championship is the highest goal in that league.

The question this year regards Niagara. Niagara, like Quinnipiac with the MAAC, plays in a new Division I league, College Hockey America. Most of the teams in the CHA are not yet competitive with the rest of Division I, and as such, Niagara’s 20-0-1 conference record isn’t particularly enlightening.

Niagara had some impressive wins this season, including a 3-0 win over nationally-ranked Boston University and a 2-1 victory over NCAA bubble team Rensselaer. However, for every impressive win, there was an equally dreadful loss. Niagara seemed to falter late in the season, and suffered a noggin-scratchin’ 2-1 loss to MAAC school Canisius, as well as a loss and an overtime win against the number-10 team in the CCHA, Western Michigan.

Overall, Niagara finished 9-6-2 against members of the four major conferences, with those opponents making up a field of pretty representative teams in Division I. The question is, is that good enough to attract the selection committee’s eye? We’ll find out for sure when the teams are announced.

The PWR has been very accurate in years past for determining the field of teams to be taken in the selection process, though the seeding process, which must concern itself with minutiae like intraconference matchups and even gate attendance, often remains a mystery. This year, we’ll consider cases with and without Niagara earning one of the remaining at-large bids.

Here are the top 14 teams in the PWR, taking into account all of the results from the various conference tournaments:

Rk Team                GP  W- L- T  Win%  Rk     RPI  Rk  PWR
1 Wisconsin 40 31- 8- 1 0.7875 1 | 0.6095 1 | 25
2 Maine 38 26- 7- 5 0.7500 4 | 0.6082 2 | 24
3 North Dakota 39 26- 8- 5 0.7308 5 | 0.5982 3 | 23
4 St Lawrence 33 24- 7- 2 0.7576 3 | 0.5886 5 | 22
5 Niagara 27 18- 7- 2 0.7037 8 | 0.5723 8 | 20
6 New Hampshire 37 23- 8- 6 0.7027 9 | 0.5921 4 | 19
7 Colgate 33 23- 8- 2 0.7273 6 | 0.5701 9 | 19
8 Boston College 38 26-11- 1 0.6974 11 | 0.5835 6 | 18
9 Boston University 40 24- 9- 7 0.6875 12 | 0.5830 7 | 18
10 Quinnipiac 28 20- 5- 3 0.7679 2 | 0.5657 11 | 16
11 Michigan State 40 26-10- 4 0.7000 10 | 0.5631 12 | 16
12 Michigan 39 26- 9- 4 0.7179 7 | 0.5680 10 | 14
13 St Cloud 37 21-13- 3 0.6081 15 | 0.5459 13 | 12
14 MSU-Mankato 35 19-13- 3 0.5857 16 | 0.5250 16 | 12

In the 10th slot is Quinnipiac, which is generally regarded as out of the tournament selection process. All further analysis will assume Quinnipiac will not receive a bid.

Now we come to the two cases: with or without Niagara.

Excluding Niagara, we have the following top 12 teams:

Wisconsin
Maine
North Dakota
St. Lawrence
New Hampshire
Colgate
Boston College
Boston University
Michigan State
Michigan
St. Cloud
MSU-Mankato

Including Niagara, we have the same first ten teams, plus Niagara to make 11. The 12th spot is a tie between WCHA teams St Cloud and MSU-Mankato. Let’s look at the individual comparison between the two:

    MSU-Mankato vs St Cloud
RPI 0.5252 0 0.5465 1
L16 10- 4- 2 1 9- 5- 2 0
TUC 7- 6- 2 1 6- 7- 3 0
H2H 0 2
COP 17- 9- 3 1 17-11- 3 0
============================================
PTS 3 3

It’s a tie between the two teams, but ties are broken by the RPI, and so St Cloud wins, by the narrowest of margins.

So the list is now:

Wisconsin
Maine
North Dakota
St. Lawrence
Niagara
New Hampshire
Colgate
Boston College
Boston University
Michigan State
Michigan
St. Cloud

According to these two options, the choice before the committee is between MSU-Mankato and Niagara, a selection which simply cannot be made by using the information at hand. The committee will hash it out, but until then it’s anyone’s guess.

Seeding is much more difficult to predict. In the past, the PWR has been an accurate predictor of teams selected, but not particularly accurate in terms of seeding.

Here’s one attempt at seeding, with Niagara (which is considered a West team, since the CHA is a “western” conference in the eyes of the NCAA) in the field.

Start out by placing the teams in their respective regions:

West                       East
1 Wisconsin 1 Maine
2 North Dakota 2 St. Lawrence
3 Niagara 3 New Hampshire
4 Michigan State 4 Colgate
5 Michigan 5 Boston College
6 St. Cloud 6 Boston University

Traditionally, the practice is to swap the bottom two schools in each bracket and reseed:

West                       East
1 Wisconsin 1 Maine
2 North Dakota 2 St. Lawrence
3 Niagara 3 New Hampshire
4 Boston College 4 Colgate
5 Boston University 5 Michigan
6 Michigan State 6 St. Cloud

This looks pretty good, except Boston College and Boston University would have a first-round matchup. That should be avoided, which is easily done by swapping BC and Niagara. In the end, we have:

West                       East
1 Wisconsin 1 Maine
2 North Dakota 2 St. Lawrence
3 Boston College 3 New Hampshire
4 Niagara 4 Colgate
5 Boston University 5 Michigan
6 Michigan State 6 St. Cloud

This particular seeding accomplishes many things, including:

Avoiding all first and potential second-round matchups between intraconference foes.

Keeping St. Lawrence and Colgate, the two New York schools, in Albany, which would increase the draw there.

Admittedly, St Cloud is sent East but would presumably be a bigger draw in Minneapolis. However, with WCHA opponents Wisconsin and North Dakota getting the two byes, it would be impossible to keep St Cloud in the West and avoid a potential second-round matchup.

If Niagara is left out, we have the following, after swapping and reseeding:

West                       East
1 Wisconsin 1 Maine
2 North Dakota 2 St. Lawrence
3 Boston College 3 New Hampshire
4 Boston University 4 Colgate
5 Michigan State 5 St. Cloud
6 Michigan 6 MSU-Mankato

With this seeding, there are no first- or potential second-round matchups between conference-mates, and again Colgate and St Lawrence are kept in Albany.

Another possibility is that the committee will give Niagara a bid, but disregard its PWR standing for seeding purposes, instead regarding Niagara initially as a sixth seed, much as MSU-Mankato would be. In this case, the seeding process proceeds exactly as in the scenario above, but with Niagara finishing up in MSU-Mankato’s spot, as the sixth East seed. That is, we get final seedings as follows:

West                       East
1 Wisconsin 1 Maine
2 North Dakota 2 St. Lawrence
3 Boston College 3 New Hampshire
4 Boston University 4 Colgate
5 Michigan State 5 St. Cloud
6 Michigan 6 Niagara

The NCAA seedings will be announced on ESPN2 at 1:30 Eastern time on Sunday, March 19.