Quantcast
Column

College Hockey:
NCAA Tourney Analysis

Editor’s note: This analysis was compiled before the Colorado College-Denver game Thursday evening, and does not account for that result.

This analysis is based on an examination of the Pairwise Ranking, the computer ranking system used by the NCAA tournament selection committee. The PWR compares each tourney-eligible team in Division I which has a .500 or better overall record with every other such team.

Each such comparison is worth one PWR point, and is decided based upon five numerical criteria. The best possible score is therefore earned by winning the comparison with every other team: currently, this is held by Michigan, with 20 out of 20 possible PWR points. The lowest possible score is 0, for a team which wins no comparisons.

We can project the 12 NCAA tourney bids via detailed looks at each criteria of comparison between all of these teams, the so-called “teams under consideration.” The key is figuring out which comparisons might change as a result of future games, and which won’t no matter what.

For example, some comparisons are very lopsided, such as Minnesota over Michigan State. This one is currently 4-1, and there is no chance of MSU reversing it. The Gophers’ advantage is simply too great.

On the other hand, some other team-vs-team comparisons are very close, so that a slight adjustment in one or more of the five comparison criteria — RPI rating, record versus teams under consideration, record in last 20 games, head-to-head record, and record vs. common opponents — could turn the comparison the other way.

Figuring out when a comparison might change is based on several years of experience in tracking these things. I could be wrong on some of my guesses, of course, but I try to be very conservative in my claims.

Four teams already have bids: Michigan, North Dakota, Clarkson and BU are in, thanks to the automatic bid the NCAA awards to the regular-season champions of the four major conferences. In addition, any team still in conference playoffs could earn an automatic bid by winning its league tourney.

All that we can analyze now is who would (and would not) get any at-large bids after the autos have been decided. In the worst case for at-large teams (the case where no regular-season champion wins its conference tournament) there will be only four out of the 12 original NCAA tourney slots open for them.

In the absolute worst case for a team fighting for an at-large bid, all four remaining auto bids would go to teams that otherwise rank below them. The important thing to remember for teams trying to get an at-large bid is that they are guaranteed to have lost at least one game this weekend. That’s why they’re at-larges, and not tourney champs (exception: Vermont, which is no longer in the ECAC playoffs and therefore cannot win or lose any more games at all).

That is, if you win all of your games this weekend, you get an auto-bid and do not have to worry about your criteria for getting selected. So, at-large candidates will tend to fall a bit from their current rankings. Whoever falls the least will probably get the bids.

Minnesota, New Hampshire and Miami all survive the absolute worst-case scenarios. In this situation, Miami and UNH will lose at most six criteria comparisons (and win 14). If four lower-ranked teams pull upsets, they still beat 10 remaining teams — and you only need to beat eight of them (since Maine is tournament-ineligible).

That is, 21 teams under consideration, minus 12 bids minus one for Maine equals eight. That’s the number of teams you have to beat to get in. Minnesota is 13-7 at worst, so they are OK too. Call these three teams “locks” for the tournament.

Vermont and Denver are not quite locks, but are “virtual locks.” If these teams lose all of the close criteria comparisons that they have, they could end up beating 12 and losing to eight of the teams under consideration.

If four upsets happen in the four conference tourneys — such as Bowling Green, Boston College, Princeton and Colorado College, then there would be eight autobids, the three locks above and only one spot left for either UVM or DU. It is impossible to tell which of these two would get the 12th spot. If one regular-season winner or a lock team wins a league playoff, then both will get in, so it is very (very) unlikley that either will miss the big dance.

MSU, Cornell and CC are clearly the “bubble” teams. MSU and Cornell could fall as far as eight criteria comparison wins and 12 losses, which is still good enough to get in. These two teams can probably only get bumped by lower-ranked teams getting auto bids.

CC is a bit worse off, possibly going 7-13, so the Tigers could simply get caught by a team currently beneath them in the rankings.

Said another way, for any of these bubble teams to get an at-large bid, the previous nine teams must have already been selected (since these three are way behind them). That only leaves three spots, so if a very-lowly ranked team gets an auto bid, or if one of the immediately lower teams can catch them on the criteria, any of these can get booted from the field.

It is too hard to tell which of the three bubble teams would get bumped, but CC is the most likely to drop down.

“In the hunt” teams are those that would not qualify today, but could get an at-large bid if things went well for them. St. Cloud is one of these, since it could win its comparisons with MSU and CC (but not Cornell) and become 9-11. If SCSU does climb up for an at-large bid, it is not clear which of the three bubble teams would get bumped, as that depends on the criteria comparisons between the three teams, most of which are volatile. But SCSU would most likely bump CC.

No one else is in the hunt, however. Princeton can overcome SCSU on the criteria, but not any of the bubble teams, so the most comparisons it could win would be seven (and you will need eight). RPI can catch MSU and Princeton on the criteria, but not SCSU or CC, so the highest it can climb is also 7-13.

BGSU and the lower teams under consideration that are not playing cannot even come close to moving up for an at-large bid. BC and Mass-Lowell cannot even finish at .500, so their only hope is to win out in the play-offs.

Of course, this does not mean Princeton and BGSU cannot get into the tourney: they just need to win their conference championships to do so. They will not get an at-large bid, however.

Seeding is not worth worrying about yet. After Friday’s results, the placement of teams within the regions will become much clearer. Then we can run actual scenarios of potential results and compute all of possible outcomes.

After all is said and done, listen to “Around the Rinks” to hear my post-mortem analysis.


The following is a self-policing forum for discussing views on this story. Comments that are derogatory, make personal attacks, are abusive, or contain profanity or racism will be removed at our discretion. USCHO.com is not responsible for comments posted by users. Please report any inappropriate or offensive comments by clicking the “Flag” link next to that comment in order to alert the moderator.

Please also keep “woofing,” taunting, and otherwise unsportsmanlike behavior to a minimum. Your posts will more than likely be deleted, and worse yet, you reflect badly on yourself, your favorite team and your conference.