clock menu more-arrow no yes

Filed under:

College Hockey's Tournament Selection Process Flawed

New, 5 comments

When asked to envision the selection process for college basketball's NCAA Tournament, you probably conjure up images of cigar smoke-filled rooms and closed-door sessions with NCAA big wigs. A secretive group whose sole purpose is to broker deals and subjectively argue the merits of whether 17-12 Connecticut really deserves an at-large bid to this year's tournament.

From everything I've read, you wouldn't be far off.  

Not so for college hockey's NCAA Tournament field of 16. College hockey doesn't believe in the subjective merit of arguing the case of Team X over Team Y in an attempt to whittle down the tournament field from 58 to 16. Instead, to do the job, college hockey employs a statistical formula known as the PairWise Rankings

This year, however, I'm here to tell you that the PairWise is failing college hockey. Take, for instance, the case of the Vermont Catamounts men's hockey team.

 

As I was taking a look at one set of recent NCAA Tournament bracketology posts circulating around the blogosphere, I was immediately draw to the number of Hockey East teams projected to be in the field of 16. The answer 3 - Boston College, New Hampshire and Vermont.

Wait, what? Vermont? The same Vermont that has won just three games in their last 10? A team that before this weekend's surprising sweep of Boston University, had managed just 10 points in 12 Hockey East contests in 2010. A team that currently sits in 7th place in the Hockey East standings with one weekend to go, with a very real possibility of missing out on the Hockey East Tournament entirely. 

Surely the PairWise rankings are mistaken. Clearly Vermont isn't the third best program in Hockey East, right?

Wrong.

Upon closer inspection, Vermont currently sits tied for 12th place in the PWR rankings. That's good for the third best Hockey East program behind BC (tied for 6th) and New Hampshire (10th). After Vermont, there a glut of Hockey East teams further down in the PairWise rankings, including Maine (17th), Northeastern (tied 21st), UMass Lowell (tied 21st) and Boston University (23rd). 

The PWR uses four criteria which are combined to make a comparison: RPI rating (a measure of strength of schedule), record against other teams in the RPI top 25 (called TUC, for teams under consideration), record against common opponents and head-to-head competition.

Here's a look at the current comparison between Vermont and Maine (arguably the true third best team in the league):

Vermont Maine
.5318 1 RPI 0 .5283
9-5-3 .6176 1 Record against RPI Top 25 0 .4750 9-10-1
10-8-5 .5435 0 Common Opponents 0 .5435 12-10-1
0-2-1 0 Head-to-Head 2 2-0-1

In this case, Vermont and Maine are tied at 2. UVM wins the strength of schedule and record against teams in the Top 25 category, the two teams split record against common opponents (both currently at .5435), and Maine gets 2 points for winning the head-to-head season series 2-0-1. In cases of a tie - as in this case - the first tiebreaker goes back to RPI, or strength of schedule, which Vermont has a slight edge.

The problem here is that strength of schedule is largely out of control of a program once the season starts. There are so few non-conference matchups that teams are largely reliant on the rest of Hockey East to boost their SOS. In the same vein, record against RPI Top 25 isn't exactly a fair measure either. Not all Top 25 teams are created equal, and each Hockey East team only gets 7 chances to play programs in non-conference play and differentiate themselves in the record against RPI Top 25 category.

Finally, the PWR doesn't features any measure of how teams are playing down the stretch; something the NCAA basketball selection committee takes a very close look at when determining who's in and who's out. In college hockey, all games are weighted equally. Not that it makes a huge difference in the comparison between Maine and Vermont, but there's certainly a difference between Maine going 5-4-1 in your last 10 games (including a 1-0-1 record against UVM) and Vermont going 3-3-4 (and 0-1-1 at home vs. Maine). 

While it may seem like this isn't much of a big deal, the difference is if the season ended today, Vermont would be in the field of 16 while Maine would be out of the tournament.  

For another comparison that hits closer to home, take the comparison between BC and Bemidji State for the coveted final #1 seed. Currently College Hockey America's Bemidji State holds the head-to-head PWR comparison over BC for the #4 slot. This would translate into the NCAA Tournament's final #1 seed.

Bemidji State Boston College
.5520 1 RPI 0 .5499
6-2-1 .7222 1 Record against RPI Top 25 0 .5000 8-8-2
0-0-0 N/A 0 Common Opponents 0 N/A 0-0-0
0-0-0 0 Head-to-Head 0 0-0-0

Here, in cases where the two teams haven't faced one another and share no common opponents, the PWR really seems to break down. The only criteria the Beavers and Eagles are compared on then are strength of schedule and record against RPI Top 25.

Regarding strength of schedule, this is a case where you are comparing complete apples to oranges. BC's strength of schedule is largely tied up in how the rest of Hockey East fares. York's squad really gets only five opportunities to play teams out of conference between the Hockey East regular season schedule and the two non-conference games in the Beanpot.

Contrast that with CHA's Bemidji State, who has the opportunity to schedule 14 non-conference games  in between snacking on the rest of the CHA. Bemidji State has gone 13-3-0 this season in the CHA, and you won't find any other CHA team even sniffing the RPI Top 25 this season. Don't get me wrong. Bemidji State has fared well in non-conference play (9-5-2) and has some quality wins under their belt, including a win over #2 Miami. However, to compare strength of schedule between a team that gets to coast through their conference schedule and play 14 non-conference games vs. a team that has to grind through the Hockey East schedule and gets to choose only 5 programs to play outside of Hockey East plus the Beanpot, well, it doesn't seem like a fair fight.

In addition, the comparison between record against Top 25 teams isn't exactly a fair fight either. As stated above, not all Top 25 teams are created equal. Plus BC has played over twice as many games against teams in the Top 25 as Bemidji State has. It's no wonder that Bemidji State owns a .722 winning percentage (6-2-1) against the RPI Top 25 while BC is only at .500 (8-8-2).

When comparing teams like BC and Bemidji State where there are no common opponents and no head-to-head results, the remaining two measures hardly seem objective at all. Hypothetically, if you asked BC to play Bemidji State's schedule and Bemidji State to play BC's schedule, I'd imagine the results would be very different.

Luckily for BC and for Maine, there is plenty of hockey left to be played. Maine has the opportunity to clinch home ice in the Hockey East quarterfinals this weekend and play their way into the PWR top 12 and earn a NCAA Tournament berth. Similarly, Vermont has the opportunity to prove me wrong down the stretch and earn a spot in the Hockey East tournament and the NCAAs. And finally, if BC keeps winning, I'd imagine there will be enough quality wins on the schedule to push BC past Bemidji State and into one of the top 4 spots in the PWR.

But perhaps it's time to put a little subjectivity back into the NCAA ice hockey tournament selection process. It seems to work for college basketball. If anything, putting the argument back into the selection process will give people more reason to talk about college hockey, giving the sport the exposure that it desperately needs.