clock menu more-arrow no yes mobile

Filed under:

From Bad to Worse: NCAA Tweaks the PWR

I was getting ready to write my annual post on why the math system the NCAA uses to calculate who gets in to the NCAA tournament, the Pairwise Ranking, makes no sense, when yesterday, College Hockey News announced the NCAA decided to make the system ever worse, by reverting back to an old definition of what a "Team Under Consideration" is. Rather than only comparing the top 25 teams in the RPI, every team with an RPI of .5000 or better will be under consideration.

So why is this change like a worse version of Hitler(no offense, North Dakota fans)? The big reason is that it pushes the number of "Teams Under Consideration" from 25 up to 34. We'll start with the obvious fact that there's no reason 58% of the teams in college hockey need to be "under consideration" for the NCAA tournament. That's more than twice the number of teams that actually get into the tournament. You may as well change the name of the TUC comparison category to the "You're Not Michigan Tech"(YNT, for short) comparison.

Second, it does nothing to address one of the PWR's biggest flaws in the TUC--now YNT--"cliff". Michigan State and Quinnipiac sit just barely above the cut-off mark, while Niagara and RIT are just shy of the mark. If those four teams flipped with each other, it could cause some major changes in the rankings. Perhaps the effects of a team dropping off the cliff will be slightly less felt with so many more teams, and consequently, results, falling into the YNT category. But even if a team plays 20 YNT games, which would be towards the higher end, two games is still a pretty significant 10% of that record.

Third, it was a joke that a game against the #1 team in the country counted the same as a game against the #25 team in the country in the TUC category. Now, a game against the #1 team in the country counts the same as a game against a team that couldn't crack the top half of the college hockey world in that comparison category. That's especially troublesome when one conference has three of the top four teams.

But my biggest problem with this change is that it greatly increases the potential for outlying comparisons to change the final results of the PWR. This was already a problem when the PWR was looking at 25 teams, and the potential only increases with the more teams that you look at.

If I'm trying to decide who gets the last NCAA tournament at-large bid between RPI #15 Rensselear and RPI #16 Western Michigan, I don't really need some sort of complicated math system to tell me that each team had a better season than #33 Michigan State or #34 Quinnipiac. So why even bother comparing them? All it does by comparing those teams is give the potential for some fluke of circumstances where one team's huge advantage in RPI, which measures the entirety of a season, is canceled out by one or two good weekends against the right out-of-conference opponent.

The current table doesn't have too many outlying comparisons, but despite one great weekend, does anybody think Maine has been better than North Dakota this year? If you had to choose between RPI #10 Union and RPI # 28 Ferris State, which would you pick? Because there's a computer that says Ferris State's season has been the better of the two. Hopefully none of those flukey outlying comparisons have any effect on the final tournament bracket, let alone influencing decisions right on the NCAA tournament bubble the way it did in 2008, but the possibility definitely exists.

If anything, the move should have been made back to the 1990's, before the PWR was published and subsequently codified, when the NCAA tournament committee was comparing fewer teams, not more. The version of the PWR they looked at only cared about the few teams on the bubble for the NCAA tournament field, and how those legitimate bubble teams compared against other legitimate bubble teams. It's dangerous bringing human judgment into the picture by letting someone make that determination of who is on the bubble, but would it make any less sense than what we have now? Because what we have now doesn't make a whole lot of sense.