The Division I men’s ice hockey committee made some changes to the selection process this year, both in the criteria used to compare teams, and the methodology used in selecting and placing them.
Using a strict interpretation of its rules, and not budging from them, the committee selected and seeded the field in a mechanical 1-16 fashion. It altered from this only to protect a pair of sacred cows: a host school must be placed in that region; and intraconference first-round matchups are to be avoided like the plague.
There is nothing wrong with this on the surface. In fact, I have personally recommended in the past that the committee just use straight Pairwise to pick the field in the tournament. The complete inflexibility, however, in the seeding process led to some injustices, and/or undesireable — and avoidable — consequences.
On another level, we have the ongoing issue of the tournament criteria itself. By tweaking, tweaking, tweaking every year, we are left with a patchwork system that’s riddled with holes. The road to hell is paved with good intentions, and every one of the tweaks had noble intent.
But we’ll tackle the first issue first; that of the inflexible seeding process.
(FYI … I shouldn’t have to say this, but I will anyway: The NCAA tournament is still great, the four regional setup is great, and we’re all going to have some great memories. But this article is just about fixing flaws and making things even better.)
It’s evident from the article linked above that the committee did everything perfectly. It created four bands of four teams, giving each band a seed, 1, 2, 3, 4. It did not move a team out of that “band,” and did everything possible to create perfect matchups of 1 vs. 16, 2 vs. 15, 3 vs. 14, etc… It only deviated from that to avoid upsetting the NCAA’s two sacred rules: Host schools must stay at their region; and avoid intraconference first-round matchups.
Further, when it was forced to deviate, it only deviated within the rigid bands. Moving a team from the 13th position to the 16th is OK, because both are No. 4 seeds, but not from the 13th to the 12th.
Never before has the committee been that mechanical and robotic about it. But, boy, when they said they were going by the numbers, they sure meant it.
In fact, they were too proper and perfect.
While I am completely in favor of a perfectly objective system, as most of the coaches are, there are still too many tricky situations created — mainly, but not only, because of the NCAA’s two sacred rules — that demand some common sense “shifting around” of things.
By creating so many rigid rules — the banding, the hosting, and the avoidance of intra-conference matchups — you paint yourself into so tight a corner, that you leave yourself no leeway whatsoever to avoid sticky situations, such as the plainly-evident Cornell one.
Cornell, the No. 1 team overall in the tournament, was forced to play the No. 14 team (No. 11 in the more-sophisticated KRACH), instead of the No. 16 team (No. 29 in PWR and No. 47 in KRACH). The quality gap between Mankato at 14 and Mercyhurst at 15, is wider than the gap between Mankato at 14 and Cornell at 1. Given that assertion, the need to give Cornell the ridiculously more favorable matchup that it earned, should have been a priority over the strict adherance to banding and avoiding intraconference matchups.
This would not be as big a problem if all the No. 4 seeds were reasonably close in ability, such as in men’s basketball. But the gap between the two WCHA No. 4 seeds, and the two autobid happy-to-be-there No. 4 seeds is enormous. And this situation is going to happen again.
To be fair, NCAA committee chair Ian McCaw said the Cornell situation was the “toughest issue.” But, it was nevertheless left unsolved.
If the committee had given itself any leeway at all, it could have found multiple ways to solve Cornell’s dilemma that, while perhaps slightly disfavoring another team, would have made up for the grave injustice given to Cornell.
One such way out arises if the committee used its old methodology for seeding teams; i.e. individually comparing teams, not using strict PWR (see sidebar for detail). However, this is just circumstance, and since I’ve argued in favor of using strict PWR in the past, I don’t want to be a hypocrite and argue in favor of the old methodology (though at least it would’ve solved this sticky dilemma).
But there was another way out.
If the committee had simply switched the seeds of St. Cloud and Ohio State, all would be solved. You make St. Cloud a No. 3 against Boston College, make Ohio State a No. 4 against Minnesota, move Mankato over to play New Hampshire, and Mercyhurst becomes a No. 4 against Cornell.
Obviously, however, the committee was more concerned with preserving these “bands” than it was in protecting Cornell. Their reasoning: they didn’t want to upset the integrity of the seeds.
A noble reasoning, perhaps, but think about this:
The committee was willing to flip Wayne State, the 29th-ranked team in PWR (39th in KRACH), with Mankato, the 14th PWR team (11th in KRACH), to avoid an intra-conference matchup — but it was not willing to flip the 13th team (St. Cloud-14th in KRACH) with the 11th (Ohio State-16th in KRACH) to protect Cornell? Just because they were trying to preserve the band?
It’s OK to switch 29 and 14, but not OK to switch 11 and 13? Just because 29 and 14 are in the same band, and 11 and 13 are not?
Does this make common sense?
What is worse, violating your banding principle and possibly slightly slighting OSU, or keeping it as is and severely negatively impacting Cornell?
This is not about Cornell, per se. This is about the philosophy that says these “banding” principles are so strict, and so limiting, that we will not sway from them one iota in order to avoid an injustice.
If this were a minor quibble, we move on. But this falls into the “major quibble” category.
This strict adherence to the banding reared its ugly head in other ways as well. The committee was unwilling to shift UNH and Cornell, and thus avoid the potential second-round matchup between UNH and BU, teams that just played in the Hockey East tournament final. In the past, the potential second-round matchup between North Dakota and Minnesota also would have been easily avoided, at very minor expense.
In the past, Maine would have stayed East as a reward for being the “better” No. 2 seed. Of course, the committee wanted to “protect” Cornell by matching it with the “worst” No. 2 seed, Boston College, but who would you rather play right now, Maine or BC?
Finally, it’s worth saying that, while the Pairwise is good — and KRACH is better — the numbers aren’t precise enough to corner yourself with such strict adherence to them. The difference between No. 8 and No. 6 in PWR is minimal, and not worth boxing yourself into such an inflexibility.
All of the above becomes even more noteworthy when you consider that many of the criteria used to create this strict 1-16 ranking are flawed. As a result, you have the consequence of boxing yourself into corners with flawed numbers. That’s almost worse than being subjective.
In fact, taking a look at KRACH, we see that Boston College is actually No. 5, and Mankato is No. 11. This gives Cornell the most difficult bracket — according to KRACH — of any other No. 1 seed.
This doesn’t even take into account a Last 16 criteria. It, or some derivative thereof that rewarded better play late in the season or in conference tournaments, could have drastically altered the field and seeds. It was taken out of the selection criteria this year, because of the flawed nature of a Last 16 criteria based solely on winning percentage.
Obviously, teams in much weaker conferences have better winning percentages down the stretch, which skewers comparisons. That’s why that criteria was taken out. However, some sort of “Down the Stretch” criteria is philosophically sound, and can be utilized so long as it’s KRACH-ified; i.e. normalized to account for strength of schedule.
Without actually crunching the numbers, this most certainly would have kept St. Cloud State out of the tournament, made Maine a worse seed, and ultimately led to more “perfect” and usable numbers.
Hockey fans and coaches don’t necessarily want the committee using subjectivity, so by reinserting a strong Last 16 (or something similar) criteria, you are getting common sense results in a completely objective fashion.
The same is true for the Teams Under Consideration criteria, which now counts any team with an RPI of .500 or better. When you factor in that RPI is flawed in and of itself, then basing TUCs on RPI is flawed. At the very least, the committee needs to consider a TUC to be a team with an RPI over .500 AND a winning percentage over .500.
Better yet, scrap RPI and the constant tinkering of its weights, and use KRACH. And then use KRACH to determine TUCs.
And most of all, let’s just stop monkeying around with all of these criteria, sit down with people who understand the ramifications of everything that’s done, and hammer out a process that works as best as humanly possible.
The ultimate point here is, if you are going to be so strict to the numbers, and adhere to them so closely, then you better darn well make sure you are using GOOD numbers! As it stands, the committee is not only painting itself into a corner with overly-rigid principles, it’s doing it with flawed numbers.
An open proposal for a new system is forthcoming.