Each week during the season, we look at the big events and big games around Division I men’s college hockey in Tuesday Morning Quarterback.
Jim: Dan, we’re at the halfway point in the season and what a first half it was.
Fans back in the stands was the No. 1 high point for me and likely almost everyone in the nation.
On the ice, I’m impressed by a few teams like Quinnipiac, Cornell and Omaha. All three have winning percentages in the top five, but guess who (or should I say what) isn’t as impressed? The PairWise. Quinnipiac is fifth, Omaha is 15th and Cornell, despite their nation’s best 9-1-1 record, is 19th.
What, in your opinion, is wrong here? (And yes, I have my own opinion, but I figured I’d give you a whack first).
Dan: Asking me my opinion about something that’s wrong with college hockey is a lot like dancing in and out of a bear trap, but I’ll try to stay calm and tame on this despite the fact that my eight-month-old daughter is experiencing a sleep regression (shoutout to all you parents out there who know exactly what I’m talking about).
The Pairwise Rankings are solely about the mathematical numbers and quality of opponents based on how they perform opposite other teams, and the biggest, major difference between those three teams are the results against certain opponents.
Cornell, for example, has five wins against teams ranked 48th or lower, and both Dartmouth and Yale are in the bottom five teams in the Pairwise, with Yale slotting in the 58th and final spot. That means the wins over Brown, Alaska and Princeton – the other three teams – don’t really rate as much as Quinnipiac’s 5-2 win over North Dakota or the tie and win over both Boston College and Northeastern. The “tie” against Clarkson from this past weekend helps, as does the win over Boston University, but the truth is that the team’s non-conference schedule, so far, doesn’t have enough weight to pull it higher than where it is right now, especially when compared to what Quinnipiac did against teams like NoDak, which is No. 3, and Northeastern, which is No. 13. Those wins weight more and, understandably, better.
Remember that every single conference has to go .500 because every single league game ends in both a win and a loss or a tie. The only way to move up the Pairwise is then to win nonconference games. But the nonleague schedule has to be good enough across the board, and that relies on two different factors.
One is the individual schedule and playing teams that are simply better than another team; that’s where Quinnipiac gets more credit in the numbers right now. The other is to have the entire league play well enough to push all of its parts to the top of the polls. That’s something I ran into with Atlantic Hockey over the better part of the last 10 years, and that was one of the reasons the reorganization of the WCHA into the CCHA helped the cumulative league, even as I hate to admit it.
In terms of Omaha, the Mavs benefit from a very, very good league, and though I readily admit my love for that team, this weekend’s split with Western Michigan is a perfect example of why they sit 15th. A sweep would’ve probably pushed them into the top 10 while simultaneously pulling the Broncos further down the list. Instead, the one win split kept Omaha afloat and didn’t pull Western down the list.
Because the Pairwise is so mathematically-based, it’s both easy and impossible to manipulate. Look at what Arizona State did two years ago and what Niagara did in 2013, and the pattern is simple – win the right games and let those opponents carry you up the rankings while you simultaneously don’t mess up by losing to teams so far behind you.
For what it’s worth, that should illustrate just how thin Cornell in particular is skating. A loss to one of those teams down the bottom of the PairWise would be catastrophic, and with half of ECAC sitting in that low percentile, it’s more of a warning sign than I think voters, myself included, give it credit for. Omaha is a little more forgiving, largely because of it will have more opportunities to play teams that can carry it through to the tournament, and Quinnipiac, by virtue of its wins, gain a couple of do-overs if it loses one of those games next month against Princeton, Brown or Yale.
Jim: Well, first, I’ll be careful to ask about the PairWise when you are I are having a drink one night as I’m afraid you’ll have a bear trap sitting behind my barstool.
But I also have the answer to the question of “How do you get Dan Rubin to answer a question in 500 words or more?”
Believe it or not, your diatribe might be missing one element of Cornell’s results to date, and that’s the fact that three of the Big Red’s wins occurred in overtime. Beating Alaska twice and Brown once in the 3-on-3 extra session looks great in the overall 9-1-1 mark, but in the long-term three of those nine wins need an asterisks. The overtime wins, as we have mentioned in this space before, only credit a team with 55 percent of a victory in the PairWise. So 55 percent of three wins against Alaska (twice) and Brown, two teams in the bottom 13 nationally of the RPI won’t be beneficial to the overall RPI of Cornell.
Which, I guess, brings us to a major issue in college hockey right now. How should we credit 3-on-3 overtime wins. Coaches were so reluctant to give credit for winning a game when it wasn’t being played 5-on-5, that the entire body in men’s hockey decided that for the purpose of the RPI (and thus the PairWise) that OT wins would be credited at 55 percent for a win and 45 percent of for a loss. From the day I first heard this, it never made sense to me.
In league standings, we’re already using a two-thirds-to-one-thirds scoring system for OT wins (though shootout wins are credited the same). So telling a team that your national ranking in the RPI and PairWise won’t even get close to what the win typically is worth in your standings seems like a disconnect to me.
I’m still in favor of using the NHL system. You go back to a two-point system and give two points for any win and one point if the loss comes in OT or a shootout. People got fired up about using this system because of the NCAA tournament. But if you’re telling me you can credit a team for 55 percent of a win currently, who can’t you credit a team 100 percent of a win and 50 percent of the loss in the RPI?
Is this more math than the average fan wants to do? Yes. But right now, Cornell is becoming a case study for why the current system need to be changed. The team with the best winning percentage halfway through a season wouldn’t make the national tournament without an autobid.
Something seems off, no?
Dan: I’m seeing myself develop a track record that’s built on anger and sadism, and a lot of our readers and folks in college hockey are going to assume that I’m this nasty, angry, little man. Outside of my Napoleon complex, I promise I’m generally a jovial guy.
Though for what it’s worth, I now have the idea to set up a bear trap behind your stool, and I’m not backing away from it.
To the system, though. I completely forgot about the percentages for the new system, and I agree that it needs tinkering because, like you said, the RPI and PWR won’t get close to what the wins mean in the standings. But instead of more math, I’ll go the other way and advocate for less math – and not just because I almost failed statistics during my undergraduate years, which, if anyone asks, is also 18 years ago.
The impact to the NCAA tournament is huge because of hockey’s mathematical selection, and I readily admit that I wanted the NCAA to envision and redraw its map through a selection committee. I think the numbers are worthwhile, but I really wanted to incorporate an eye test into the conversation more than what we see. A clean break felt right, but instead we’re left with a system that, regardless of how we slice it, needs adjustment.
Reinventing hockey seems to be a conversation I’ve had more and more lately, and it was the subject I wrote a week or two ago. That was more about adapting to the way we’ve become more remote as a society and how the digital enhancements that were adopted as part of college hockey’s remote year last year are now more commonplace. From a content perspective, I feel like last year, with all of its warts, rocketed us into a more digital world, and some of those effects are more permanent.
From a pure hockey standpoint, nothing beats the return to the rinks, but were there other aspects of last year that should have been kept, either from a macro or micro level?
Jim: Well, I have to disagree that anywhere near resembling an “eye test” for the selection process would be a good thing.
I detested last season, trying to guess how six individuals, all of whom I know and respect, would select 16 teams to play for a national championship. I like the fact that every year all of the Division I men’s coaches get together at a convention, determine by a majority how to select their tournament, and then use that mathematical formula to choose the at-large teams in the field.
Someone, believe it not, it feels like wanting to use a mathematical approach makes me a curmudgeon. Could that be correct?
On the other side of your question – what happened last year and what we might take from it to continue using – I have to admit, there isn’t a lot. I detested every bit of the difficulties associated with last season. I guess I appreciated how some leagues went to a single elimination tournament, something Hockey East did. I loved how some digital content was easier to obtain (again, looking at Hockey East). And I’d love those things to continue. But I can’t think of a ton of other things from last year that I loved.
I’m guessing you asking that question means you have a few that I’m missing. Go ahead and school me!
Dan: So let me redefine what I was trying to say on that question. I have thought long and hard about last season, and under no circumstances do I want to go back to that world. I do, however, think there are some ideas that can turn into net positives if they make sense for the future.
I’ll start with a couple of obvious ones. First, Atlantic Hockey switched to a pod-based, divisional format that I loved for standings purposes. Air Force’s geography obviously makes this difficult for a full-time solution, as does an odd number of teams, but I liked the idea of the divisions playing one another prior to the championship weekend. It was an updated format from the disastrous 2011 postseason that used single-elimination games before the quarterfinals reseeded, but the prospect of having the teams play a divisional format with a full-time, league-wide schedule intrigues me. That’s not to say it would work, but from an experimental standpoint, I would dabble for a year or two.
Second was the return to campus sites for conference championships. I don’t think this makes a ton of sense for a couple of leagues, namely Hockey East, but I was sternly behind the WCHA’s move to campus sites less than five years ago. I thought it made sense for several leagues, and I loved the concept of allowing the best remaining seed or the league champion to automatically clinch a home game for the championship.
I have no problem admitting I was a little disappointed to see a mass return to neutral sites. Again, I don’t know about long term viability, but I would’ve liked to have seen it evolve into a new line of thinking. From a Hockey East perspective, what I’m saying makes no sense because the league is so centrally located, but it’s interesting to think about.
The third one is a little bit more obscure. Paula and I discussed the midseason tournaments and how the formats just don’t offer the same pizzazz as years past, and the Great Lakes Invitational, in particular, was a sore subject for its format that ensures Michigan and Michigan State won’t play like they did when they met in Detroit at the old Joe Louis Arena.
I thought about it this past week, and I realized that my problem with the postseason tournament could potentially resolve itself with the incorporation of a new-style bubble like what the NCHC pulled off in Omaha last year.
That doesn’t mean the lack of fans, but what I’m proposing is a weekend series of games that follows a little bit of what we normally see on the junior hockey circuit. Every league team would then play a series of matchups at that team’s arena as a sort of way forward to incorporate a kickoff weekend or multi-team event. I’m not saying to propose certain rivalries play at the event, but I’d be interested in the success of an early season event at a campus site for the entire league as a kickoff.
The most obvious hurdles center around revenue sharing for teams that lose a home game, and I have no idea how leagues could pull it off. It would just be interesting, to me, to watch teams meet and battle over a full weekend with 4-5 games over three or four days.
Maybe I’m crazy for thinking any of this stuff could work, but hey, outside the box isn’t the worst thing in the world.
