The Register Citizen (Torrington, CT)
NCAA hoops should learn from hockey selection process
I’m a Baseball Hall of Fame voter. I am aware everyone knows more than me.
Does everyone also know more than a computer and the mathematical models and algorithms used to rank college athletic teams? If not, if there really are formulas that can work, why do we need people in back rooms determining who gets to play in national collegiate championships?
That question has been on my mind since the NCAA Men’s Ice Hockey Committee sent a memo this week to coaches saying that while it still will use some elements of the PairWise Rankings this season, vast schedule changes caused by COVID-19 have necessitated the use of a human component in the 2021 tournament selection.
It took a pandemic for hockey to be like football and basketball.
The PairWise algorithm, used for more than two decades to set the 16-team NCAA field, still can be used within conferences. Yet with so few inter-conference games, regional advisory committees made up of mostly coaches, according to College Hockey News, will “assist in the observation and evaluation of teams and provide recommendations to the national committee.”
Sports fans who don’t follow college hockey often are surprised that no one goes behind closed doors like they’re picking the pope and wait until a puff of white smoke reveals that Notre Dame and some Power Five conference school got in and some mid-major got jobbed …
Sorry, got a little cynical there.
Basketball had RPI and now has NetRankings to do much of the heavy lifting for the committee. Yet Selection Sunday annually turns into a food fight over the No. 1 seeds and the last four teams left out of the field. Football once used computers to help in the BCS selection. While advanced analytics are studied, they play no formal role with the College Football Playoff committee’s decision making. That’s pretty obvious as the cartel dutifully blocks out non-Power 5 schools.
Hockey did the right thing going human this one time.
“You can’t use the PairWise this year,” Quinnipiac coach Rand Pecknold said. “It’s completely ineffective when teams play an insular schedule.”
Quinnipiac, 13-5-2 and tied with Robert Morris for 10th in PairWise, has gotten in eight nonconference games. With Brown, Cornell, Dartmouth, Harvard, Princeton, Yale, RPI and Union not playing, however, the Bobcats are left with six ECAC games each against St. Lawrence, Clarkson and Colgate.
The Hockey East is playing entirely within the conference and has developed a week-by-week grab bag of scheduling. The NCHC also has played entirely within its conference and, except for a few out-of-conference games, the Big Ten, too.
“The PairWise almost has to be thrown out the window this year,” said UConn coach Mike Cavanaugh, whose 8-8-2 team is 23rd in PairWise. “Without out-of-conference play, it’s really a useless statistic. So it’ll be like the ’80s where guys got in a room, old-school, and decide who goes and who doesn’t.
“For a long time it has been cut and dried. No one making arbitrary decisions, and unfortunately this year that’s going to be the case. I think it’s going to be really important for the league commissioners to advocate for their teams.”
Are you one of the coaches on a regional committee?
“No,” Cavanaugh said. “Thank God.”
Hockey developed an objective formula through the 1990s, has gone through revisions, and no one should mistake it as perfection. There’s another system, KRACH, that some college hockey aficionados insist is better. UConn is 19th and Quinnipiac is 15th in KRACH. Pecknold, who has a vote in one of the weekly polls, believes his team is seventh or eighth in the nation right now, “definitely top 10.” He also pointed out Cornell and Harvard of the ECAC would have been stacked if they played and could have pushed back QU.
After the six automatic conference bids, PairWise is used to pick the final 10 spots. RPI, record vs. common opponents and head-tohead are essentially the three criteria. Since 2013-14, quality win bonuses and home-road weighting have been added.
My point is it’s left to objective criteria and people know the score throughout the season.
“I like the PairWise,” Pecknold said. “I think it is cleaner. I think it’s easier for the fans to see what is going on. I think it’s pretty effective.”
“It’s transparent,” Cavanaugh said. “You know where you stand. You know what you have to do. I think it’s pretty fair. And for the most part I think PairWise is an accurate representation of who should be going.”
Pecknold points out that even within leagues, teams are playing unbalanced schedules this season. AIC (PairWise sixth) and Robert Morris, Atlantic Hockey leaders in what is considered the weakest league, aren’t playing each other. Quinnipiac swept AIC this season.
Of its 12 nonconference games, Quinnipiac usually plays 8-10 against Hockey East. Pecknold points out his team has won 60-70 percent of those games in the past five years and it has helped with seedings.
“Maybe some Hockey East teams will say Quinnipiac hasn’t played as tough a schedule,” Pecknold said. “I respect their decision to play within conference (because of COVID). I just don’t think you should punish us because we didn’t have the opportunity to play them. There are already teams lobbying a little bit with people on the committee.”
Let’s take a look at women’s basketball and the first official “16 reveal” by the NCAA committee this week. Texas A&M was 13th in the NET yet fifth (a No. 2 seed) with the committee. Baylor is fourth in the NET yet 10th with the committee. I’m wondering if Baylor is a 1 seed and it’s officially a 3 right now. That can change in the coming month, but there are significant gaps. Who’s wrong? The people who fed the NET criteria into the computer? Or the people watching the games and judging?
In years when UConn dominated, the seedings didn’t matter. Yet if this first committee ranking holds up in Region 1, the No. 1-seeded Huskies could play Tennessee in the Sweet 16, Arizona/Baylor in the Elite Eight and two among South Carolina, Stanford and Louisville in the Final Four. So seedings matter plenty this year.
The UConn men are on the bubble at the moment. With James Bouknight back, the Huskies can play their way into the tournament starting Saturday at Villanova. And with the COVID discrepancies, certainly human minds are needed in 2021 to mold the tournament.
“I think we’re too reliant on technology nowadays to begin with,” UConn coach Dan Hurley said. “I’ll roll with human error. So much of the metrics have some type of major flaw with it. The NET ratings this year, which overvalues road games and hurts teams that didn’t get the volume of nonconference games ...”
No argument on this year. But the future?
“All the different metrics have flaws,” Hurley said.
“I’m going to go with human beings. Hopefully, with deep backgrounds in basketball.”
As Pecknold did remind us, the first year Quinnipiac went Division I, it was ninth in the PairWise. All the Bobcats’ games played were Metro-Atlantic.
“We actually had to get voted out of the tournament and I don’t think it has happened in all the years since,” Pecknold said. “It was the right decision. We hadn’t played one game outside the MAAC. No one would play us, which is a whole different story.”
It took time to knock the kinks out, and NCAA hoops is far advanced beyond 2000 hockey.
I submit serious consideration to using the NET for the at-large bids and/or seeding the top 16 basketball teams within the next three to five years. Continually tinker with it to find the best criteria. In the end it will be the fairest way. Certainly, it’s the most objective.
Of course, it would interfere with a cartel-rigged College Football Playoff format that should be at least eight teams and more inclusive.
And it would interfere with the industry of bracketology leading to Selection Sunday.
Every person who steps into a committee room has a bias — if not for a conference, for a certain piece of analysis — and then argues it in the heat of the moment. The algorithm will be cool and consistent every time.
Maybe the powers of college sports don’t want the most objective system. Maybe deep down, the most powerful want to retain the most power. And deep down, maybe the fans want to be entertained all winter by Joe Lunardi and retain the privilege of crying that their school got robbed any given Selection Sunday.
Me? I prefer objectivity. Or maybe I’ve come to trust machines more than man.