Bracket Watch: Will some peculiar NET ratings cause seed madness?

The NCAA moved to the NET ratings system before last season, and for the most part, it worked swimmingly in Year 1. There weren’t many controversial choices for inclusion into the field of 68 last March, and 14 of the top 16 teams in the NET on Selection Sunday reached the Sweet 16.

So the NCAA ran the NET back this year without any changes to its secret formula. And while it remains a useful tool that’s infinitely better than the old RPI, let’s just say that the honeymoon period for NET might be over.

Year 2 has exposed some glaring quirks in the system. Start at the top, with teams that continue to be overvalued. Duke, despite losing three of its past five games — two of those to unranked teams — has barely budged out of its top-six spot. But the love for the Blue Devils is far less egregious than the ongoing respect given to Arizona. The Wildcats ranked 10th in the NET as of Thursday afternoon. The number 10 is also how many losses they had going into that night’s game against Washington State. They also entered the night tied for sixth in the Pac-12, with an uninspiring 3-7 Quad 1 record.

Meanwhile, the top team in the Pac-12 standings going into Thursday was UCLA. The Bruins have been red hot yet have not really been able to move up in the NET. They ranked 76th on Thursday afternoon, one spot ahead of 15-14 Davidson and seven spots behind the Pac-12’s last-place team, Washington. UCLA’s low ranking makes some sense, given how bad Mick Cronin’s team played for much of the year before turning things around. But you would hope the system would better be able to respond to a team racking up six Quad 1 wins versus a Washington team whose Quad 1 win on Thursday at Arizona State was its first since Baylor on Nov 8.

Here’s another example of a NET oddity: Kentucky (24-6) ranks 22nd, behind Texas Tech (18-12) and West Virginia (20-10), though neither of those Big 12 teams has done much for weeks. And Kentucky won at Texas Tech, not to mention its wins over Michigan State and Louisville.

The reason for these apparent glitches is pretty obvious. Unlike the RPI, which was simply a results-based stat, the NET incorporates predictive measurements such as offensive and defensive efficiency. A team such as Kentucky is just not very efficient on either end (33rd offensively and 49th defensively, per KenPom.com), while Duke and to a lesser extent Arizona are. There’s value in those numbers — who doesn’t love KenPom? — but results probably should matter a little more when determining who deserves an NCAA Tournament bid.

Look at this way. Jeff Sagarin’s ratings system, which is one metric on the committee’s team sheets, weighs results more heavily. Some of his rankings on Thursday also seemed much, much saner:

No. 11 Kentucky
No. 12 Duke
No. 36 Texas Tech
No. 37 Arizona
No. 39 West Virginia
No. 61 UCLA
No. 81 Davidson
No. 99 Washington

The NCAA, luckily, is not oblivious to these issues. Dan Gavitt, the NCAA’s senior vice president in charge of men’s basketball, said on Jeff Goodman’s Good N’ Plenty podcast this week that the selection committee wanted two years of data on the NET before making any changes. Gavitt added that NCAA officials have identified a few areas that could use tweaks and that they may make some adjustments to the NET formula this spring.

The NCAA stresses that the NET is simply a sorting tool and that the selection committee has several other metrics – including Sagarin and ESPN’s BPI and yes, KenPom — to consider. The committee proved last year that it wasn’t glued to NET numbers: Temple (56th in NET), Arizona State (63) and St. John’s (73) received at-large bids, and UNC-Greensboro (60) was only knocked out by Oregon winning the Pac-12 tournament. Meanwhile, two teams with top 35 NET rankings, N.C. State (33) and Clemson (35), were left out.

Yet the NET also creates the quad system that is so important in judging teams. The challenge this year for the committee will be determining what’s real and what’s not in those NET numbers and figuring out how to seed teams such as Providence, UCLA and Kentucky, which have big wins but also damaging Quad 4 losses. But those don’t have to be the end of the world either. The only two teams that got at-large invitations last year with a Quad 4 loss were Baylor and Arizona State; they each had two black marks in that quadrant. Baylor was a No. 9 seed, while Arizona State went to the First Four.

Bottom line: The NET remains a significant step forward as the committee’s main sorting instrument. But it could use a few mechanical repairs in the offseason. Especially if Arizona gets a top-four seed.

Some other notes on this week’s bracket:

• Programming note: Beginning Monday, Bracket Watch will be coming at you on a daily basis all the way through Selection Sunday. It’s go time.

• The 2-seed line was the biggest challenge this week. Dayton and Florida State remain there, but Seton Hall lost at home to Villanova on Wednesday and Maryland has been fading a bit.

We were strongly considering moving Kentucky up to the 2-seed line after the Wildcats’ win over Auburn last Saturday. Then UK lost at home to Tennessee, and Auburn lost at home to Texas A&M. So that’s a no-go.

In the end, we kept Seton Hall there because the Pirates’ 10 Quad 1 wins are third-best in the country behind Kansas and Baylor. And we welcomed Michigan State to the floor just below the penthouse. The Spartans have nine losses, but they’ve beaten Iowa, Maryland and Penn State — the latter two on the road — in their past three games and have only one loss all season outside of Quad 1. Michigan State also ranks third in Sagarin, fourth in BPI and seventh in KenPom to go along with its No. 7 NET ranking. That looks like a 2-seed profile to us.

• Speaking of the 2-seeds, Dayton is making a strong push to overtake San Diego State as the fourth No. 1 seed. The Flyers won convincingly at Rhode Island on Wednesday for their fifth Quad 1 win, one more than the Aztecs, and their offensive efficiency numbers are historically good. Putting Dayton as the No. 1 seed in the East and San Diego State as the No. 2 in the West fixes some geographic concerns as well. We’re not quite ready to go there yet, but it’s getting closer.

• Last week, we wrote that Kansas would still go to Indianapolis (Midwest) and Baylor to Houston (South) despite flipping positions on the No. 1 line. But as the overall No. 1 seed, the Jayhawks would get to pick which region they would want to headline. We missed Bill Self’s comments that hinted of a preference for Houston, which makes sense on a couple of levels. So we’re going to go with that change now, until we (hopefully) hear more from Self.

• Texas, which has won five straight including three against teams in the field, is in the First Four this week. The Longhorns have avoided bad losses all season and are a decent 5-7 in Quad 1. Purdue also creeps back in at 16-14. Tuesday’s win at Iowa was huge and goes nicely with earlier victories over Michigan State, Wisconsin, a season sweep of Indiana and a home win over those same Hawkeyes. Also, Texas won at Purdue on Nov. 9 in a game that seemed big at the time, then almost immediately dwindled from consciousness and now is relevant again. Gotta love college basketball.

• The latest episode of Bracket Madness, The Athletic’s podcast dedicated to all things bracketology, will be available on Friday morning. Be sure to also check out Tuesday’s episode with Mr. Bubble Watch himself, Eamonn Brennan, as we discussed how UCLA and Texas have added plenty of insanity to this bubble. Find both episodes wherever you get your podcasts.

• This week’s bracket pits the South vs. the East and the Midwest vs. the West in the national semifinals, based on the order of my No. 1 seeds. An asterisk denotes a projected conference automatic bid qualifier.

South Region (Houston)

East Region (New York)

Midwest Region (Indianapolis)

West Region (Los Angeles)

Multi-bid conferences

(Top photo of Duke’s Wendell Moore and N.C. State’s Braxton Beverly: Grant Halverson / Getty Images)