When the NCAA announced the creation of a new metric to gauge college basketball teams, it was billed as an all-encompassing formula that would displace traditional metrics like RPI and effectively measure teams throughout the season.
“What has been developed is a contemporary method of looking at teams analytically, using results-based and predictive metrics that will assist the Men’s Basketball Committee as it reviews games throughout the season,” said Senior Vice President of Basketball for the NCAA Dan Gavitt to NCAA.com. “While no perfect rankings exist, using the results of past tournaments will help ensure that the rankings are built on an objective source of truth.”
Gavitt added the following in his statement to NCAA.com: “The NCAA Men’s Basketball Committee has had helpful metrics it has used over the years, and will continue to use the team sheets, but those will now be sorted by the NCAA Evaluation Tool,” Gavitt said. “As has always been the case, the committee won’t solely focus on metrics to select at-large teams and seed the field. There will always be a subjective element to the tournament selection process, too.”
That last line is key.
If the tournament process was going to be subjective anyway, why create a new metric? Was RPI not serviceable in discerning which teams were worthy of a tournament berth before the committee made the final decision? With this sentence, the NCAA essentially admitted it wasn’t going to use the NET rankings in the selection process.
Luckily, we still have the rankings available to create a tournament bracket based off the NET. This bracket will help us identify some positives with the new metric, highlight some of its flaws and give certain schools ammunition about getting snubbed on Selection Sunday.
Here’s what the NET bracket looks like with the teams the committee put in the field.
To create this bracket, I took the teams the committee put in the field and sorted them from top to bottom based on NET. I then assigned each team a number 1-68 to represent its overall seed in the field. Based on how the committee actually placed teams in this year’s tournament, I placed the new corresponding teams in the same spots. Teams 1-4 received a No. 1 seed, teams 5-8 received a No. 2 seed, and so on. After creating the NET bracket from the teams in the 2019 field, here are some observations.
THE NET DID WELL ON A PER-SEED BASIS
If we take the NET to be an all-encompassing metric and assign seeds to each team based solely on ranking, the NCAA committee wasn’t far off. To do this, I compared each team’s seed according to the NET ranking versus the seed it received from the committee. For example, Virginia Tech was a No. 4 seed in the eyes of the committee. With a NET ranking of 11, however, the Hokies would be assigned a No. 3 seed. This results in a seed difference of 1.
After sorting the tournament teams from top to bottom and assigning each team a seed 1-16 based on NET, I calculated the seed difference for each team based on the number it received from the committee. When averaging the seed difference across all 68 teams, the average seed difference comes out to 0.735 either way. This means the committee sorted the field it selected very similarly to how it would’ve been sorted by NET. However, that doesn’t mean it got seed lines correct.
STRONG NET MID-MAJORS GOT HOSED ON SEED LINES
Gonzaga was a No. 1 seed in either case, but the Bulldogs got the fourth overall seed from the committee instead of the second they would’ve received based on NET. In what I personally think is the most egregious placement, a 31-3 Houston squad got knocked down two seed lines from a No. 1 seed to a No. 3 seed. The committee knocked Wofford, a No. 4 seed based on its NET ranking, to the 7-line. Buffalo, a 31-3 team with a road win against then No. 13 West Virginia and a win against Syracuse, found itself on the 6-line despite NET saying it was a No. 4 seed.
Even the lower mid-majors got screwed. New Mexico State, a No. 10 seed in the eyes of NET, found itself on the 12-line. Baylor, one spot ahead of the Aggies in the NET ranking, received the No. 9 seed NET said it should get. Saint Mary’s, which needed a conference tournament run to secure a bid despite a high NET, got pushed down three seed lines from the No. 8 NET said it should get to a No.11. Murray State, a No. 10 seed based on NET, is on the 12-line.
Mid-majors Nevada and VCU got fair treatment based on NET, moving one seed line each in opposite directions. Turns out, the NET actually boosted mid-majors, something the committee wouldn’t have done and didn’t do.
POWER-5 SCHOOLS GOT A BOOST FROM THE COMMITTEE
On the flip side, the Power-5 schools got a bump from the committee respective to their NET ranking. LSU and Michigan each moved up a single seed line in the actual bracket. Kansas State, a No. 6 seed based on NET, was put on the four-line. Marquette also got bumped up two seed lines from No. 7 in NET to No. 5 in the eyes of the committee.
And then there’s Syracuse.
The Orange were not exactly on the bubble this year, but they were a No. 10 seed according to NET. The committee put Syracuse on the 8-line, a two-spot jump. I’m not sure what dirt Jim Boeheim has on the committee, but it’s worked wonders in recent years. Instead of facing a No. 7 seed, the Orange now square off with a No. 9 seed.
Minnesota, with a NET ranking of 61, should’ve been on the 12-line. Instead, the Golden Gophers got up to the 10-line. The team got a huge win over Purdue in the Big Ten conference tournament, but that shouldn’t be the difference between playing a No. 5 seed versus a No. 7 seed. The NET pushed back on subpar Power-5 schools, but the committee propped them up anyway.
WHAT PENN STATE TEAM DID NET TRACK ALL YEAR?
The Nittany Lions finished 5oth in the NET, just missing out on an at-large berth if we went strictly off the ranking to fill out the field. Penn State was 14-18 on the season and still placed ahead of 30-win New Mexico State, 30-win UC Irvine, 28-win Liberty and 28-win UNC Greensboro in NET. With win percentage being one of the NET’s factors, how does a team with a sub-.500 record beat out 30-win squads? Those 14 victories must’ve been something special.
THE NET DOESN’T ACCOUNT FOR INJURIES
I’m talking specifically about Zion Williamson in this case. Duke was awarded the No. 1 overall seed despite being third in NET. This ranking still puts the Blue Devils on the top seed line, but it might not give them as favorable region as the top overall spot would. Virginia lost to Duke twice with Williamson healthy, so the Cavaliers got dropped from the top overall seed based on NET to second overall. North Carolina, which was a No.2 seed based on NET, beat the Blue Devils twice without Williamson before being edged out by Duke in the conference tournament with the star forward back. The Tar Heels got the third overall seed for their wins, but the committee factored in Williamson’s return more than NET.
CONFERENCE TOURNAMENTS DON’T MATTER OUTSIDE OF AUTO-BID OR BUST TEAMS
Unless a team is looking for the automatic berth, the conference tournament is essentially meaningless on the seed line for both the NET and the committee. Auburn, the SEC conference tournament winner, was a No. 5 seed in both scenarios. Michigan State, the Big Ten conference tournament winner, would’ve been a No. 2 seed either way. How about Iowa State from the Big 12? The Cyclones were on the 6-line based on NET and the committee.
Duke was a No. 1 seed in both scenarios and Oregon won the automatic bid from the Pac-12. The Ducks would’ve just missed out as an at-large team in a bracket solely on NET ranking. Washington, the loser in the Pac-12 conference tournament final, actually jumped up to the 9-line from the 11-line it had in NET.
NC STATE, CLEMSON, AND OTHER HIGH NET TEAMS HAVE A LEGITIMATE GRIPE
It’s one thing to put St. John’s, a team ranked 73rd in NET, in the field as the worst at-large. It’s another to do it with NC State (33rd in NET), Clemson (35th in NET) and Texas (38th in NET) not being included in the tournament as well. A 40-spot difference in NET is too much to ignore for what is supposed to be an optimized, all-encompassing number.
If Power-5 schools can’t get help from the NET, what chance do Furman (41st in NET) and Lipscomb (49th in NET) have of getting at-large berths? This is the biggest evidence of either the NET not being accurate enough or not being used at all by the committee.
If NC State’s resume was so flawed, how did it get ranked so highly in the NCAA’s own algorithm? If the algorithm is correct, why didn’t the committee give the Wolfpack an at-large bid over St. John’s, among other teams? Of the 36 at-large teams, 21 had a higher NET than NC State. The Wolfpack weren’t going to be the best at-large team by a wide margin, but NET had them in the field.
If we took all the at-large teams based off NET, the following teams would be in the bracket: NC State, Clemson, Texas, Furman, Memphis and Nebraska. These teams would replace St. John’s, Arizona State, Minnesota, Seton Hall, Temple and Ohio State in the current field.
THE NET IS EITHER NOT OPTIMIZED OR NOT NECESSARY (OR BOTH)
Based on these observations, the NET ranking is not optimized. If NC State’s flawed resume gave it a 33 in NET and Penn State landed at 50 despite a sub-.500 record, the formula is not as accurate as intended. If that’s the case, the question now becomes whether the NET ranking is necessary. When it unveiled the NET ranking, the NCAA clearly intended for it to be a replacement for metrics like RPI. Gavitt said the NET wouldn’t be the sole metric used, but the at-large field saw zero impact from the evaluation tool claimed to solve the committee’s problem ranking teams.
The NCAA committee is naturally subjective and will continue to be, as Gavitt mentioned in his statement. A subjective process doesn’t need any metrics, especially one that seems far less all-encompassing than the NCAA would want us to believe.
The committee wants its hands on the bracket, regardless of how computers and models rank teams. In its current iteration, the NET is as flawed as previously models. It does give some mid-major teams their fair due and pushes back against subpar Power-5 teams, but the committee clearly didn’t use that information when filling out the field.
The solution here is simple: either fix the NET to use it or get rid of it entirely. If the committee doesn’t need the tool, we don’t need it either.