College basketball followers love ragging on the NET rankings. Die-hard fans often believe their favorite team is undervalued by the NET (short for NCAA Evaluation Tool), and the daily ranking fluctuations leave fans questioning the logic behind the NCAA’s system.
Pitt drops 3 spots in the holy NET Rankings after an 8-point home win, putting them just behind 14-13 Florida and 15-13 Oregon. The NCAA has this metric figured out! pic.twitter.com/7jAlYFzERR
— Kyle Saxon (@DixonDisciple) February 22, 2023
The NCAA includes a detailed explanation of what the NET tries to measure, but fans nonetheless argue that their team should never tumble down any ranking system after a victory.
My extremely nerdy and mildly spicy take? The NET rankings are not broken. In fact, I actually like them. Long live the NET!
How’s the NET calculated, and what’s its point?
The exact NET calculation is unknown, but the NCAA shares that the NET has two key components, which are the Team Value Index (TVI) and net efficiency.
TVI includes game results, game location, and outcome. Net efficiency (offensive efficiency minus defensive efficiency) is adjusted to account for strength of an opponent and game location.
The NET tries to incorporate wins and losses, strength of schedule, and efficiency data to spit out a metric that rates teams based on both results and predictive metrics.
A team’s specific NET ranking isn’t what the NCAA Tournament Selection Committee uses to form its bracket. Instead, the purpose of the NET ranking system is to sort teams into quadrants to create a resume comparison tool for the committee.
“I think having a computer do that is better than human rankings because I think computers capture noise better than humans,” said Jim Root, a college basketball expert and founding member of Three Man Weave. “We want to know the value of wins and losses. Your own NET ranking is less important.”
“The NET must be broken, how can a team like Saint Mary’s with only two Quad 1 wins be 7th in the NET”
I don’t know how many times I’ve felt the urge to “kindly remind” these people that the quadrants don’t decide the NET, the NET decides the quadrants
— Bauertology (@Bauertology) February 22, 2023
The more Quad 1 wins — home wins against teams ranked 1-30 in the NET, neutral site wins against teams ranked 1-50 in the NET, and road wins against teams ranked 1-75 in the NET — the better. The more Quad 4 losses — home losses against teams ranked 161-363, neutral site losses against teams ranked 201-363, and road losses against teams ranked 241-363 — the worse. There are also Quad 2 and Quad 3 games, with Quad 2 wins and losses being viewed more favorably than Quad 3 and 4 victories.
You get the idea. And again, a team’s exact NET ranking is not how the committee determines NCAA Tournament selection or seeding.
Last one I think is important—the media probably puts more meaning into the NET than the committee does. NET ranking matters, obviously. It’s the primary sorting tool from which we get quadrants. Its impact is undeniable. But Burnett said it’s not the judge and jury of selection. pic.twitter.com/8MuK8tIgVd
— Matt Norlander (@MattNorlander) February 18, 2022
“Computer models cannot accurately evaluate qualitative factors such as games missed by key players or coaches, travel difficulties, and other effects of specific games,” the NCAA website says.
See! The NCAA even admits that computer models aren’t everything.
“Each committee member independently evaluates a vast amount of information during the process to make individual decisions,” the website continues. “It is these qualitative, quantitative, and subjective opinions — developed after hours of personal observations, discussion with coaches, directors of athletics and commissioners, and review and comparison of various data — that each individual ultimately will determine their vote on all issues related to selections, seeding and bracketing.”
A team’s NET ranking doesn’t mean as much as fans believe. Rutgers, for example, made the NCAA Tournament last season with a NET ranking of 77.
Predictive metrics upset fans
The NET factors in a team’s adjusted efficiency, a critical measure used by predictive analytics sites like KenPom.com.
“The first thing you should know about this system is that it is designed to be purely predictive,” KenPom’s website says. “If you’re looking for a system that rates teams on how ‘good’ their season has been, you’ve come to the wrong place.”
KenPom tries to give an understanding of “how strong a team would be if it played tonight, independent of injuries or emotional factors.” ESPN’s Strength of Record metric offers a better idea of how solid a team’s season has been when solely viewing wins and losses.
Efficiency measures help add context to wins and losses, suggesting that not every outcome is created equal. Let’s take a look at Clemson’s men’s basketball team, for example. The Tigers are 2-0 against Florida State (9-21) this season.
In a road game against the Seminoles on Jan. 28, Clemson won 82-81. The Tigers dropped from 58th to 60th on KenPom by the time they played their next game. In a home game against the Seminoles on Feb. 15, Clemson won 94-54. The Tigers improved seven spots on KenPom by the time their next game rolled around.
On paper, Clemson won both games, but predictive metrics viewed the Feb. 15 win more favorably because it was a more complete team performance than the Jan. 28 nail-biter. The two wins count the same in the conference standings, but most fans would agree the second win was more impressive.
Clemson is 69th in KenPom as of Wednesday morning, boasting a 21-9 record. It’s easy to see why fans would question how Clemson can be 69th when West Virginia is 19th despite a 17-13 record.
That’s due in large part to West Virginia’s tougher schedule, as the Big 12 is better than the ACC this season. West Virginia is undefeated in Quad 3 and Quad 4 games, while Clemson has four losses in Quad 3 and Quad 4 games. The Mountaineers really don’t have a bad loss — they even have a 25-point blowout win over ACC-leading Pitt — while the Tigers have four losses to teams outside the KenPom top 165, and they don’t have the quality wins to offset the bad losses. One of Clemson’s losses is an 18-point defeat to Loyola-Chicago, which is just 9-19 on the season and outside the KenPom top 250.
The ACC could be looking at zero Top 4 seeds in the NCAA Tournament along with zero teams in the NET & KenPom Top 30.
— JBR Bracketology (@JBRBracketology) February 25, 2023
The Tigers are also 5-2 in games that went to overtime or were decided by three points or fewer. Five of their eight losses are by at least eight points.
“I get that it’s not the most intuitive thing, but if you’re losing by a lot and winning by a little, it probably means you’re closer to middling than awesome,” Root, an Action Network contributor, said.
Predictive metrics offer betting insights
Sportsbooks use power ratings, home-court advantage, injuries, and other factors to set betting lines. Predictive metrics like KenPom yield insight into what future sports betting lines might look like.
“You check KenPom, Bart Torvik [projections], and traditionally if you average those, you’re going to be within a point of the line,” Root said.
Metrics like KenPom also include occasional outliers. Ohio State was in the KenPom top 40 going into a Feb. 12 game with Michigan State, despite an 11-13 overall record and 3-10 mark in Big Ten play. The Buckeyes were competitive in losses to good teams, keeping them afloat in the metric.
The metric has since corrected itself as the Buckeyes’ level of play has declined. They recently lost four consecutive games by an average of 17.3 points, and Ohio State currently sits at 65th in KenPom’s rankings.
NCAA Net Rankings continue to be hot garbage.
No. 67 Ohio State (11-16, 3-13)
Quad 1: 3-11
Quad 2: 2-4
Quad 3/4: 6-1
No. 72 Wisconsin (16-11, 8-9)
Quad 1: 5-6
Quad 2: 5-4
Quad 3/4: 6-1#Badgers
— Evan Flood (@Evan_Flood) February 23, 2023
“It’s not perfect, and I think that’s people’s response,” Root said. “They’re like, ‘Well, here’s two examples of KenPom being wrong.’ Of course that’s going to happen. There are 360 teams and 8,000 games in the year, so there’s going to be a lot of incorrect examples on a micro basis.”
KenPom numbers don’t mean that every team with a better adjusted efficiency rating will beat teams with worse efficiency metrics – college sports are highly enjoyable because of upsets – but it provides insight into matchups. It’s valuable data for the NCAA Selection Committee, and it’s valuable for bettors ahead of March Madness as they fill out their brackets.
“I don’t mind fans being against them, because for a fan it’s more important who we beat, who we lost to. But I would implore them to understand why the predictive stuff is important for valuing how difficult a future game might be,” Root said.
Since 2003, every men’s basketball national champion has finished the season in the top 40 of both KenPom’s offensive and defensive efficiency metrics. As it currently stands, that trend would take notable contenders Tennessee, Gonzaga, and Baylor out of the running for the national title.
In the last 10 years, only one team — UConn at No. 40 in 2013-14 — went on to win the national championship after being outside the top 10 of KenPom’s rankings on Jan. 1. Many of the betting favorites fit that criteria, including notable teams like Houston (+650 to win the national title on FanDuel), Kansas (+800), Alabama (+900), Purdue (+1000), and UCLA (+1000).
While not always loved by fans, predictive metrics hold value. They’re worth considering when betting on college basketball games this March, especially if you’re scrolling through mobile sportsbooks looking for a national title futures bet.
Photo: Ryan Hunt/Getty Images