On the list of people you might have expected to see quoted in an article about college basketball and the array of numbers at the core of the NCAA Tournament selection process, I imagine you wouldn’t have expected to encounter the words of the late British economist Ronald Coase.
“If you torture the data long enough,” he once said, “it will confess to anything you like.”
Who knew he was such an expert on Big 12 basketball?
Yes, I’m kidding, and yes, I know it’s dangerous to joke about such serious matters as March Madness. Everyone wants to be in. Not everyone gets the chance. And few, even among those affected, have a full understanding of why their team isn’t on the ideal side of the NCAA “bubble”.
And it might be as difficult to figure out this season as any in recent memory, because the metric at the core of the process – the NCAA Evaluation Tool, or NET – is speaking a language some outside the Big 12 apparently haven’t been able to translate.
MORE: Latest projections for men's bracket | Women's bracket
“You can manipulate the NET,” Clemson coach Brad Brownell told WCCP radio in February. “And whether people want to say the NET is the be all, end-all, it’s all anybody talks about. That’s why it is so powerful.
“(The Big 12) nonconference scheduling, they are playing 300-level teams and winning by 40 and 50 points to increase their offensive and defensive efficiency numbers, which is a big part of the NET tool. So that’s why you see teams trying to win by 30 and 40 points instead of putting in their walk-ons.”
Brownell is not entirely incorrect, but he’s also not entirely correct. And there is so much left unsaid in all of that.
Led by point guard Jamal Shead and the champion Houston Cougars, the Big 12 did, collectively, play poorer non-conference schedules on average than teams in the other major conferences. We’ll excuse the Pac-12 from the discussion, because it’s their last year and because they’ve spent most of it testing the boundaries of the word “major”.
Big 12 teams played 7 Quad 4 opponents per team, compared to 5.3 for the ACC, 5.6 for the Big Ten, 5 for the SEC, and 4.8 for the Big East. Quad 4 is generally the weakest 25 percent of teams in the NET. Those are the games Brownell is complaining about and that underscore his point.
On the high end of the scale, members of the Big 12 played fewer Top 50 NET opponents. But the ACC did not rank first in this category. Instead it was the Big East, which played 2.63 Top 50 NET opponents per team, followed by the ACC (2.6), SEC (2.57), Big Ten (2.07) and Big 12 (1.85).
When it came to playing the best of the best, it turns out the ACC was no better than the Big 12. Each played an average of 1.27 Top 25 NET teams. It was the SEC that played the most (1.78), followed by the Big East (1.45) and Big Ten (1.35).
And then there’s the matter of actually winning games. That's the most important component of any computer ranking, so it's not as easy as making a declaration.
The ACC was 5-14 against NET Top 25 teams, 10-10 against the Top 50, an overall winning percentage of .385. The Big 12 was 6-8, 4-8 (.385). The SEC was 7-18, 5-6 (.333). The Big Ten was 8-11, 5-7 (.419). The Big East was 7-9, 8-5, and thus the only league with an overall winning record against Top 50 opponents (.517).
(Note: The order of the NET rankings can change from moment to moment, and certainly day to day. These numbers were compiled over the course of a week. But all teams were in the Top 50 or Top 25 categories when examined, and the nature of such rankings means they remain in those categories or close).
There were members of the Big 12 that are enjoying winning seasons that did not build their success on a foundation of powerful non-league opposition. Iowa State’s highest-rated opponent was No. 54 Virginia Tech, and the Cyclones lost that game. Their highest-rated victim was No. 57 Iowa. Iowa State now stands 13-4 in the Big 12 and No. 8 in the NET. Texas Tech, now No. 37 in the NET, did not beat anyone outside the conference ranked higher than No. 108 Northern Iowa. For NET No. 38 TCU, the highest-ranked victim outside the Big 12 was No. 124 Arizona State.
Big 12 teams, however, were not disadvantaged by playing in a league with a large number of non-achievers, as has been the case with the ACC. Outside the league, Wake Forest defeated No. 27 Florida, but after that it was all the way down to No. 90 Rutgers. Pitt’s highest-ranked victim was No. 147 West Virginia, and the Panthers lost at home to No. 157 Missouri. Inside the ACC, Pitt was swept by No. 84 Syracuse and lost at No. 94 Miami. Wake lost to No. 98 Florida State, No. 130 Notre Dame and No. 125 Georgia Tech, the last of those at home.
The ACC has three teams more than two games under .500. The Big Ten has one. The Big 12 has two.
All of this begs a few questions that frustrated ACC members might want to avoid:
DECOURCY: Expanding the NCAA Tournament is messing with perfection
First: If members of the ACC were aware the NET could be manipulated, why weren’t they trying? The premier conference in college basketball before its decision to expand in 2003, the ACC received only 10 March Madness bids combined in 2022 and 2023, compared to 17 for the Big Ten, 14 for the SEC and 13 for the Big 12.
When Pitt was contending for the ACC title a year ago -- the Panthers lost share of the regular-season title with a two-point road loss to Miami on the final day of the regular season – they barely were able to scrape into the NCAAs as a First Four entrant.
All this would have created a situation that called for urgency. If the solution to this were as easy as playing poorer non-league opponents, why didn’t the ACC take that approach?
“It’s something that we have to figure out,” Capel told the Pittsburgh Post-Gazette. “Because it’s being talked about, it will probably be changed next year. I think where we have to be better as a league is trying to get out in front of maybe where this thing is going and then adjust from that standpoint. You have to always be able to adapt.”
Second: If teams can game the NET, does that mean they also are gaming KenPom? Statistician Ken Pomeroy’s ratings at KenPom.com are considered the game’s standard. His statistical data is indispensable to reporters covering the game and to a wide variety of programs. But there is so much similarity in numerical ranking from KenPom to the NET to ESPN’s BPI, to suggest a league has gamed one is to say it has done so across the board.
As of Thursday morning, the NET rankings, KenPom and BPI contained the same top 11 teams, albeit ordered slightly differently. Of the top 20 teams, 19 were the same in NET, KenPom and BPI. The NET has KenPom’s No. 19 Michigan State at No. 22. BPI has KenPom No. 20 San Diego State at No. 32.
But there’s been no apparent expression of displeasure among coaches about the influence of the KenPom or BPI rankings, each of which is displayed prominently on every team sheet (resume) presented to the NCAA Tournament selection committee.
Third: If the ACC coaches are bothered that there’s a predictive metric at the core of the selection process, why have there been no public complaints about this? The NET is not a standard that provides the final say on team selection; it is used by the committee as a grouping tool. It provides a measurement of accomplishment through the Quad system and through examination of any particular result. But if coaches are unhappy about any predictive metric – one that emphasizes efficiency and dominance or, simply put scoring margin-- being so important in the process they have not acted to have the process altered.
If the committee were to settle on a results-based metric for their standard, there is bad news for those angry at the Big 12 this season. Those metrics appear to love the league even more. Five of the league’s 14 teams are in the Strength of Record top 10, nine are in the top 35, and only three are 70 or lower. The ACC has one in the top 10, -- North Carolina and star guard RJ Davis -- along three in the top 35. Eight of its members are 70 or beyond.
If you think about it, maybe Professor Coase was wrong.
No matter which numbers the ACC is working over, the data won’t say anything nice about the league.