019: Nothing but NET

More conference comparisons using everyone's favorite metric punching bag.

This is an attempt to share how the NET rankings have changed over time across the top seven conferences in college basketball.

Notes on the data analysis before we begin:

• Conferences are Big 12, Big Ten, SEC, Big East, Pac-12, Mountain West, and ACC

• Compares NET ratings published on December 28, 2022 and today (March 1)

• Subtracts the December rating from the March rating:

  • negative number means rating is worse

  • positive number means rating is better

• Compares these differences to the kenpom ratings on those same dates

• Shares the median¹ change in rating across the different conferences to account for outliers or bad teams (not using the average!)

  • this takes all the changes for each team in a conference, and computes the median change in rating across those teams

• Shares the difference in NET and kenpom rating for each team sorted by conference

Conference comparisons

• The ACC is the curious conference with these numbers. The league’s median rating change is 12 spots better in the NET and 25 spots worse in kenpom.

• Take Florida State as an example:

  • 149th in kenpom on December 28 and 206th today (-57)

  • 248th in NET on December 28 and 218th today (+30)

• Change across other leagues is modest. The Mountain West is the only conference with a better median rating change in both metrics, +20 in NET and +9 in kenpom.

• New Mexico is a bubble team and a fun example:

  • 65th in kenpom on December 28 and 46th today (+19)

  • 12th in NET on December 28 and 48th today (-36)

All teams

The chart below lists all teams with the change in NET rating and kenpom rating between December 28 and today (March 1).

As an example, Boston College:

• 69 spots better in the NET and two spots worse in kenpom

  • 227th in NET on December 28 and 158th today (+69)

  • 164th in kenpom on December 28 and 166th today (-2)

What does this all mean?

The NCAA Evaluation Tool is opaque.

The NET gets blurrier as you start to dig in and compare numbers across conferences. There is certainly reason to be suspicious of the NET ratings in December. And maybe more reason to be suspicious in how those numbers can change over the next three months.

Of course, the committee of 12 people might not use the The EZ TeamSheet Solver (credit to John Gasaway) and ignore TeamSheets. Who knows?

Over the next 10 days, the bubble will intensify. The best way to get off the bubble is to simply win games.

If you’re interested in other tools information to understand which teams get in or out of the tourney, a couple recommendations outside of the NET and kenpom:

• Bart Torvik: the tools on the site are elite. You can game out conference title odds or build your own bracket. The WAB (wins above bubble) metric is super handy, and the TeamSheet ranks are too.

• Seth Burn: puts together an aggregate of WAB, Parcells (his own metric), and Pythag across different metric systems. Firm believer in eliminating the committee, plus Seth has collaborated with @5th_Factor in the past to showcase how quadrants are bad.

---

Here is the code to generate the tables and numbers. The kenpom ratings require a subscription (you can subscribe here) and these were fetched using hoopR, part of the sportsdataverse.

If you see a mistake or question the math, good for you! Please let me know in the comments, so I can learn something and fix it.

1. Possible using the median change in ranking is a poor approach. It felt more accurate than the average because of some wacky outliers (hi, Louisville!)