NCAA Basketball: A look back on season one of the NET rankings

Last August, the NCAA unveiled its new ratings system and changed NCAA Basketball. Here’s a look back on the first year of the NET rankings

One year ago, the NCAA released its new NET rankings system, the acronym standing for “NCAA Evaluation Tool”. The system, designed as a replacement to the aging RPI that had been used for nearly four decades, was a cooperative effort between the NCAA, coaches, analytics experts, and Google, per a release on the NCAA website. According to that release, the NET was optimized using machine learning – a far cry from the simplistic RPI, which has a very simple and transparent formula.

It’s that second part – transparency – that may be the most jarring difference between the two metrics.

A team’s RPI rating is quite easy to calculate. The team’s winning percentage accounts for one-quarter of the rating, while their opponents’ average winning percentage comprises another fifty percent. The remaining 25% relies on the average winning percentage of your opponents’ opponents. That’s it. Any Tom, Dick, and Harry with access to team schedules can pump this out in no time. No controversy, no mess.

But what about the NET?

The NCAA’s new system is made up of many different components now, far beyond simple winning percentage. The most notable new inclusion, perhaps, is the implementation of offensive and defensive efficiency, a clear step in the general direction of the analytics movement. These numbers, which can be adjusted and calculated in various ways, are the foundation of many leading analytics systems, including Ken Pomeroy’s ratings.

The NET also factors in data such as game score, location, and strength of schedule. Additionally, wins and losses are judged on their quality, ostensibly using the quadrant system that has been in place since 2017. This system, for the uninitiated, divides wins and losses into four levels of quality based on the location of the game and the current NET ranking of the opponent. The quadrant system proved to be significant in the NCAA Tournament selection process last March.

At least it seemed significant. Honestly, it’s hard to know what is significant and what is not.

The NCAA has not yet released specifics on how all the data is adjusted and weighted in the final product. And it probably never will. In fact, teams’ actual NET ratings are not displayed publicly – only their ranking. By omitting the actual rating, the NCAA has made it difficult to quantify the disparity between two teams. Perhaps Team #1 is rated only 0.1 points higher than Team #2, but ten points higher than Team #3. But the NET rankings only give you the order of the teams, so you don’t get the complete picture.

And that lack of clarity is what caused some resistance to the NET in its first year as the new, official metric of the NCAA.