Skip to content

Kentucky Sports Radio

University of Kentucky Basketball, Football, and Recruiting news brought to you in the most ridiculous manner possible.

Understanding the NET, the NCAA’s replacement for the RPI

Earlier today, the NCAA announced a major upgrade to the tournament selection process: the NET, or the NCAA Evaluation Tool. Starting this season, the NET will replace the RPI as the primary metric the selection committee uses to choose and seed teams for the tournament.

Drew gave you an overview of the NET earlier, but here’s a little bit more about how it works and what it means for Kentucky.

The Basics

According to the NCAA, the NET relies on game results, strength of schedule, game location, scoring margin, net offensive and defensive efficiency, and the quality of wins and losses. It was developed with input from the Selection Committee, the National Association of Basketball Coaches, basketball analytics experts, and Google Cloud Professional Services (!). Developers used late season games in test sets to predict outcomes and optimize the system until it was as accurate as possible.

It’s important to note that the NET isn’t the only metric the selection committee will use; it simply replaces the RPI as the sorting tool. Tournament team sheets will still use the quadrant system, ESPN’s BPI and Strength of Record, along with rankings by Kevin Pauga, KenPom, and Jeff Sagarin.

Scoring margin now matters…sort of

One big difference between the RPI and NET is that the NET accounts for scoring margin, whereas the RPI only ranked teams by wins, losses, and strength of schedule. Scoring margins now matter, BUT they’ll be capped at 10 points “to prevent rankings from encouraging unsportsmanlike play, such as needlessly running up the score in a game where the outcome was certain.”

Is it just me or does ten points seem like a low cap? The difference between a ten-point win and a 20-point win is huge. I’d raise that to 15, but that’s just me.

When a game takes place does not matter

The NET will factor in the location of the game, but it won’t consider the date, so Kentucky vs. Duke on November 6 will be just as important as a game in the SEC Tournament.

Yet another reason to beat Duke.

You’ll be able to check the NET daily

Starting in late November, NET rankings will be publicly available and updated daily through Selection Sunday and again after the tournament’s over. That means you’ll see it A LOT on this here website.

Unfortunately, we may never really know exactly how it works

CBS Sports’ Matt Norlander spoke with Dan Gavitt, the senior vice president of basketball for the NCAA, who said the NET algorithm is powered by artificial intelligence, so it’s not something that can fully be explained to the public, nor will the organization run it on previous seasons for comparison. So, essentially, you won’t really know how the sausage is made.

But, it’s better than the RPI, right? Baby steps.

[mobile_ad]

Article written by Mrs. Tyler Thompson

No, I will not make you a sandwich, but you can follow me on Twitter @MrsTylerKSR or email me.

9 Comments for Understanding the NET, the NCAA’s replacement for the RPI



  1. Bluebloodtoo
    3:24 pm August 22, 2018 Permalink

    Ten points did seem low. A close game inside the 1 minute mark can yield 10 point differential. I hope they did some analysis to define that 10 point margin.



  2. Bluebloodtoo
    3:25 pm August 22, 2018 Permalink

    I also agree with not sharing the algorithm. If they did, it would/could be manipulated by coaches.



  3. somerset bill
    6:35 pm August 22, 2018 Permalink

    I agree that the 10 point cap is strange and kind of useless. Seems more like a 20 point cap would be a better indicator of dominance without padding of scores. A 17 point win is pretty definitive. But no 17 point win is the product of a team running up the score on a weaker one.

    As far as the algorithm goes, if the rankings are going to be updated daily, then the algorithm will soon be completely transparent and every last factor and weighting percentage will be known to the nth digit — probably before January.



  4. ukjaybrat
    8:28 am August 23, 2018 Permalink

    yeah 10 points is low. I’ve seen really close games (2-3 points) that end up in 8-9 point wins bc of free throws, which is why gambling on basketball is a fool’s bet.

    i agree with others it should be 15 or 20. no one has ever “run up the score” to a 17 point win. heck cal doesn’t even put the bench in unless we’re up 25+ (or there’s less than 60 seconds left in the game)



  5. ukjaybrat
    8:32 am August 23, 2018 Permalink

    “When a game takes place does not matter”

    I have to disagree with this to a certain point as well. i think anything in the first month should have a little less weight for obvious reasons. after that? fine – everything can be the same. maybe this is a biased attitude because my team is full of new freshman we rely on every year. but it seems like even with upperclassmen the first month is more about building toward the rest of the season than winning basketball games.



  6. rickshelton
    10:06 am August 23, 2018 Permalink

    The algorithm should be fully known since any secrecy leaves room for meddling. For example, what determines a “quality win”? Is that based on a human ranking or expectations based on past performance or a program’s reputation? Will the NET prevent the selection committee from creating matchups that are, for example, unfair for Kentucky because their good for TV ratings?



    • ukjaybrat
      11:45 am August 23, 2018 Permalink

      i think a lot of people are misunderstanding the part about not sharing the algorithm. they aren’t “not sharing” to hide something. they are “not sharing” because the Google Cloud Professional Services is powered by artificial intelligence. there’s no amount of explaining they could do that the general public would understand how it works.



  7. Luether
    2:58 pm August 23, 2018 Permalink

    Speak for yourself, brat…



    • ukjaybrat
      3:00 pm August 23, 2018 Permalink

      well, i made three comments and you didn’t reply to any of them. so i don’t know what you’re referring to. so good job there