The NCAA's New "NET" Rankings Are Off To A Rough Start
By Tully Corcoran
If you’re going to talk about college basketball rating systems, you have to start by acknowledging the inherent goofiness of the endeavor.
There are 353 Division I basketball teams all playing 30-plus games against a small but varied portion of those 353 teams in a sport with a high level of variance in performance from one night to the next. And we’re trying to come up with a math formula that’s going to tell us just how impressed we ought to be with Kansas’ win over Iowa State on a Tuesday night in February, as compared to a Creighton win over Cal three months earlier.
But that’s the task the basketball world has assigned itself, and the NCAA has rolled out the latest in college basketball ranking technology.
The main function of the “‘The NET,” as I’ll be calling it, is to replace the RPI, which everyone hated, with a new thing everyone will hate. Because, again, this process is fundamentally ridiculous.
We’re only a few weeks into the college basketball season, so the small sample of data is naturally going to make these rankings look dumber in November than they will in March.
It’s just … look at these.
- Ohio State
- Virginia
- Texas Tech
- Michigan
- Gonzaga
- Duke
- Michigan State
- Wisconsin
- Virginia Tech
- Loyola Marymount
- Kansas
- Belmont
- Nevada
- Nebraska
- Iowa
- Auburn
- Maryland
- Houston
- Notre Dame
- Purdue
- North Carolina
- Radford
- Pittsburgh
- Kansas State
- San Francisco
Let’s compare this formula-based ranking system to what we’ll use as a proxy for “public opinion” or “the eye test” — The Associated Press Top 25.
- Gonzaga
- Kansas
- Duke
- Virginia
- Nevada
- Tennessee
- Michigan
- Auburn
- Michigan State
- Kentucky
- North Carolina
- Kansas State
- Virginia Tech
- Iowa
- Florida State
- Ohio State
- Texas
- Oregon
- Purdue
- Texas Tech
- Buffalo
- Wisconsin
- Villanova
- Maryland
- Mississippi State
What jumps out here is that The NET’s No. 1 team, Ohio State, is 15 spots lower in the eye test, while the AP’s No. 2 team, Kansas, is No. 11 according to The NET. Kansas is undefeated with two wins over AP Top 10 teams, and a win over The NET’s No. 7 team, Michigan State.
The NET and the AP also have big disagreements over Texas Tech, which is No. 3 in NET and No. 20 in the AP, Tennessee (unranked/6th), North Carolina (21st/11th), Kansas State (24th/12th), Houston (18th/unranked), Texas (unranked/17th), Wisconsin (8th/22nd), Loyola Marymount (10th/unranked), Belmont (13th/unranked), Nevada (13th/5th) … just about the whole thing, really.
It’s our good fortune that Warren Nolan calculated what the RPI rankings would be this year. So here those are.
- Kansas
- Georgia Southern
- Tennessee
- Loyola Marymount
- Duke
- St. John’s
- Texas
- Gonzaga
- Auburn
- Radford
- Murray State
- Ohio State
- The Citadel
- Nevada
- Duquesne
- Kansas State
- Wisconsin
- Virginia Tech
- Texas Tech
- Houston Baptist
- Oklahoma
- Belmont
- Central Florida
- Creighton
- Florida State
Clearly, the RPI has some goofy outliers too. Houston Baptist and The Citadel are probably not going to be anywhere near the Top 25 a month from now, if even a week. But the AP voters and the RPI are pretty much on the same page about Kansas, Duke, Ohio State, Kansas State, Virginia Tech, Wisconsin, Tennessee, Texas Tech, and some others. The big difference is with Gonzaga, which is No. 1 in the AP and No. 8 in the would-be RPI, and that neither Kentucky nor North Carolina, 10th and 11th in the AP, don’t appear in the RPI top 25 at all.
Beauty is in the eye of the beholder, and so are basketball metrics. But at this point, I can’t make a judgment one way or another in a duel between RPI and The Net. In both rankings, the Top 10 is a mess, even after accounting for the outliers.
These are just the numbers, I know. The output of a formula. A human being can look at The NET and, through observation and reason, conclude there is no way Michigan State should be ranked ahead of Kansas.
A mathematical formula lacks that capability, and its value shouldn’t be judged against human consciousness. That’s why we have the AP, the selection committee and Twitter.
A formula should only be judged against other formulas. Baking a cake is science, too, but a person still has to come up with a good recipe if anybody’s going to want to eat it.
And The NET, at this point, could use a little salt or something.