Michigan ranked 4th in initial NET Ranking release

2 Likes

First NET rankings released (replacing the RPI this year): https://www.ncaa.com/rankings/basketball-men/d1/ncaa-mens-basketball-net-rankings

OSU #1, Michigan 4th, MSU 7, and Wisconsin 8

4 B1G teams in the top 8, 8 in the top 20. Definitely a better start for the conference, and will make a ton of difference come tourney time if the B1G can keep this up for the next month or so.

This is the new NCAA Selection Committee metric. When it was announced I wrote about how the lack of transparency was concerning. That hasn’t really changed.

This graphic is basically all the information the NCAA has provided about what goes into the NET.

Regardless of how much the black box nature of the metric annoys me, it bodes well for the Big Ten that so many teams are ranked highly right now.

If the Big Ten can nab a few key wins off of the ACC this week, conference should be in a strong position come March.

It is also ridiculous but not surprising that we don’t get any sort of rating measure. There’s no indication of how far 1 is from 2 or 5 is from 50. Strictly numbered ranks.

It is also probably way too early to release this. If you ran KenPom without preseason factors right now it would probably look wonky. The RPI after Thanksgiving probably looks wonky every year.

1 Like

What is the deal with the plus/minus signs in Net Efficiency? Should they be opposite for offense and defense. Eg, steams are good on defense and bad on offense.

Under the O and the D? They are just explaining how possessions are calculated:

FGA + TO - OR + .475 * FTA is the standard for estimating possessions in college hoops.

1 Like

The interesting thing to me is that Section 2 essentially takes margin of victory/loss into account, which was something the NCAA has never allowed in football and wasn’t in the RPI calculation. In the BCS era, all the football computer algorithms had to be switched so as not to account for margin. And now it’s in the NET for all intents and purposes.

It will be interesting to see if some scores are run up come late February/early March by bubble teams…

Kind of surprised that OSU is so high in their rankings, but when you consider they beat two decent teams on the road and their margin of victory was 9 points in each (thereby avoiding the collapsing margin of victory penalty in NET), it sort of makes sense.

Correct… but you also have “efficiency margin” which is essentially normalized margin of victory… just uncapped. And then also “margin of victory” capped at 10 points. The whole thing makes no sense.

1 Like

Well, yes. Would you expect a system that had Loyola Marymount ranked ahead of Kansas to make sense?

I’m worried about process not results. Any ranking system that doesn’t factor in last season is going to have some wacky results early in the year.

You don’t see these sort of outliers in something like KenPom because he puts so much emphasis on the preseason stuff to prevent those sort of outliers (and make things more predictive).

I accept the #4 ranking on this. Great news also 4 B1G teams in the top ten. Espn has B1G second only to ACC in non conf win %. Hopefully the ACC Challenge will reverse that stat.

Too early.

So at what point do the “results” of the “process” have to pass the eyeball test, not to mention the laugh test? NET is already being dismissed and not taken seriously by people who actually watch the teams play.

Well I’m not saying releasing the numbers on Nov. 26th is smart.

Loyola Marymount is 4th in the RPI right now, but if we were still using the RPI then no one would look at that

http://barttorvik.com/rpi.php

The bigger problem is that we have no idea how NET is calculated beyond some basic talking points. It would be a little bit more convincing if they released last year’s end of year numbers like Gasaway suggests.

The RPI had many issues but it got things in the ball park by the end of the year. We also knew exactly how it was calculated.

Agreed on your penultimate paragraph. Best way to prove its worth is to use it on last years teams.

1 Like