You all make things way too complicated for me. All I know is teams ratings improve based on how much they exceed or fall short of expectations each game…basically how they perform against the spread, only it’s an unknown spread based on their NET ratings. But you can pretty well look at Vegas spreads and know if a team helped or hurt their raw rating.
I completely realize it’s off/def efficiency and yada yada but the result is the same. What others have said about road wins being a bonus is probably right too because I’ve noticed that…so probably if a home team should be a 2 point favorite they’re 4 in NET or something like that. And again completely realize that’s not how it actually works but in terms of results it’s pretty close.
I think it’s a terrible way to actually rate teams because the nature of basketball is some teams beat up on inferior competition but can’t beat teams ahead of them. It’s how VA Tech is consistently over ranked. But since they’re only supposed to use it to categorize opponents, it’s not a bad system. Could just as easily use Kenpom though haha
True or False: your team’s NET ranking does not matter for NCAAT selection, but the NET rankings of your opponents does matter in evaluating the quality of your team’s wins and losses?
The Wayback Machine has got it covered. As of Tuesday afternoon, Miami was ranked 44th. Right now they’re 38th. Torvik has them going from 34 → 26. So a pretty similar jump but a lot less than the 20 places from NET, though both KP and Torvik had them ranked lower than the NET.
Either way it seems like NET rankings are closer to raw efficiency numbers than whatever the advanced analytics guys are doing.
It goes on the team sheet with other rankings (KenPom, BPI, SOR, etc), but they claim it has no primacy (so kind of false, but has some truth)
Your opponents’ NET rankings determines your quadrant record, which I think is much more important. (so definitely true, and more important for opponents than your own)
@Raleigh_Hoo would know all this much better than me…
I’d guess that they used a machine learning technique to figure out how to weight the different inputs, and perhaps they rerun the process after each year, so the weighting vary slightly year-over-year.
Given that the entire thing is super opaque, there’s no telling how interpretable it is or isn’t. The opacity of how it’s calculated is my biggest beef with the NET.
I believe this interview will be between some of the posters on LRA’s two favorite people… it may be an epic conversation that can be used for banter for years to come!
Mea culpa: thought our offense would be ahead of our defense. So far looking like the reverse.
Probably biggest surprise is the team leaning into a way more disruptive version of pack line. Didn’t see that coming but probably should have with Reece, Rohde, and Dunn. Thought the lack of a go-to 5 and having to play Groves would hurt us more, but so far it’s been manageable.
