It’s been tinkered with since 2018. NCAA consulted Google and no longer uses winning %, adjusted winning %, and scoring margin as of 2020. Wouldn’t be too surprised if they adjusted it even more after that.
NET Rankings are Flawed
- Thread starter BillBrasky4Cy
- Start date
It’s been tinkered with since 2018. NCAA consulted Google and no longer uses winning %, adjusted winning %, and scoring margin as of 2020. Wouldn’t be too surprised if they adjusted it even more after that.
ah, good to know. I pulled the tweet from their page explaining NET, which was updated Jan 27. The NCAA really sucks at keeping things updated.
College basketball's NET rankings, explained
Here's everything you need to know about college basketball's NET ranking system.
NET uses net efficiency on a ppp basis. So, it's processing the numbers differently, but it very much uses margin of victory. If you win by 10 and both teams were efficient on offense, that means both were inefficient on defense. The reverse is true for a 10 point win where both teams were inefficient on offense. It is impossible for net efficiency to not correlate almost perfectly with margin of victory.
Though they can and should be using some of the same metrics, the NCAA tourney committee has a different objective. Right or wrong, the committee is not necessarily supposed to pick the at large teams that have the best chance to win, they are supposed to reward teams for what they've done in terms of W/L and SOR throughout the season. Otherwise, they would have a strong emphasis of "last 10" to get really hot teams into the tourney.
This always annoys me that they list scoring margin and net efficiency as separate metrics. Other than capping it at 10 points, I would love someone to be able to explain situations where scoring margin will not correlate almost perfectly to net efficiency on a ppp basis. Because other than a tiny number of instances (half times and technicals) possessions between teams are going to alternate, a scoring margin ranking and ppp net efficiency ranking are always going to be almost identical. Every game each team is going to have within 0-2 possessions of one another. It is not mathematically possible to a huge margin of victory and have a small net ppp difference. It's not mathematically possible to have a small margin of victory and a large net ppp difference.
Sorry, there's no way beating a bubble team at home when the regular season is over 80% complete should jump you up that many spots as beating OU did, especially when you look at the minimal movement we've seen from winning tougher games. There's just way too much weighting for margin of victory and net efficiency, which are essentially the same.
With that said, the initial selection signaled big time that the overall rankings are not used very closely, and that they are looking at quad performance very strongly.
That graphic is old. They removed the 5th factor (MOV) a couple of years ago.
I get what you are saying about MOV correlating to the efficiencies numbers, and I agree. But I would bet RPI and average MOV also correlate really well. Pretty much any ranking and MOV is going to correlate, because the better teams are going to have a higher MOV on average. That's why it's used in almost every predictive model.
I understand your point about making that big a jump in the NET based on one game this late in the season, but we actually made a bigger jump in RPI. Any model you come up with is going to have quirks you have to deal with.
From the selection process handbook:
The committee selects the 36 best teams not otherwise automatic qualifiers for their conference to fill the at-large berths.
That bold is not mine by the way, it's theirs.That's pretty much all they say on what teams should be selected at large. So you tell me what you think "best" means, and I'll tell you what I think "best" means, and neither of us is wrong. It's subjective, and so is the process of selecting the teams. It's up to each committee member how they want to weigh all the different data they have and what they think ultimately makes one team better than the other.
Unfortunately "best" is far too often skewed by what is best for my particular league.
That's good to know regarding MOV being removed. It doesn't make sense to include in my opinion as long as NE is included. And I'm not saying MOV and rankings or NE correlate closely. Yes, they all do. What I'm saying is MOV and NE are essentially the same. You can have low MOV or low NE but be great in RPI or other systems because you win all the close games and lose in blowouts. So it's possible to have some significant disparity. What I was saying is that MOV and NE on a ppp basis mathematically can not have significant disparity. It's not possible. That's why when people proclaim that there is not MOV in NET, it's just semantics.
I do think "best" is a good way for them to state it, as it gives a little wiggle room. I think within that "best definition" there is still a bit of a split between best resume vs. most dangerous/most likely to go on a run. That's where you can get some significant differences in selections and in seeding.
Generally speaking I think the NET is a good tool to create some guidelines for the quads. I'm OK with taking the actual ranking itself into significant consideration, I just hope that outliers are treated as such.
ISU drops a spot from 34 to 35 after a win because Loyola jumps several spots after beating Net 315 Loyola, that is some funny stuff.
coulda used xavier over providence last night
We won’t move much from this point on. We will end up in the 30-40 range regardless of results. Unless we win out including at Baylor. Then maybe we’d be like 25th.
Iowa State has played 19 Q1 and Q2 games with 15 of those being Q1. NET is facked, at some point who you beat and who you lose to has to matter. In general, NET doesn't really care for the best league in the country. After the top 4 in the Big 12 the NET drop off is pretty sharp which makes little sense.
Which is total BS. Evansville literally quit in that game. They were a very generous foul call away from not breaking 30. Evansville's total lack of effort and interest had way more to do with that than anything Loyola did.
Side note: Since 1950, Evansville has lost a game by 32 or more points 20 times. Five of those have occurred this season. (Source: Evansville Courier & Press)
And word out of Evansville is that they aren't ready to fire Lickliter or if they do they will just give the job to his top assistant. Insanity.
Why be content with finishing 10th when you can start finishing 12th next year...
Nope. Enrollment is dropping like a rock and they are too prideful to realize that they don't fit in The Valley anymore and should probably drop to the OVC if not D3. It's a mess.
They are going to pull off the almost impossible feat of finishing dead last in the Valley in every sport except volleyball (7th) and Men's Golf (4th).
Iowa State picks up its 9th Q1 win and they are rewarded with DROPPING two spots! NET is f*cked.
The officiating in the last 2 min probably cost us at least 2-3 spots. Maybe more. We should have won the game by at least 10 and NET seems to love efficiency over wins and losses.
