Newest NET ranking (ie, the one that counts)

[WiseWillie]

...and as a side note for more on the NET here's a Q&A online that might answer some questions.

https://ncaaorg.s3.amazonaws.com/championships/sports/basketball/d1/women/2023-24D1WBB_NETFAQ.pdf

Note the comment in one of the Answers about the use by the NCAA of "category" or "categories" not "quadrants" or "quads"

 

[Bone Dog]

...and as a side note for more on the NET here's a Q&A online that might answer some questions.

https://ncaaorg.s3.amazonaws.com/championships/sports/basketball/d1/women/2023-24D1WBB_NETFAQ.pdf

This is very helpful. Thank you.

For me, the interesting question is the Team Value Index. The Adjusted Net Efficiency makes good sense to me. For the TVI, we get this explanation:

"Team Value Index is the results-oriented component of the NET, ranking more highly those teams that played and beat other good teams, factoring in opponent, location of the game and winner."​

It's a little vague, and I get that there's an AI at work behind it making it too complicated to lay out a simple algorithm. What I wonder is why the TVI doesn't simply consist of the NET of the opponents. Perhaps there's a sort of logic loop [?] implied in this which makes it undesirable. In any event, the focus on efficiency in the ANE seems like a better approach than merely assessing W-L and MoV.

 

[sun]

I’m not understanding why the point total is of importance. There is nothing a team can do to “catch up” but keep winning, which is every team’s goal.

My point is that NET is more important than any poll because the polls aren’t used in seeding for the tour. But NET is only one criterion; UConn’s overall strength of schedule and performance in conference play may end up being more valuable

Huskynan, If you've ever watched auto racing, toward the end of a race the announcer will provide the lap times of the 2 lead cars.
For instance the announcer will say car A is 10 seconds ahead with 7 laps to go, & car B is running laps 2 seconds faster than car A.
This reveals that car B could possibly catch up to car A in 5 laps with 2 more laps to go after that.
Then it's up to the drivers & the traffic on the track to see which car willl win.
But at least the viewer is being informed in advance that it's mathematically possible for car B to catch up to car A if there aren't any crashes.
It can signal that the ending is worth watching & will be exciting.
If the viewer is kept in the dark about the lap times then they can lose interest in the outcome which isn't good for the TV ratings & advertisers.
The NET is leaving out information that could show emerging trends regarding how solid of a lead that UConn has on its #4 NET ranking relative to other teams close to them in the rankings.
It could generate more interest by letting fans see the daily fluctuations in a team's actual NET score vis-a-vis opposing teams scores instead of simply seeing their rankings.

 

[Plebe]

Just looking at the bullet points on that, at least we have no bad losses and were competitive in the three losses (by 11, 11, and 12 points). Hopefully we can be perfect in conference. Wins against ND and South Carolina + a Big East title locks up a #1 seed IMO even if there are a bunch of other P5 teams with less losses.

“Competitive in losses” is obviously a relative concept, but it's a stretch to say we were competitive in those games. We were quite thoroughly outplayed in all three, more so than what the final scores would indicate.

While we've beaten several NET top 50 teams, what UConn's resume is most lacking is high-quality wins. Our only win over a NET top 25 team is Louisville. The Notre Dame game looms large for our resume.

A win at South Carolina? Hope springs eternal, I suppose.

 

[Bone Dog]

We were quite thoroughly outplayed in all three, more so than what the final scores would indicate.

I think that's clearly true. But does the NET algorithm distinguish this? I haven't looked at any efficiency numbers for those games, so I'm curious if that would reveal the lopsided character of those games for the algorithm.

 

[sun]

I think that's clearly true. But does the NET algorithm distinguish this? I haven't looked at any efficiency numbers for those games, so I'm curious if that would reveal the lopsided character of those games for the algorithm.

I recall reading an article about an NET update that said that (about) a 10 point win was in effect the maximum benefiit that would be weighted within the formula.
That was because blowouts had previously been given more weight, but apparently that was changed.

 

[Plebe]

I think that's clearly true. But does the NET algorithm distinguish this? I haven't looked at any efficiency numbers for those games, so I'm curious if that would reveal the lopsided character of those games for the algorithm.

Not as far as I know. Sport stat geeks have long talked about “game control” (in-game score margin and win probability) but I've never seen any report that this is specifically tracked by the NET.

However, I'm sure this is something the committee members would be watching for and, when the time comes, discussing. Obviously there's a huge difference between a 20-point deficit that shrinks to 10 in the final 3 minutes vs. a one-possession game that becomes a 10-point win in the final minute.