You Can't Fool The NET. UConn still 4th.

[Tenspro2002]

There was a stretch of years where they played at least two tournament games a year in NC. Sometimes 4

I think one season Duke didn't leave NC until the Final Four. The ACC Tournament, first two rounds and Sweet 16 / Elite 8 were all played in their home state

KU gets a lot of home games as well

UConn, on the other hand, played more games in the opposing team's state than anyone

Ohio St
Florida
UCLA
NC
TX
George Mason
Mich St (OK, this was in the FF)

And we lost to MIss St on a court they just won the SEC Tournament Title on

PS: This is off the top of my head

In 2011 we won regional games against SDSU and Arizona in Anaheim.

I guess in 2014 we got some payback getting to play in MSG.

 

[Hunt for 7]

You’ll have to point me to what you’re talking about with your facts.

Find print or social media content prior to Wednesday February 7, 2024 that specifically noted Danny Hurley as a contender for National or Big East COY. You stated it was being discussed quite frequently while suggesting I was making things up. Spoiler: I did this and I know what you will find.

 

[7774]

Find print or social media content prior to Wednesday February 7, 2024 that specifically noted Danny Hurley as a contender for National or Big East COY. You stated it was being discussed quite frequently while suggesting I was making things up. Spoiler: I did this and I know what you will find.

So if you did this then you know you're wrong? I'm not sure what we're doing here.

Shot Takers: And the midseason college basketball awards go to ...

Dana O'Neil and Brian Hamilton give their picks for top transfer, player of the year, Final Four teams and more at midseason.

theathletic.com

 

[Hunt for 7]

American...11-8. Kind of funny we have well over half of that conference's tourney wins

So UConn I believe got two ncaa bids while they were in the American so with two bids we ended up playing in a total of 8 games. Each bid for us was worth 4 tournament games. All of the other bids the AAC have resulted in 11 games played. During that period how many other AAC teams played more than the minimum one game which each bid represents. I have completed some research related to another debate and although not complete I am finding anywhere between 2.5 to 3 games per bid suggests the conference is performing quite well related to games played per bid.

 

[Palatine]

Yeah but, what about the Dunkel? You can fool them.

https://www.dunkelindex.com/ranking/ncaa-basketball/2024

 

[Waquoit]

The values assigned to the algorithms are still subjectively decided by humans.

I swear some folks still haven't grasped that.

 

[Hunt for 7]

I swear some folks still haven't grasped that.

I thought AI was supposed to remove that human input and develop “it’s” own eye test. If AI is learning that means everything it spits out is not based on just human input. As I understand it that is what makes it so dangerous. It starts to think on its own. Hey for me I am all for it because I believe the ncaa committee is unable to think for itself.

 

[tzznandrew]

None of this bothers me all that much. We will get at worst, reasonable respect with our seeding for the tournament and from there, it's up to us to perform.

On a side note, I saw a chart that had us with (I believe) nine Q-4 wins (considerably more than those ahead of us). I think we need to cut a couple of those out but I can somewhat understand the thought process at least for this year. With normal seasons for UNC, Gonzaga, Texas and Indiana, we should have had more than sufficient strength of scheduling with an excess of Q-4 games thrown in.

I've been banging this drum all year and been told there's nothing to see here...

Ultimately, because this team is as good as it is...this probably won't affect us at all this year (difference between #1 overall and #1 playing Brooklyn and Boston is meaningless), so those posters (whom I generally respect but think are wrong on this) will be right for this year. And any year we're this good, it won't matter. But...the odds that we're 24-2 next year are pretty low.

 

[Waquoit]

I thought AI was supposed to remove that human input and develop “it’s” own eye test. If AI is learning that means everything it spits out is not based on just human input. As I understand it that is what makes it so dangerous. It starts to think on its own. Hey for me I am all for it because I believe the ncaa committee is unable to think for itself.

These algorithms aren't AI. In fact, true AI does not exist yet. What they call AI is just marketing of supercomputing.

 

[JohnFSilver]

Metrics are a tool to help separate teams that are of equal eye test. UConn is so ouch better than everyone else, I do t think the rankings should be justified to put UConn 2nd, third or fourth since the eye test clearly makes them superior.

 

[Tenspro2002]

I've been banging this drum all year and been told there's nothing to see here...

Ultimately, because this team is as good as it is...this probably won't affect us at all this year (difference between #1 overall and #1 playing Brooklyn and Boston is meaningless), so those posters (whom I generally respect but think are wrong on this) will be right for this year. And any year we're this good, it won't matter. But...the odds that we're 24-2 next year are pretty low.

DePaul being historically terrible was not accounted for. If we knew we would be playing true cupcake (#300-caliber) games in January and February, we probably would have scheduled differently.

 

[cohenzone]

The values assigned to the algorithms are still subjectively decided by humans. There is zero difference between a computer poll and an eye test for validity or truth.
You just substitute direct eye contact for a program which tells a computer to compile a list based on categories and values you assign to those categories.

I think that’s what he said in your quote. But whatever. Assigning a lower value because your really good team didn’t nuke versus merely flatten a really bad team is a distinction only a computer system would do and some BY fans.

 

[uh-no]

I think that’s what he said in your quote. But whatever. Assigning a lower value because your really good team didn’t nuke versus merely flatten a really bad team is a distinction only a computer system would do and some BY fans.

efficiency systems, such as NET, KP, whatever, almost ALWAYS devalue

• points margins past a certain gap. There's no predictive value of a 30 vs 40 point win.
• games between teams with large disparities. Uconn beating depaul should mean not a whole lot in the metrics since they are so differently rated. (now if defaul played CLOSER to uconn thn expected, it would)

So in the end, people are far overstating how much uconn's results against "really bad" teams have mattered. Far more impactful are things like:

• getting trounced by SHU
• only beating st john's by 4...at home
• only beating xavier by 5
• only beating providence by 9
• only beating butler by 9

It's not that those are bad wins/margins....but those are the places uconn could have done better to be atop NET.....not beating depaul by 50 instead of 40

 

[UConnRock]

efficiency systems, such as NET, KP, whatever, almost ALWAYS devalue

• points margins past a certain gap. There's no predictive value of a 30 vs 40 point win.
• games between teams with large disparities. Uconn beating depaul should mean not a whole lot in the metrics since they are so differently rated. (now if defaul played CLOSER to uconn thn expected, it would)

So in the end, people are far overstating how much uconn's results against "really bad" teams have mattered. Far more impactful are things like:

• getting trounced by SHU
• only beating st john's by 4...at home
• only beating xavier by 5
• only beating providence by 9
• only beating butler by 9

It's not that those are bad wins/margins....but those are the places uconn could have done better to be atop NET.....not beating depaul by 50 instead of 40

Every team has a handful of games that are closer than expected. Beyond Houston’s 3 losses, Houston has had close games with Xavier, A&M, Cincy, OT against Texas. I also don’t look at 9 point conference wins to be negatives

 

[uh-no]

I also don’t look at 9 point conference wins to be negatives

Yes, but data has shown that such margins are predictive in value. IF we want to be #1 in efficiency metrics, we need to do better in some of those games. There's no conspiracy here.

I can't speak for NET since it has all sorts of weird crap in it, but there are two major things that repress uconn's efficiency metrics WRT houston:
1) houston's SoS is about a point better. That's more than half the difference between our two ratings
2) despite both teams playing at a glacial pace overall, houston is about a possession slower than uconn, meaning if uconn and houston were to both beat the same team by, say, 10 points, houston would get more credit for having built that 10 point margin in fewer possessions.

These two things almost surely explain effectively the entirety of the difference in raw efficiency.

This is all academic, though. Efficiency metrics aren't be-all-end-all....and naturally uconn looks by far the best of the top teams right now. All it does is tell us is that houston, uconn, purdue, and arizona are pretty much head and shoulders above the rest, and despite some of the musings that we'll beat them easily (which we might), they're apt to put up a good fight. Splitting hairs between 1 or 4 is a bit silly, IMO.

 

[Waquoit]

Metrics are a tool to help separate teams that are of equal eye test. UConn is so ouch better than everyone else, I do t think the rankings should be justified to put UConn 2nd, third or fourth since the eye test clearly makes them superior.

Purdue loses to a team with a rookie coach, pretty much were handled the entire game. But Purdue fan is still touting resume because we have more Q4 games and Purdue has more Q3 games. Say what?

 

[cohenzone]

efficiency systems, such as NET, KP, whatever, almost ALWAYS devalue

• points margins past a certain gap. There's no predictive value of a 30 vs 40 point win.
• games between teams with large disparities. Uconn beating depaul should mean not a whole lot in the metrics since they are so differently rated. (now if defaul played CLOSER to uconn thn expected, it would)

So in the end, people are far overstating how much uconn's results against "really bad" teams have mattered. Far more impactful are things like:

• getting trounced by SHU
• only beating st john's by 4...at home
• only beating xavier by 5
• only beating providence by 9
• only beating butler by 9

It's not that those are bad wins/margins....but those are the places uconn could have done better to be atop NET.....not beating depaul by 50 instead of 40

Right. And none of that means a thing game by game. Beating Marquette by almost 40

efficiency systems, such as NET, KP, whatever, almost ALWAYS devalue

• points margins past a certain gap. There's no predictive value of a 30 vs 40 point win.
• games between teams with large disparities. Uconn beating depaul should mean not a whole lot in the metrics since they are so differently rated. (now if defaul played CLOSER to uconn thn expected, it would)

So in the end, people are far overstating how much uconn's results against "really bad" teams have mattered. Far more impactful are things like:

• getting trounced by SHU
• only beating st john's by 4...at home
• only beating xavier by 5
• only beating providence by 9
• only beating butler by 9

It's not that those are bad wins/margins....but those are the places uconn could have done better to be atop NET.....not beating depaul by 50 instead of 40

it’s all just nipping around the margins in some imprecise attempt to be more or less more precise. In the end, it isn’t all that much better than the eye test. But it does provide fodder for sports boards.

 

[uh-no]

Right. And none of that means a thing game by game. Beating Marquette by almost 40

it’s all just nipping around the margins in some imprecise attempt to be more or less more precise. In the end, it isn’t all that much better than the eye test. But it does provide fodder for sports boards.

Sure. Eye test is usually good....but it's also often biased in cases where results haven't tracked with how well a team has played. If a team loses by a couple to a bunch of top teams, many people will use the "eye test" and say they aren't good....but in reality, they're probably a sleeper.

Consider last year's uconn team. COnventional eye test wisdom was that uconn was cooked after their woeful league start. Analytics never dropped them less than 6th. That's the kind of thing it helps with....allowing decoupling of game to game variance inherent in any random-ish event (not uniform, but there is still a "random" variable to some degree), from human intuition and recency bias.

Neither is perfect, and they make good checks on eachother. On the one hand, uconn is likely not 2 points worse than houston as the metrics would indicate, but also, kentucky was never the top 6 team the polls pinned them as (metrics never even having them top 15).

 

[cohenzone]

Sure. Eye test is usually good....but it's also often biased in cases where results haven't tracked with how well a team has played. If a team loses by a couple to a bunch of top teams, many people will use the "eye test" and say they aren't good....but in reality, they're probably a sleeper.

Consider last year's uconn team. COnventional eye test wisdom was that uconn was cooked after their woeful league start. Analytics never dropped them less than 6th. That's the kind of thing it helps with....allowing decoupling of game to game variance inherent in any random-ish event (not uniform, but there is still a "random" variable to some degree), from human intuition and recency bias.

Neither is perfect, and they make good checks on eachother. On the one hand, uconn is likely not 2 points worse than houston as the metrics would indicate, but also, kentucky was never the top 6 team the polls pinned them as (metrics never even having them top 15).

It would be interesting to know what NET would have been for the Napier NC team.

 

[AZHuskiePop]

Sorry for the scrap. My computer tends to push its own buttons.

Here's the rest.

Well, we're still 4th this morning in the NET rankings. As far as yesterday's 28-point romp over a top Quad1 opponent, move along, nothing to see here.

Those relying on the eye test to rank teams are urged to get the eyes examined.

Of course, the game was at Gampel, or at least as far as the NET is concerned, it was. Home court is home court.

Except when you travel 3000 miles to Seattle to beat the pants off a Quad 1 Gonzaga team. Then home court for the opponent becomes a neutral court.

Oh, well.

As my old buddy use to say “ figures don’t lie but liers fo figure”
As an old Stat guy who was accused of excess creativity I never trust a result without examining the raw data .

 

[uh-no]

It would be interesting to know what NET would have been for the Napier NC team.

NET ends up tracking decently close to other efficiency metrics such as KP, and uconn was 25 entering the tournament. I don't think there's any metrics gonna explain that team...other than "defense which went from pretty good to crushing after nearly getting bounced by st joes"

 

[cohenzone]

NET ends up tracking decently close to other efficiency metrics such as KP, and uconn was 25 entering the tournament. I don't think there's any metrics gonna explain that team...other than "defense which went from pretty good to crushing after nearly getting bounced by st joes"

They didn’t meet many, if any, eye tests either. Erratic as heck and boom, they jell.

 

[tzznandrew]

DePaul being historically terrible was not accounted for. If we knew we would be playing true cupcake (#300-caliber) games in January and February, we probably would have scheduled differently.

That's 2 out of 9. Look at all the other top NET teams. Houston has 7 Q4 games. BYU and Iowa State have been excoriated for having 9 like we do.

Top 10 games vs. Q4:
Houston: 7
Purdue: 3
Arizona: 2
UConn: 9
Alabama: 3
Tennessee: 5
Auburn: 5
Iowa State: 9
UNC: 5
BYU: 9

You have to get to 15 St. Mary's (in a real mid-major) before you find another team with as many Q4 games as us (they're 10-0). And then its 23 Gonzaga (11-0). Then it's 32 Indiana State (9-1). Then it's 46 Drake (9-0).

In the NET top 50, only 4 teams played more Q4 games, and only 7 played as many or more.

 

[uh-no]

That's 2 out of 9. Look at all the other top NET teams. Houston has 7 Q4 games. BYU and Iowa State have been excoriated for having 9 like we do.

Top 10 games vs. Q4:
Houston: 7
Purdue: 3
Arizona: 2
UConn: 9
Alabama: 3
Tennessee: 5
Auburn: 5
Iowa State: 9
UNC: 5
BYU: 9

You have to get to 15 St. Mary's (in a real mid-major) before you find another team with as many Q4 games as us (they're 10-0). And then its 23 Gonzaga (11-0). Then it's 32 Indiana State (9-1). Then it's 46 Drake (9-0).

In the NET top 50, only 4 teams played more Q4 games, and only 7 played as many or more.

which Q a team falls in is an output of NET. not an input.

 

[tzznandrew]

which Q a team falls in is an output of NET. not an input.

I know. And? We played 9 Q4 games which is among the most in the NET top 50, and while it may not burn a 24-2 monster, going forward we may want to consider (for fan interest and seeding) not having as many games against Q4 teams as programs like Drake.