Manipulating the NET

[jncuse]

not true though.. nothing says you have to have the same number of possessions within1.. it just tends to work out that way.

You have the ball you get an intentional/flagrant foul you get shots and the ball. is that one possession or 2?

Or you score and the other team doesnt inbound the ball, thats not a possession. or you cause a tie up and the other team fails to get it in, thats not a possession.

In theory a team could have zero possessions in a game and in some HS games it is skewed by 10+

If you don't inbound the ball its a turnover. That equals one ********* possession.

 

[upperdeck]

If you don't inbound the ball its a freaking turnover. That equals one ********* possession,. ****.

no, its not a possession. By rule. Which is why i sited the tie up rule.. when a tie up happens the change of possession changes the arrow.. If you don't inbound the ball the arrow doesn't change because the possession never happens until you inbound the ball. Inbounding is a turnover but not a possession.

 

[littlereg]

I was looking at some of those Big12 teams and a few of their resumes are awfully similar to our 2007 team that was penalized for scheduling a week OOC. Texas and Oklahoma specifically.

don’t worry though they beat eachother up so that means they’re great!!!

 

[Sudano]

You can try it and see what it gets you. I think it would be even worse. And then all of that can be blowed up once a team has a hot night and shoots 75% from three. A transitive property in college basketball means that SUNY-ESF could be better than UConn - we just can't know because they haven't played, but ESF has beaten teams that have beaten teams that beat UConn! Until there's a ~180 game season where every P6+ team plays every other team 2-3 times there'll always be tradeoffs and teams and fans whining about what's holding them back: We get screwed by the refs! Team W's coach was stealing signs! We were shorthanded! They shouldn't get credit for that win, Team Y had food poisoning! Team Z's guards hated each other ...etc ad nasueum.

The best way (so far) to measure a team's performance within the limited data we're stuck with is to measure how they performed against their opponents relative to how all of the other teams played them. You measure how a team scores vs how do they defend per possession versus opponents compared to other teams.

I don't care about NET in particular, but the idea behind the various efficiency metrics IS NOT to get precise and accurate results, it's about getting close enough to separate the most wheat from the chaff.

To add on to your point. It's not deciding who wins the national championship, it's deciding who are the top 68 teams and seed them to then play off for the national championship. Let's not overblow what this is ... It's treating the decision between 68 and 69 like it's the decision between 1 and 2+. (It's really lower seeds because P5 teams are higher)

 

[Red2003]

I'll give you a quick stat -- in 66 games against Q4 teams, the B12 was 39-23 against the spread (with 4 pushes) and there were many more massive covers than shortfalls. In fact the 6 teams highest in the B12 standing went 22-4 against the spread. (I started an an analysis in the gambling thread the other day - see link below) That's not a small sample to do really well in - its not only 5 games.

The easiest way, and the traditional way for a conference to benefit from NET or RPI, has simply been to play well out of conference as a group in terms of win and losses (obviously with schedule consideration). The B12 played really well in Q1 and Q2 games last year OOC, and avoided bad losses and they got the tourney seeding and selection love they earned. ACC failed in that regard in 2022, 2023 (both miserably) and 2024 (although to a lesser extent). It's hard to manipulate NET when you only win 73% of your Q3 games like the ACC did this year.

That being said something is clearly happening in the B12 this year. Their Q1+Q2 record as compared the SEC, ACC, MWC (as I had calculated earlier) did not stand out as much better-- like it did last year. They did excel in Q3 games (winning 96% of 26 games). But something had to happen for them to be doing so well in NET.

And its in those Q4 games where they are performing extremely well over a not so small sample of 66 games. Is it manipulation or simply being better in those Q4 games? Hard to tell.

But I can tell you there is one type of game that is easiest for a team to control its narrative margin wise -- especially good teams that find a way to go 22-4 in those games.

Sports Gambling Thread

3-1 last night (Group 1: 28-14, Group 2: 32-14) Plays for Tonight Illinois at Michigan (+3) Vermont at Binghamton (+7) Seattle at Stone Cold Steve Austin (-6) St Thomas at South Dakota (+6) Gonzaga at Pepperdine (+12.5) Regarding the top 25, the "Old-School" auto bet (Top 25 at home...

syracusefan.com

I assume that FSU/Louisville in particular tanked the whole conferences noncon early on.

 

[Orangezoo]

I assume that FSU/Louisville in particular tanked the whole conferences noncon early on.

The bottom 5 all helped.

 

[LoveShack]

not true though.. nothing says you have to have the same number of possessions within1.. it just tends to work out that way.

You have the ball you get an intentional/flagrant foul you get shots and the ball. is that one possession or 2?

Or you score and the other team doesnt inbound the ball, thats not a possession. or you cause a tie up and the other team fails to get it in, thats not a possession.

In theory a team could have zero possessions in a game and in some HS games it is skewed by 10+

Regarding possessions these box scores, usually posted by Orange79 (thank you by the way) in the game threads list possessions. Feel free to go back and check them. They are most always equal.

 

[Cusefan0307]

problem with the NET is you only get .6 credit for a home win where you get 1.4 for a road win. Too skewed

This was the same with the RPI. I can see .8 and 1.2, but come on that’s dumb.

 

[pwong]

This was the same with the RPI. I can see .8 and 1.2, but come on that’s dumb.

How is it ridiculous that road wins are weighed more than home wins? It’s much harder to win on the road than at home

 

[Orangezoo]

In case folks missed it just go look at the MOV numbers on the previous page. It’s blatantly obvious what’s driving the NET - even more than I would have expected honestly.

 

[Orangezoo]

This was the same with the RPI. I can see .8 and 1.2, but come on that’s dumb.

With that said it’s not making a huge difference when you look at Wake for example.

 

[Red2003]

it’s actually pretty simple. By using margin of victory as a weighted variable that’s capped, it reduced the weighting of the efficiency metrics that use MOV and thus even with that redundancy, kept the MOV impact somewhat controlled. Now since they removed the MOV as a stand alone variable the efficiency metrics have a higher weight and thus you see these impacts play out more.

In order for this to balance out they have to nerf some of the efficiency weighting in favor of wins and losses , road record etc.

Another reason for this is teams with more losses see a jump too given this is a pretty big spike in MOV and for efficiency. If we beat the Ville and win by 20 or more we can jump 5 spots or more most likely especially when our efficiency numbers would get a bigger boost than some others would. As JN pointed out as well road and higher margin victories combined get an even bigger boost.

Love it that you said nerf. Definitely you're a gamer ha. Definitely agree by the way.

 

[jncuse]

it’s actually pretty simple. By using margin of victory as a weighted variable that’s capped, it reduced the weighting of the efficiency metrics that use MOV and thus even with that redundancy, kept the MOV impact somewhat controlled. Now since they removed the MOV as a stand alone variable the efficiency metrics have a higher weight and thus you see these impacts play out more.

In order for this to balance out they have to nerf some of the efficiency weighting in favor of wins and losses , road record etc.

Another reason for this is teams with more losses see a jump too given this is a pretty big spike in MOV and for efficiency. If we beat the Ville and win by 20 or more we can jump 5 spots or more most likely especially when our efficiency numbers would get a bigger boost than some others would. As JN pointed out as well road and higher margin victories combined get an even bigger boost.

So is it official that they had a cap on MOV before this year, or a few years back?

Just interested. Missed any official confirmation or notice. Would certainly make sense as this was the first year that I have measured OOC performance by conference and it didn't equate with NET fully as expected.

 

[CorduroyG]

I dont wanna hear anyone defend or justify or try to explain NET ever again. A team that is 13-14 has a better NET than a team thats 19-10. Theres no defending that. Its a broken system.

And margin of victory should not decide anything.

 

[Cusefan0307]

How is it ridiculous that road wins are weighed more than home wins? It’s much harder to win on the road than at home

You think a road win should be weighed 133% more than a home win? That's absurd. I suggested .8 and 1.2 because that makes a home win worth 66% of a win and that is what cbb home winning percentage is since the turn of the century.

Some of you guys really get butt hurt and take it personally when any of these metrics get called out and I'm not even anti analytics or anti NET. I know how the formula works, but lets weigh a road win vs Louisville more than a home win vs UNC.

 

[Orangezoo]

So is it official that they had a cap on MOV before this year, or a few years back?

Just interested. Missed any official confirmation or notice. Would certainly make sense as this was the first year that I have measured OOC performance by conference and it didn't equate with NET fully as expected.

So I think it’s been a couple years that they removed it as a specific variable and thus capped it which by looking at some of the numbers I’m seeing with regard to MOV it’s allowing MOV to be a massive factor in the NET. Which of course by law of averages is going to keep the top at the top but once you get below 20 or so then it can get wild. I don’t see where other factors are playing a big enough role to where you run up the score and win home games by a good margin and minimize margin on the road and you’ve got yourself a solid Net even if your W/L are meh.

Back to your original question again- There was an article a couple years ago on the “new “ formula and removing the specific MoV with 10 pt cap that I recall but it wasn’t on any media blast or formal announcement I recall.

 

[jncuse]

I dont wanna hear anyone defend or justify or try to explain NET ever again. A team that is 13-14 has a better NET than a team thats 19-10. Theres no defending that. Its a broken system.

And margin of victory should not decide anything.

The key thing to remember though whether its RPI or NET, and no matter what they put out, neither are a summary or a score of your resume, nor are they an important consideration on a standalone basis.

There is a reason teams from the P6 with RPI'/NET's as low as the mid 30's and as high as the 70's, have got in the tournament. They look primarily at quality of wins / bad losses, and those can run fairly divergent of your NET/RPI.

 

[shandeezy7]

So is it official that they had a cap on MOV before this year, or a few years back?

Just interested. Missed any official confirmation or notice. Would certainly make sense as this was the first year that I have measured OOC performance by conference and it didn't equate with NET fully as expected.

Yes, they changed it in 2020. There were originally five standalone factors that went into NET ranking. One of them was margin of victory, which was capped at 10 points in the formula. There are now only two components of the NET Ranking:

The remaining factors include the Team Value Index (TVI), which is a result-based feature that rewards teams for beating quality opponents, particularly away from home, as well as an adjusted net efficiency rating.

So margin of victory was removed as a standalone factor, but instead of it being on its own and capped at 10 points, it's now baked into the net efficiency rating with absolutely zero controls on it.

DI men’s basketball committee announces change to NET in 2020-21

The NCAA Division I Men’s Basketball Committee announced a change to the NCAA Evaluation Tool metric, which debuted during the 2018-19 season and replaced the RPI as the primary sorting tool.

www.ncaa.com

This is the tweet explaining the NET Ranking system when it was initially unveiled (you have to click on it to see the whole graphic - thanks, Elon):

 

[Red2003]

I dont wanna hear anyone defend or justify or try to explain NET ever again. A team that is 13-14 has a better NET than a team thats 19-10. Theres no defending that. Its a broken system.

And margin of victory should not decide anything.

You should check out reddit. Every day there is a post on how it's being gamed. There are still many big12 and Mountain West fanbases that believe that it was much needed experience beating Lindenwood level teams 5 times by 50 points. 1 or 2 times not enough "experience" needed 5-6 sometimes 7 or 8 of those games

 

[jncuse]

You think a road win should be weighed 133% more than a home win? That's absurd. I suggested .8 and 1.2 because that makes a home win worth 66% of a win and that is what cbb home winning percentage is since the turn of the century.

Some of you guys really get butt hurt and take it personally when any of these metrics get called out and I'm not even anti analytics or anti NET. I know how the formula works, but lets weigh a road win vs Louisville more than a home win vs UNC.

There is probably (I would say is but I am only 98% confident in what I am writing below) an element of double counting going on when you get road wins.

Here is the problem - if you are going to go with a system that is primarily MOV as your baseline.

It's already giving you a 7 or 8 point "adjustment" for playing on the road instead of at home. Remember the home court advantage is somewhere between 3.5 to 4. So say you win on the road, you probably get a margin reward, and then you get a road win kicker.

I tracked the biggest movers in NET week over week, for about 3 weeks (which is possible with Warren Nolan since they always show the NET at end of prior week). One example is post #8 in the thread below.

Gonzaga +20 (2 Q4 Road Wins by 67 points)
Drake +20 (2 Q4 Wins, one on the road, won by 70 points)
Virginia +16 (2 Q2 Wins, one road, won by 17 points)
Colorado +14 (Q2 Win + Q4 Win, both home, won by 49 points)
Syracuse +12 (Q1+Q2 Win, one road, won by 14 points)
Providence +11 (Q4 Road Win, won by 38 points)
New Mexico +11 (Q1 Win + Q4 Road Win, won by 32 points)
St. Mary's +10 (Q1 Road Win at San Fran)

It's pretty clear that road wins are by far the biggest mover. Road wins should be valued more, but it seems the double counting really comes into play here.

An Early 2024 Bracket Matrix Follow (or other Tourney Stuff if you Want)

As of this morning per the matrix = 0 (which of course is right call) Before tip off next game = ?? It won't be high, but it will not be zero either. We have made clear progress this week in our resume with a Q1 and a Q2 win (I would very much disagree with the few people that put us in, but...

syracusefan.com

 

[Orangezoo]

So doing some math here - had we been closer in losses and better in cupcake wins and had a total net MOV between 100 and 150 we would be somewhere around 40 to 45 currently.

 

[jncuse]

You should check out reddit. Every day there is a post on how it's being gamed. There are still many big12 and Mountain West fanbases that believe that it was much needed experience beating Lindenwood level teams 5 times by 50 points. 1 or 2 times not enough "experience" needed 5-6 sometimes 7 or 8 of those games

The ACC played 45% of its OOC games against Q4 teams.
The MWC played 42% of its OOC games against Q4 teams.
So the ACC tried the cupcake game a bit more than the MWC.

The Big12 took it to another level and played 55% of its OOC games against Q4 teams. And probably told its team to run it up as much as they could based on the fact that the top 6 teams in the conference (as of now) were 22-4 against the spread in those games.

 

[jncuse]

I think you have to look and see point differential is bigger than road wins in these examples honestly. I see large MOV here first road wins second. If Road wins carried such weight then Wake would not be top 30 in the NET.

When you look at B12 teams you can see where their awful non conf (almost all home games) at least somewhat dilutes their large MOV advantages but not much.

The biggest factor is margin - but there is obviously a not so minimal factor for road wins as well, otherwise we would see some home blowouts on the list.

And they were always minimal representation of those (home blowouts) when I did they list. So it was the margin and then the double reward kicker for getting the win.

And if you are already using a margin system that inherently auto adjusts for playing on the road vs home, then its a problem.

 

[Orangezoo]

There is probably (I would say is but I am only 98% confident in what I am writing below) an element of double counting going on when you get road wins.

Here is the problem - if you are going to go with a system that is primarily MOV as your baseline.

It's already giving you a 7 or 8 point "adjustment" for playing on the road instead of at home. Remember the home court advantage is somewhere between 3.5 to 4. So say you win on the road, you probably get a margin reward, and then you get a road win kicker.

I tracked the biggest movers in NET week over week, for about 3 weeks (which is possible with Warren Nolan since they always show the NET at end of prior week). One example is post #8 in the thread below.

Gonzaga +20 (2 Q4 Road Wins by 67 points)
Drake +20 (2 Q4 Wins, one on the road, won by 70 points)
Virginia +16 (2 Q2 Wins, one road, won by 17 points)
Colorado +14 (Q2 Win + Q4 Win, both home, won by 49 points)
Syracuse +12 (Q1+Q2 Win, one road, won by 14 points)
Providence +11 (Q4 Road Win, won by 38 points)
New Mexico +11 (Q1 Win + Q4 Road Win, won by 32 points)
St. Mary's +10 (Q1 Road Win at San Fran)

It's pretty clear that road wins are by far the biggest mover. Road wins should be valued more, but it seems the double counting really comes into play here.

An Early 2024 Bracket Matrix Follow (or other Tourney Stuff if you Want)

As of this morning per the matrix = 0 (which of course is right call) Before tip off next game = ?? It won't be high, but it will not be zero either. We have made clear progress this week in our resume with a Q1 and a Q2 win (I would very much disagree with the few people that put us in, but...

syracusefan.com

Just re read this and yeah it’s doubling down. That said you don’t even need the road wins as much as you should given home MOV boosts get the job done in terms of overall Net ranking.

 

[Orangezoo]

The biggest factor is margin - but there is obviously a not so minimal factor for road wins as well, otherwise we would see some home blowouts on the list.

And they were always minimal representation of those when I did they list. So it was the margin and then the double reward kicker for getting the win.

Yeah I deleted after re reading your post saying the same thing essentially. So the kicker boosting the margin more gets you that higher overall margin gain likely inflating net MOV.