NET and KenPom Tracker 23-24 (SU = 84 3/9/24)

[bnoro]

If you don't understand statistics and methodology, then just don't come into this thread.

It's almost impossible to build a perfect system to rank over 350 teams that don't have common opponents and play at different levels of competition.

There's nowhere that says you have to be a top 50 NET team to make the tournament. It is simply a tool to compare teams with similar resumes if you can't decide who makes a few spots.

It's also currently incomplete data, so there really isn't any sense dwelling on it until we get to the last few games of the season.

Win games, and it takes care of itself.

 

[orangefog]

If you don't understand statistics and methodology, then just don't come into this thread.

It's almost impossible to build a perfect system to rank over 350 teams that don't have common opponents and play at different levels of competition.

There's nowhere that says you have to be a top 50 NET team to make the tournament. It is simply a tool to compare teams with similar resumes if you can't decide who makes a few spots.

It's also currently incomplete data, so there really isn't any sense dwelling on it until we get to the last few games of the season.

Win games, and it takes care of itself.

Wait, we don't HAVE to respond to all threads?

 

[Jzsshuttles]

Tell us where Syracuse should be*. All of the smarty pants on here who think that NET and KP and any other ranking site is fundamentally flawed at best or complete and utter trash at worst should be able to hammer our a consensus top 100 in about 10 minutes. GO!




*That seems to be the most important thing here. Since if SU was ranked 10th no one would be complaining.

You’re acting like a nympher..be better then the bobber bro cmon

 

[Orangezoo]

If you don't understand statistics and methodology, then just don't come into this thread.

It's almost impossible to build a perfect system to rank over 350 teams that don't have common opponents and play at different levels of competition.

There's nowhere that says you have to be a top 50 NET team to make the tournament. It is simply a tool to compare teams with similar resumes if you can't decide who makes a few spots.

It's also currently incomplete data, so there really isn't any sense dwelling on it until we get to the last few games of the season.

Win games, and it takes care of itself.

Haven’t read all responses but will say it’s a flawed model and always will be with an inability to find an ideal balance in how some variables are weighted.

It wouldn’t hurt however to find a better way to represent margin of victory and balance it a bit more in terms of comparing against the modeled expectation. All that said knowing the conference warfare will eventual result in something totally different and thus it all changes anyway.

 

[dasher]

No, it's not. You and the rest of the luddites don't understand what any of it is used for or how it works and refuse to do so because your favorite team is not being ranked (though I think "rewarded" is the more apt term you desire) the way you think they should.

jncuse has done a tremendous job explaining how these systems work and what they are used for, but it has obviously gone over your head and the other people on here. I'm sorry these systems don't match up with how you think they should. The vast NCAA conspiracy against Boeheim and Syracuse is still here and now it has MATH in its clutches!

Still no one has even listed a top-25. C'mon this should be easy for you folks. Where should syrause be? Top 10? Top 30? 44?

Sure anything can be improved, but I'll tell you right now that KP, NET, and Evan Miya are all waaaaay better than anything else out there including three guys in a room in North Carolina. Revising any system that ranks 350 teams to make your favorite team rank differently screws up the rest of the rankings. That's what the "luck" column is for. It shows the variance.

We aren't a top 25 team. We aren't that good. But any system that has Pitt rated ahead of us, has flaws. Period.

 

[djcon57]

If you don't understand statistics and methodology, then just don't come into this thread.

 

[jncuse]

You seem to be the one whining about people pointing out inconsistencies IMO. That is all perspective though and its hard to tell tone through text, much is lost.

I don't think its an unrealistic expectation that statistic based algorithms would take constructive criticism and adjust to better represent what they are trying to show.

I think human beings forget that in the natural world (the one we live in) there are no numbers. There are proportions, things can be expressed in mathematical terms but there are no numbers.

I think math like many things can be a useful tool. That said these systems are simply numbers games. You input the numbers and they perform calculations based on programs humans devised. How is that different from humans sitting in a room looking at the results of the season and figuring out the last 4-10 spots that reasonable people could differ on. It is my opinion that the difference in the end is that the these numbers games simply add a layer complication and give those making the decisions a scapegoat. I prefer personal accountability. Reasonable minds can differ, it often comes down to different order of importance. That is what happens when selecting teams with or without all of these indicating algorithms. I prioritize the human element, you seem to prioritize the numbers.

But now I'll answer seriously.

Part #1: Before I start, as I have asked you a few times how would you have the committee select teams. Just give them the W-L of each team on a sheet and that's it? Now pick 36 at larges, based on what the committee has watched and whatever way they decipher records. The eye test and standard W-L records? What else -- perhaps I am missing something.

For those claiming it should be based on what people saw in the games... even if someone loves college basketball how much college basketball can someone reasonably watch.

And then we have teams from lower conferences that are pretty strong, could be at large contenders but are shown less - Princeton, Drake, Indiana St, Grand Canyon, SMU, St Mary's. Are you watching all these teams play?

And if you are watching all these teams play. How often are you watching them play? 3-4 times each.

If you watched Syracuse play against Oregon (as it was on National TV against nothing else), Miami, and Georgetown, you might say hey this is a good team. But if your sample was Tennessee, UNC, and Pitt you would think far differently about this team. Let's take math completely out of it, and talk about human sampling with eyes. Are human samples of 3-4 observations not problematic in themselves?

Again no reasonable human can watch a large amount of games, of EVERY team in the process.

Which leads me to the conclusion. Humans need some data/metrics to make decisions. This in every form of life. It doesn't mean those decisions are being taken over by computers. It just means humans are being provided help to make better subjective decisions.

Part 2 coming later today.

 

[jncuse]

I think you are moving the goal posts. People aren't arguing the score or the outcome. That is where numbers are a useful tool as I acknowledged in my post. It's the weight being given to what it being done to numbers and statics after words that I think is misguided.

Of course I was being facetious. I am given a longer response in parts. I don't think NET has nearly the weight you think in selecting teams. That will be in part 2 and part 3 of my response.

 

[JimBoston]

I think the metrics will start to level out. Right now, many teams have only played 3 or 4 road games (not neutral site) so far, but that is changing. And for schools in major conferences, they have more opportunities to improve their stats as their schedule gets tougher. For example, Clemson looks over rated due to their weak OOC schedule.

Remember how Q1 games are determined:

home games: ranked 1 to 30
neutral games: ranked 1 to 50
road games: ranked 1 to 75

Right now, Syracuse has only 3 Q1 (currently) remaining:

I still believe over the next 13 games:

10-3 Syracuse is in
9-4 bubble
8-5 out as that probably means a bad loss.

 

[jordoo]

If you don't understand statistics and methodology, then just don't come into this thread.

It's almost impossible to build a perfect system to rank over 350 teams that don't have common opponents and play at different levels of competition.

There's nowhere that says you have to be a top 50 NET team to make the tournament. It is simply a tool to compare teams with similar resumes if you can't decide who makes a few spots.

It's also currently incomplete data, so there really isn't any sense dwelling on it until we get to the last few games of the season.

Win games, and it takes care of itself.

You don't understand so get out of here.

Nothing is perfect is somehow a legitimate reason to not point out what could be improved.

So why is the NET needed then? If it was accurate enough to split hairs then it would be a good tool to differentiate between the last couple of teams. If it an imperfect system used to rank 350 teams, as you described, then it should not be relied on to determine a small difference between two teams.

Its a work in progress, the numbers will make more sense as teams continue to prove them wrong.

Winning cures all ills! (agreed)

 

[jordoo]

Of course I was being facetious. I am given a longer response in parts. I don't think NET has nearly the weight you think in selecting teams. That will be in part 2 and part 3 of my response.

Nice, I'll wait to read all 3 and try to respond later.

 

[upperdeck]

The issue isnt the ranking so much as mostly not understanding how teams that seem to be similar in many ways get so widely different rankings. Why do we have quad 1 vs 2 vs 3? why not just use the ranking in its entirety?

if team A plays team 30 ten times on the road and goes 3-7 and team B plays team 31 on the road 10 times and goes 7-3 for example? Math says 30-31 are not all that different but view thru this team A is gonna be seen as better and get credit for quad 1 wins and team B wont and be dinged not playing quad 1s at all.

 

[orangefog]

I request that an explanation of each metric (NET, Kenpom...) be presented in the form of a catchy song similar to the School House Rock series.

 

[Cstate]

I think the metrics will start to level out. Right now, many teams have only played 3 or 4 road games (not neutral site) so far, but that is changing. And for schools in major conferences, they have more opportunities to improve their stats as their schedule gets tougher. For example, Clemson looks over rated due to their weak OOC schedule.

Remember how Q1 games are determined:

home games: ranked 1 to 30
neutral games: ranked 1 to 50
road games: ranked 1 to 75

Right now, Syracuse has only 3 Q1 (currently) remaining:

I still believe over the next 13 games:

10-3 Syracuse is in
9-4 bubble
8-5 out as that probably means a bad loss.

Clemson beat 4 of the teams listed on Andy Katz’s power 36 list that just came out during their OCC schedule. I think you are basing your view of them based on their recent league loses due to extremely poor shooting nights (mostly JG3 and PJ Hall). However to say they look over rated I’m not so sure. They just went to FSU and beat them while they were 5-1 in league play.

 

[storange]

Nothing to do with computers or Quad whatever’s and all the other crap, but just to show how crazy this all is Katz has SU 36 ( courtesy of OE post ) and no Clemson and Palm has Clemson a 4 seed.

 

[Cusefan0307]

Nothing to do with computers or Quad whatever’s and all the other crap, but just to show how crazy this all is Katz has SU 36 ( courtesy of OE post ) and no Clemson and Palm has Clemson a 4 seed.

Katz is doing a power ranking. Palm picks teams like the selection committee would. Unfortunately Palm is closer to being accurate than Katz is.

 

[upperdeck]

one metric is how are you playing now vs one is how have you played all season.

Say OC you play all the top 10 teams and beat them all on the road by 30.. then in low level conf you go 4-14 and finish last. That would be an interesting test of NET.. You would have the most quad 1s in the country and be 14-14 with a good or bad schedule?

 

[storange]

Katz is doing a power ranking. Palm picks teams like the selection committee would. Unfortunately Palm is closer to being accurate than Katz is.

I’m aware of that, my point being s about perception.

 

[jncuse]

Re - Pitt. I'd rather understand the implications moving forward rather than whining about the flaws in Individual NET (which there certainly are). And Pitt is very important to us, because we are relying on 2 quality wins instead of 1. They are overrated, we are underrated. Yes? No? OK, what does it mean though going forward.

What NET gives you today because of its system flaws relating to margin elements, it will generally taketh at least part of it away because of its same system flaws relating to margin. This is a good thing for Syracuse, not so good for a team that has benefitted from it like Pitt. Its easier to offset the system flaws that hurt you for the next 14 games. Its hard to maintain the system flaws that benefitted you for the past 18 games.

The NET is measured at the end of 32 games not 18. Outlier games now become less relevant by the end (each game goes down from 6% to 3% of your possessions).

When you have been hurt by negative margin elements like Syracuse, if you just win moving forward and avoid blowouts in losses your NET will progress forward. If we go 8-5 to close, and don't get blown out in 3 of those games NET should move forward because those margin negatives go down in value

Pitt is the opposite, and its much more concerning for them (and us as Syracuse fans) The impact of its absurdly good play in cupcake games on NET, can only be maintained if they do similar stuff to close the ACC. They either have to step it up from a W-L perspective or they need to have big blowout wins when they win and only lose close ones. It was huge that they beat Duke, but our Q1 and Q2 win against Pitt is in serious jeopardy, because of the giveth and taketh away elements of this thing.

 

[jncuse]

The issue isnt the ranking so much as mostly not understanding how teams that seem to be similar in many ways get so widely different rankings. Why do we have quad 1 vs 2 vs 3? why not just use the ranking in its entirety?

if team A plays team 30 ten times on the road and goes 3-7 and team B plays team 31 on the road 10 times and goes 7-3 for example? Math says 30-31 are not all that different but view thru this team A is gonna be seen as better and get credit for quad 1 wins and team B wont and be dinged not playing quad 1s at all.

Generally the committee scrubs teams wins and losses closely for the final "x" spots as part of their subjective process. They are not going to overrate a win because its 49 vs 51, or shouldn't based on what they say. They have all weekend to subjectively scrub this stuff (I assume they are doing something all weekend) they are not going to go hard line on teams on the borders. You often also hear them use the term "Wins over fellow tourney teams or bubble teams"

They use quadrants as a guideline to help them decide who clearly get in, and they are also relevant of course when comparing bubble teams... but its not a hard fast this team is #49 its great win, this team is #51 its not a great win.

Someone could argue that offsets part of my point on Pitt above at that is fair enough. But I do the consequences of them falling out of the 75, and moving down to say 81 could be pretty severe on our resume.

 

[baggerbob]

Haven’t read all responses but will say it’s a flawed model and always will be with an inability to find an ideal balance in how some variables are weighted.

It wouldn’t hurt however to find a better way to represent margin of victory and balance it a bit more in terms of comparing against the modeled expectation. All that said knowing the conference warfare will eventual result in something totally different and thus it all changes anyway.

The biggest flaw in the system is they don't start each season with a clean slate, and go from there.
Especially with the transfer portal, most teams are a lot different then the previous year. There is no way to tell how a transfer fits into a new team without playing a few games.

 

[Fly Rodder]

We aren't a top 25 team. We aren't that good. But any system that has Pitt rated ahead of us, has flaws. Period.

These potatoes make terrible salsa.

This is more or less what people are posting on this thread.

 

[kirbivore]

Nate Oates is former math teacher and has figured out how to game the NET system.

 

[VikesCuse84]

JBR Bracketology just released his updated seed list has Syracuse as one of the last four byes as an 11 seed.

https://x.com/jbrbracketology/status/1749505819832353196?s=46&t=-qXhiwt3p299211b-AtyXA

 

[storange]

View attachment 236810JBR Bracketology just released his updated seed list has Syracuse as one of the last four byes as an 11 seed.

https://x.com/jbrbracketology/status/1749505819832353196?s=46&t=-qXhiwt3p299211b-AtyXA

I've always lived and died by JBR Bracketology, there the best !!!!