Thats exactly what the pinheads on here think. They just dont have the balls to say it . They need to be laughed off this stage.
NET Nonsense (Part-20)
- Thread starter ib4ky_rivals111873
- Start date
Thats exactly what the pinheads on here think. They just dont have the balls to say it . They need to be laughed off this stage.
How about this for context - Duke only had 3 losses before that, and at 5 losses still have one of the best win-loss records in the country. The two teams behind them have 8 losses and 6 losses. The fact that those two losses were recent means nothing to the NET.
If it is so clear, transparent and understandable - why don't you condescend to the underlings and quickly show us the math that keeps Duke at the same ranking after their last two terrible double digit losses? I'll even spot you a little wiggle room since no one has complete vision into the formula. Just do a quick dynamic statistical analysis of weighted averages and your assumptions that let's them hold their place. This is easy math after all man! Thanks!
Please keep it simple for us dumb folks - even use shapes and simple graphs as you decide what would best show the unintelligent oafs.
Anybody defending any part of the NET just looks stupid at this point. It's so incredibly flawed
Every time I consider quitting drinking, I read through a NET thread. It ain’t worth it. I’m usually pretty blitzed by the end of the second page.
Incorrect. You can't see the movement in the actual RATINGS, because they don't publish them. One of many legitimate complaints about the NET. But expecting that they MUST move down in the RANKINGS is naive and betrays an ignorance of how computer models work.
It is indeed flawed. Any ranking system - any computer model, any human poll, the selection committee's seeding - is flawed. NET in particular is definitely flawed and of questionable value. It's a proprietary formula, the criteria used and the weights given to them are seemingly arbitrary, model outputs are not published - only the order of the teams. There doesn't seem to be any validation of the model against any criteria, it's just thrown together and it is what it is.
But people also don't understand what it's meant to be, or how computer models work. For one, NET does not appear to be designed to be a power ranking or a predictive model, but rather a model of resume strength. People also don't seem to understand that it's not a simple ranking where the model's output is just "you're number 5 and they're number 6." This is in large part the NCAA's part for not publishing the model's rating output, but people don't seem to get that it's entirely possible for the gap between #5 and #6 to be as large as, say, the gap between #6 and #12. A team might be SO FAR AHEAD of the next team, that they can fall a lot in the rating but not fall in the RANKING. And it was very clear that Duke was way ahead in the rating. You saw it in OTHER computer models, and it makes subjective sense given that they had fewer losses than others, a relatively strong strength of schedule with few awful teams on the schedule, and a large number of blowouts propping up their numbers.
I know and we still just don’t understand it.
Or we do, and we think it’s a total meaningless crock of sht.
It’s always been useless. Academics deciding seeding based on faulty systems that can’t properly function and we wonder why there’s so many problems.
Perfect world?
Ncaa tournament is no more than 32 teams.
tournament selections and seeding is decided by former NBA HOF’s that played at least a year in college. They’d makeup a group of 16 or so. No NET, no RPI, No BPI.
Take the bodies of work and allow experts to handle it.
I don’t ask a group of English Majors to perform open heart surgery because they read the statistics.
Care to show the math? Not for every team - just Duke - and not for the whole season - just recent.
As someone who works with math, finance, forecasting and statistics daily - I don't take offense to being called ignorant - hopefully no more than you do for the tunnel vision, small sample way you view the NET.
In all serious tho.
It's 2020.
Forget NET. We know there's issues there. But I can't understand why we are really having the whole stats vs people that "know the game" argument. It's as if the Moneyball era in baseball never really happened.
Or this weird perception that if you are into the stats you obviously have not played the game.
Or even stranger the ones that think the people into the stats don't "watch the games".
It's old fashion thinking. Get over the stereotypes.
It's 2020.
Forget NET. We know there's issues there. But I can't understand why we are really having the whole stats vs people that "know the game" argument. It's as if the Moneyball era in baseball never really happened.
Or this weird perception that if you are into the stats you obviously have not played the game.
Or even stranger the ones that think the people into the stats don't "watch the games".
It's old fashion thinking. Get over the stereotypes.
OK, that's actually going to be pretty easy to do, in a way I'm pretty sure you'll be able to understand. Like you said, the formula is hidden so let's take a look at another computer system that uses efficiency metrics, KenPom.
On February 18, before the first of Duke's 2 losses, Duke's Adjusted Efficiency Margin was 27.64, which put them in 2nd place. The 5th place team was Baylor, with an AdjEM of 26.82. So, there was less than one point separating #2 and #5. The gap from #5 to #6 on Feb 18 was 2.73 - more than three times as large as the gap between #2 and #5.
After Duke's 2 losses, they've fallen to #5 in KenPom with their AdjEm now at 24.69 - they've lost 2.95 points off their AdjEM! But now remember, originally the gap between 5 and 6 was 2.73 points, so very close to the amount Duke fell in KenPom.
It's easy, then, to see the plausible scenario in NET where Duke could have fallen SIGNIFICANTLY in the model's rating, while still being just a little ahead of the team they were ahead of. Maybe instead of being ahead by 2.7 and falling 2.9, this model had them a little different and they were ahead by 2.8 and fell by 2.7. Again, this would be a LOT better if the NCAA was transparent and published their data, not just the ranking.
Not at all.
Research the academics that pick the selections and tell me how many played organized basketball in their lives. It matters, yes. That’s why seeding is so screwy, they don’t understand anything other than reading faulty systems because one thing they do understand in our culture of “only math matters” is math. So they lean on it and use it because they’re out of their element.
You playing sports, enjoying numbers, understanding the game ect isn’t what the problem is.
I didnt say it was clear, transparent, or understandable. Nice try tho. Welcome to 2020 when stating facts is considered condescending. Yall completely make up stuff about the NET then call people who correct you condescending. No one called you guys dumb or simpletons, people are just correcting the misconceptions and out right lies people are telling.
I've seen doubling down before , but in this thread we see the rare quintuple down. And still counting.
Duke's scoring margin is 15 points a game, they're 4th in the country. Our scoring margin is 9 points a game, we're 28th in the country in scoring margin. Capping it 10 hurts them more than us.
As if you are not speaking in generalizations? Justifying the NET algorithm with no math or proof? Justifying it with your blind faith and telling others to? Like that? Just give a simple example, mathematically, as to how this would actually work. That's all. Not offended and well acquainted with the year 2020. No one is making stuff up - in fact just asking to have one - only one - scenario mathematically justified (Duke). Nice try convincing without the math... but just a simple sample is welcomed.
i think you have said this after each Duke loss or something similar...and comparing losing in the FF losing late season to a sub .500 team...omg, did you really...?
I was pissed when I seen Florida at 31 today......1 more damn spot at 30 and we would have 7 quad 1 wins.
How stupid is it that 1 spot in the NET by Florida makes a huge difference in our profile.
Also we need Tennessee to win some games but they also play us which hopefully they lose but we need them to stay in the top 75 so we keep that quad 1 win. (And we get a quad 2 win if we beat them at home)
How stupid is it that 1 spot in the NET by Florida makes a huge difference in our profile.
Also we need Tennessee to win some games but they also play us which hopefully they lose but we need them to stay in the top 75 so we keep that quad 1 win. (And we get a quad 2 win if we beat them at home)
This is really the only example we need to show it's ridiculous. Not only did they not drop. They stayed at 5 or something really high.
Any system that allows such an outcome is beyond flawed. It's just completely broke.
once...accident
twice... coincidence
third loss with zero to little movement is a pattern of a ****** up system...
Look dude, I get it. You just want to rant about something you dont like and you aren't interested in understanding it or why it does what it does. That's pretty obvious, because somebody with a math/stats background is easily able to follow.
I think part of the problem is the range in the quads. Winning on the road against a team that might not make the NCAA Tournament shouldn't count as a Quad 1 win. I think 75 teams is just to many to be considered the best road wins. Here are the current quad wins below.
In my opinion, they should be something more like this.
Q1 1-20, 1-35, 1-50
Q2 21-50, 36-75, 51-100
Q3 51-100, 76-125, 101-150
Q4 101+, 126+, 151+
This would actually reward teams that play the best competition and not marginal teams that might not even be in the NCAA Tournament like the bottom of the Q1 road wins currently are. I also think Q4 should cover a lot larger range than it currently does. Some teams get away with playing mediocre teams but are rewarded because they play just the right ones. Teams in my Q4 range are bad teams and should be counted as such. I wouldn't be against adding a little bit to my numbers but the current ones should be changed.
For instance, beating Fresno State (9-18) is currently a Q3 home win while beating Oral Roberts (10-13) is currently a Q4 win in the NET. They are both really bad teams and you shouldn't be rewarded for playing one over the other one.
In my opinion, they should be something more like this.
Q1 1-20, 1-35, 1-50
Q2 21-50, 36-75, 51-100
Q3 51-100, 76-125, 101-150
Q4 101+, 126+, 151+
This would actually reward teams that play the best competition and not marginal teams that might not even be in the NCAA Tournament like the bottom of the Q1 road wins currently are. I also think Q4 should cover a lot larger range than it currently does. Some teams get away with playing mediocre teams but are rewarded because they play just the right ones. Teams in my Q4 range are bad teams and should be counted as such. I wouldn't be against adding a little bit to my numbers but the current ones should be changed.
For instance, beating Fresno State (9-18) is currently a Q3 home win while beating Oral Roberts (10-13) is currently a Q4 win in the NET. They are both really bad teams and you shouldn't be rewarded for playing one over the other one.
I follow - the example you gave - and I understand the gaps you stated - were for teams - no matter the gap in a single piece of the formula from KenPom (not the NET) that were three places apart - #2 and #5. In the NET Duke example the teams are basically 50 places apart and 100 places apart respectively from Duke... and Duke moved NONE. That is a problem. Maybe with the NET - Maybe with a bias (intended or unintended) in the NET... Still it is statistically VERY improbable if not impossible. That's all.
Whether the NET is a static formula or dynamic formula - it is still virtually impossible - A static formula is easily isolated - a dynamic formula could/should swing wildly with teams far apart - less so as the sample grows - but still large differences create most movement - key is movement - which Duke has not done.
Agree to disagree - best to you. Either way in my book Duke Sucks, is protected and is a fraud.
Last edited:
But what about the ones who follow it and understand it and believe it’s trash? The NET is a great example of why these systems do not resemble the games.
I have plenty of complaints about it myself, and have articulated them here.
Seems you also "forgot" to address the other points that show the holes in your analysis. Convenient.
None of those seem that odd on the surface. I would expect SFA to move more than Florida or Marquette after a win simply because they are closer to the mean of teams. There are 353 D1 teams so the closer your ranking is to 176 the more you can expect to see movement. The further you get from being ranked 176 the harder it will probably be to change your ranking assuming CBB teams are normally distributed.
According to Sagarin predictor, UK would be favorites by about 5 against the 25th ranked team, Wisconsin, @Rupp. While against the 75th ranked team, Yale, UK would be about a 5 point favorite @Yale. Despite being ranked 50 spots apart, the games are of roughly equal difficulty due to the effect of homecourt advantage.
Winning on the road is hard. Road games against nontournament teams are often just as, or even more difficult, than home games against top 25 teams.
Just an observation.......................................
Team-A is one spot higher than Team-B in the latest NET Rankings.
Team-A is in last place in their Power conference.
Team-B is in first place in their Power conference.
Team-A is the NET tenth rated team in their conference.
Team-B is the NET ninth rated team in their conference.
Team-A has won two of their last twelve games.
Team-B has won ten of their last twelve games.
Team-A has beat three teams in top-30.
Team-B has beat three teams in top-25.
Team-A season W/L Pct .500
Team-B season W/L Pct .621
Team-A: DePaul
Team-B: UCLA
Team-A is one spot higher than Team-B in the latest NET Rankings.
Team-A is in last place in their Power conference.
Team-B is in first place in their Power conference.
Team-A is the NET tenth rated team in their conference.
Team-B is the NET ninth rated team in their conference.
Team-A has won two of their last twelve games.
Team-B has won ten of their last twelve games.
Team-A has beat three teams in top-30.
Team-B has beat three teams in top-25.
Team-A season W/L Pct .500
Team-B season W/L Pct .621
Team-A: DePaul
Team-B: UCLA
Team B (UCLA) also has losses to 123 Hofstra and 267 Cal St. Fullerton (worse loss than Evansville btw) while Team A’s (DePaul) worst loss is Buffalo at 139. Seems like the real outlier is the fact that UCLA is in first in their conference but 9th in NET in their conference which is odd. You also forgot to mention that while UCLA has won 10/12 they lost 6/7 right before that