New NET Rankings

[BigMoeMusketeer]

Now it is obviously very early, but wow, these new rankings aren't doing the Big East any favors.

#29 St. John's
#36 Creighton
#38 Villanova
#45 Butler
#76 Georgetown
#78 DePaul
#85 Seton Hall
#99 Marquette
#101 Xavier
#115 Providence

Adds some credibility to those saying it is a 4-bid league, and also some to those of us (unfortunately) that think our Muskies will be missing the Big Dance in 2019.

Again, however, very early!

[bjf123]

This early, I’d venture to guess that this ranking is about as relevant as the preseason polls.


Sent from my iPhone using Tapatalk

[GoMuskies]

So the NCAA has found a rating system that is even MORE useless than the RPI? Impressive.

[noteggs]

https://www.cbssports.com/college-basketball/news/ncaas-new-net-rankings-debut-ohio-state-shockingly-no-1-with-duke-no-6-and-kentucky-no-61/

What do you mean? This wouldn’t be your top 25 with Kentucky being on the bubble?

[BigMoeMusketeer]

This early, I’d venture to guess that this ranking is about as relevant as the preseason polls.
Sent from my iPhone using Tapatalk

The only reason I think it is slightly more relevant than what you're suggesting, is that if there aren't significant jumps between now and the first of the year, you'll have 18 games against teams much lower ranked than you're used to, or than what you were expecting. Just like the RPI, while the non-conference was often dismissed as "too early", it set the tone greatly for what things looked like down the road. When the Big East would have 5-7 in the top 30 of the RPI, all of those conference games were far more valuable than they would have been if they were all 30-100.

I don't think this early ranking is reason for alarm, per se, but as a whole, the Big East needs to do a lot of good work over the next month to improve the opportunities for the league games January - March.

[drudy23]

It's the same as early season RPI rankings....they need time.

[GoMuskies]

It's the same as early season RPI rankings....they need time.

Yes, much like the RPI this will likely only be a moderately shitty rating system by March.

[X-band '01]

The Big East was 16-5 during the exempt bracketed tournaments (or 17-6 if you include the two nonbracketed games that Georgetown played). Xavier was the only Big East team to lose a pair of games.

In a vacuum, there's no shame in the Wisconsin and Auburn losses, but the San Diego State loss really hurts. There's little (if any) margin of error remaining in the noncon schedule.

Creighton and Seton Hall were pleasant surprises after their Gavitt Game losses. Even Marquette did okay with the Louisville and a somewhat respectable loss against Kansas.

[xukeith]

Remember the early or min 1990's when the NCAA released it's RPI and our X team was#1? Oh when it looks good, fans love it and spin statistics galore in our favor. But when it doesn't look so good, "This is Awful!"

Beauty is in the eye of the beholder. Let's cool off until late December. Root like crazy for every BE team and every X OOC opponent.

[GoMuskies]

I don't think it's awful because of where it ranks us. I think it's awful because it's clearly awful. Just look at the rankings. I think Nate Silver called it the worst list of schools in any ranking he's ever seen. Hard to argue.

[scoscox]

Every computer ranking is all over the place at the beginning of the year. The RPI has plenty of crazy early season rankings. My first reaction was to laugh and say these are ridiculous, but I'm sure they'll probably even out as the year goes on

[noteggs]

I agree for the most part with “even out” over course of the year (I hope), but the quote that concerns me is (especially the last sentence about quality opponent) “Another lingering issue is adjusted winning percentage, which seems to have overlap in the overall formula. Teams are getting double credit for its wins and losses, and taking some of the data accrued in the team value index's criteria. There is also the issue of the 1.4/1.0/0.6 weighted values and how those can't possibly represent difficulty depending on specific venues (specifically on the road). Bringing even more concern into adjusted winning percentage is the fact that quality of opponent isn't accounted for.”

As GO stated, it’s not about X and at this point I don’t see a huge improvement. Only time will tell.

[xukeith]

I don't think it's awful because of where it ranks us. I think it's awful because it's clearly awful. Just look at the rankings. I think Nate Silver called it the worst list of schools in any ranking he's ever seen. Hard to argue.

The rpi this early in the year also is awful.

[xukeith]

Louisville is 117! If that makes you feel better. It is WAAAYYYY too early for computer rankings. Just keep rooting for OOC and BE teams.

[GoMuskies]

The rpi this early in the year also is awful.

The RPI is awful in March, too. As this thing will be.

[xukeith]

The RPI is awful in March, too. As this thing will be.

So solution is any team with 2 votes in AP or USA today poll is an at large bid? Regular season winners get bids?
Stop using computers?
Only use ESPN "experts" or Fox "experts"?

[GoMuskies]

Or, you know, use good tools that are already in existence like KenPom and Sagarin.

[xukeith]

Agree.
https://kenpom.com/
has the best ratings of all the computer ratings.

[RyanblockXU]

Agree.
https://kenpom.com/
has the best ratings of all the computer ratings.

Kenpom is the best system IMO however they do include his own preseason rankings and for the first 2 months, those built in preseason numbers weigh pretty heavy into the ranking.


Kenpom list our offense at #24 in the country and our defense at #153

However when you look at another great site for college basketball stats: Sports reference CBB (https://www.sports-reference.com/cbb/) they show non adjusted offensive and defensive rank.


With out preseason Bias our offense is ranked #78 and our defense is ranked #327th.


those are pretty huge differences. One shows how we are Truly performing ( CBB REF) and the other shows how they believe we will ultimately perform for the whole season (kenPom)

[xubrew]

The issue with the KenPom rankings is that they really do not take merit into account, and really aren't supposed to be a measurement of merit. Ken Pomeroy was actually on the advisory committee for the NET, and he himself said that his rankings shouldn't be used as any sort of a base by the committee. He isn't attempting to rank teams by their accomplishments. He is trying to rank teams on current win probability.

The other issue with predictive metrics is that no one knows exactly how they're calculated, and the person publishing is never going to tell you. They also make changes to them throughout the semester to better yield the results they're looking for. As a for instance, if a study reveals that a team who shoots 80% from the line wins the game 75% of the time (I'm just making that up and know it probably isn't true), then someone who is doing a predictive ranking can adjust to that, and rank teams who shoot that well from the free throw line higher whether they're winning their games or not. And, it's not uncommon for them to make changes like that in the middle of the year if they think doing so will help the accuracy of their win probability.

As an interesting aside, Pomeroy also had a heavy influence on the quad system. Not so much the concept, but rather what teams should fit into which quads.

Anyway, as far as the NET is concerned, I think that they should absolutely release the formula, and think it's kind of ridiculous that they haven't. I get that it is very complex and very complicated, but so what? Their reason for not wanting to, at least what I've been made to understand, is that they don't like the idea of someone in the media calculating it for previous seasons and then criticizing how the selections may have been different. That's just complete nonsense because it's a ridiculous thing to be concerned about, but at the same time it makes perfect sense because the NCAA is oftentimes concerned about things that are ridiculous.

But, I do know some things about it that people who like KenPom should also like. It is a hybrid of a merit ranking and an predictive ranking. Unlike KenPom, it starts out at zero because there is no concern at all of being an accurate predictor or performance right now, but rather a reliable and fair ranking in March, but there is a predictive ranking variable to it. There is also an artificial intelligence (or machine learning, I think they're calling it) element to it. This will make it virtually impossible for even a math genius to re-engineer. Sagarin is brilliant (at least I think). Ken Pomeroy is also brilliant. But as brilliant as they both are, there is almost no way they'd be able to reverse engineer the other's ranking at any given time. The same will be true with the NET. Until they tell us, we just won't know.

Now, what the machine learning is supposed to do is identify outliers within the variables and adjust to it accordingly. For instance, if a team has a 28-2 record, but has played the 300th overall SOS, then it is going to know that the record is inflated and not put as much weight on it. (I just made that up, but am simply trying to explain what the machine learning component is supposed to do). If a team has a really high offensive and defensive effeciency rating, but has an average win/loss record, average SOS, and average number of quality wins, then it will identify the high efficiency ranking as an outlier and not weight it as much. (I know that it's almost impossible to have a high efficiency ranking and an average record, but again I'm just trying to think of examples).

So...we shall see. My gut tells me that it's probably better than the RPI. But, at the end of the day all it really does is decide how information will be sorted on a team sheet. So, maybe it really doesn't matter that much. I'm also somewhat familiar with some of the people on this year's committee, and none of them really strike me as numbers crunchers, so it may not end up mattering at all. At the end of the day, I think things are going to play out pretty much the way they always have. But, I guess we don't know for sure until it happens.

[xukeith]

I never had issues with the rpi and admired how they put analysis of 4 quadrants and more points for road wins.

[xubrew]

I never had issues with it either. As a power ranking it sucked, but since the committee never used it as a power ranking and just used it as a very basic and general rating I saw no problem with it.

[paulxu]

Net Rankings/Current RPI (Warren Nolan)

St Johns 33/15
Nova 37/73
Marquette 39/21
Creighton 47/20
Butler 49/51
Xavier 62/108
DePaul 70/77
Seton Hall 86/61
Georgetown 87/63
Providence 99/76

What might be of a concern, is Warren Nolan shows W/L vs RPI Group 1 (maybe that's the grouping from the team sheets)
The conference according to him is 3-12 in that comparison.
In fact, the conference only has 18 losses, so 12 of them came from Group 1.