UDPride Discussion Forums    
     

Go Back   UDPride Discussion Forums

» Log in
User Name:

Password:

Not a member yet?
Register Now!
» Advertisement
UDPride Discussion Forums

UDPride Discussion Forums (http://www.udpride.com/forums/index.php)
-   Mens Basketball (http://www.udpride.com/forums/forumdisplay.php?f=3)
-   -   The RPI is Dead. But is the NET a Slam Dunk? (http://www.udpride.com/forums/showthread.php?t=32297)

Chris R 08-22-2018 02:35 PM

The RPI is Dead. But is the NET a Slam Dunk?
 
New Article: http://www.udpride.com/forums/showthread.php?t=32296

Some of my thoughts on the death of the RPI and creation of the NET. I have a lot of fear that we're getting into dark areas where those being evaluated by more and more complicated arithmetic have no means to fully audit and fact-check the tools they are being scrutinized with.

Let's see if the NCAA has more to say. Because more needs to be said.

Alberto Strasse 08-22-2018 03:02 PM

Does the NCAA have a Financial Interest in Favoring
 
the power 5 football schools in the NCAA basketball tournament?

CE80 08-22-2018 03:16 PM

To me the biggest flaw with picking the at large, and seeding the NCAAT teams is the how the disparity of opportunities for "good" wins between teams is handled. More opportunities should yield more wins and the bar should be higher than a team with less opportunities for "good" wins. Will NET change that? I doubt it and therefore the more things change the more they will stay the same.

N2663R 08-22-2018 04:48 PM

"Does the NCAA have a financial interest in favoring"
Quote:

Originally Posted by Alberto Strasse (Post 556953)
the power 5 football schools in the NCAA basketball tournament?

Probably not as much as the networks who are paying for the broadcasting rights. I'll bet they "influence" the NCAA to have as many marquee names as possible in the tournament. They also hope for 1-2 "Chicago-Loyola" teams to build intrigue. They certainly don't want a lot of Bowling Green's - Weber State's - Eastern Kentucky's - Richmond's clogging up prime time matchups.

NJFlyr71 08-22-2018 09:06 PM

So far I haven't seen any evidence of Russian tabulated point adjustments to be included into the NET. However, any conservative leaning schools (don't worry UD an't one of them) will be unceremoniously discarded from consideration.

Now back to shoe contracts and the ESPN power poll. :rolleyes:

NJFlyr71 08-22-2018 09:09 PM

Quote:

Originally Posted by N2663R (Post 556961)
"Does the NCAA have a financial interest in favoring"

anything that brings in over $1B like in BILLION dollars? Hell Yea!!!

And twice on Sunday! Except during NFL FB games.

Brad S. 08-22-2018 09:54 PM

Quote:

Originally Posted by CE80 (Post 556955)
To me the biggest flaw with picking the at large, and seeding the NCAAT teams is the how the disparity of opportunities for "good" wins between teams is handled. More opportunities should yield more wins and the bar should be higher than a team with less opportunities for "good" wins. Will NET change that? I doubt it and therefore the more things change the more they will stay the same.

Unless they do away with the selection committee, any ranking tool developed is more of the same.

Either seed the tournament by your tool, or develop a better one.

Every sport, every tournament...

Furio 08-22-2018 11:20 PM

Gets rid of those pesky sites like DanceCard pointing out the glaring inconsistences in their selections. Secret Skynet computer model. Lol

Chris R 08-22-2018 11:40 PM

The money at stake -- millions and literally billions -- is what drives my insistence on transparency in this entire redux of the RPI to the NET. Im not sure I have ever seen ESPN's Power Index formula but it might be out there. I know I have never see Sagarin's. He has all kinds of variations including Elo Chess, etc. Ken Pomeroy may publish his offensive and defensive efficiency formulas somewhere in the corners of his web site, but I have not scoured for them. If all of these third-party tools now being used OFFICIALLY in conjunction or as an insertion into the NET, all of them must come clean on arithmetic if they have not already done so.

There is too much money involved for there to be any secrets among the number crunchers that ask all member NCAA schools to take their fairness and objectivity at face value. The Power-5 schools, the conference TV networks, and CBS Sports are in this to control the chess pieces as much as possible and make profits hand over fist. Profit motive oftentimes makes otherwise good people consciously or subconsciously slant the playing field of fair play in ways that are hard to detect above the waterline.

And since all those participating in the NCAA tourney are NCAA members -- many of them funded by public taxpayers -- they have a right to act as auditors of the auditors of objectivity. Without that check-and-balance, the speculation of corruption will collapse the confidence the system is attempting to provide.

We already know the NCAA makes occasional calculation mistakes because we've had to correct them on it several times over the years. I presume the other third party metrics are not infallible and make occasional mistakes too. If nothing else, openness ensures we can correct the obvious oversights like these. But it also guarantees there's no deep-state collusion or favoritism or bias baked into the cake. With the kind of money schools are capable of making with deep NCAA tourney runs, brand awareness, and merchandising, there's no room for speculation of any kind. If the NCAA more or less stays hush on all of these metrics, its like admitting facts not in evidence into the courtroom.

We want to see the arithmetic formulas. All of them. You told us what you're doing, now tell us how you're doing it.

Bat'71 08-23-2018 09:34 AM

IMHO, the NCAA knows it has a "perception of fairness" problem and hence this attempt to provide some added hocus-pocus to their calculations to justify the inclusion of up to nine teams from Power 5 Conferences. This fig leaf is not the answer, however. Given that all the Conferences now have their own tournaments at the end of each season that also rake in millions of greenbacks, the NCAA should eliminate all but the top three to five teams from consideration for the Big Dance from any and all Conferences. This is the only way, IMHO, to keep the Big Boyz honest...or as honest as they could be given the amount of money involved. Three Card Monte shysters have nothing on the NCAA and it's "Committee". So we know it ain't gonna happen in our lifetimes.

cj 08-23-2018 01:35 PM

It would be nice to compare the NET with the old RPI over the next few years to see where they differ and which one is the most "accurate" predictor.

MrFlyerFanatic 08-23-2018 03:27 PM

I find it strange that the NCAA claims to have no intention of populating last year's results into their new toy.
I'd be willing to bet that it's already been done during the development process. I wonder why they don't want that information disseminated?

Chris R 08-23-2018 05:16 PM

In rummaging the Internet today to try and find more information on the NET, I found a few sources that indicated the following. Did not verify if these are true:


1. The NET will be its own calculation and will not roll into its calculation any third party calculations provided by others, though third party metrics will still be used as supplements to the NET in the War Room.


2. There are about five different metrics (Levels) that will be used in the calculation. A couple are obvious holdovers like W/L Record, and SOS. But there are no specifics on how anything beyond W/L record is to be calculated.


3. The other metrics like offensive and defensive efficiency et al will also be calculated internally by the NCAA and in some fashion with the help of Google. Again however, no specifics on what the darn calculation is. Without the arithmetic we have no idea what the numbers will mean.


4. No word on what the weight of each "Level" will carry either. The RPI was 25% W/L, 50% Opponents' Record, and 25% Opponents' Opponents' Record. But how much will W/L carry in the future? Or offensive efficiency? Or SOS? Or home/neutral/away factors? No one has any idea.


It's like commanding the Theory of Relativity from the mountaintop without coughing up E=MC2. These are not small details.

The Fly 08-23-2018 05:39 PM

If the NCAA thinks this will eliminate the debate and controversies that come now with nearly every at-large bid, they’re nuts. Transparency is everything. If they think this represents a fair and improved method of selecting and seeding teams, than share the calculation details. Otherwise, it just reinforces the perception that the Selection Committee is a cabal, driven by money and the P5 ... which, of course, it is. But at least until now we could see and understand the various formulas they used (or ignored, as the case may be).

UD62 08-23-2018 05:41 PM

I wonder of this new metric will be posted and updated as the RPI is/was. Whatever is used, it should be available to the general public on an ongoing basis.

Sea Bass 08-23-2018 07:44 PM

Quote:

Originally Posted by cj (Post 557016)
It would be nice to compare the NET with the old RPI over the next few years to see where they differ and which one is the most "accurate" predictor.

they could easily show the past few years. They have chosen not to do that for whatever reason but I am skeptical.

OSU Flyer 08-23-2018 08:08 PM

Quote:

Originally Posted by Chris R (Post 557030)
In rummaging the Internet today to try and find more information on the NET, I found a few sources that indicated the following. Did not verify if these are true:


1. The NET will be its own calculation and will not roll into its calculation any third party calculations provided by others, though third party metrics will still be used as supplements to the NET in the War Room.


2. There are about five different metrics (Levels) that will be used in the calculation. A couple are obvious holdovers like W/L Record, and SOS. But there are no specifics on how anything beyond W/L record is to be calculated.


3. The other metrics like offensive and defensive efficiency et al will also be calculated internally by the NCAA and in some fashion with the help of Google. Again however, no specifics on what the darn calculation is. Without the arithmetic we have no idea what the numbers will mean.


4. No word on what the weight of each "Level" will carry either. The RPI was 25% W/L, 50% Opponents' Record, and 25% Opponents' Opponents' Record. But how much will W/L carry in the future? Or offensive efficiency? Or SOS? Or home/neutral/away factors? No one has any idea.


It's like commanding the Theory of Relativity from the mountaintop without coughing up E=MC2. These are not small details.

So you have no way of replicating NET like you do for the RPI?

FlyingArrow 08-24-2018 10:56 AM

I very strongly believe that margin of victory should not be used at all to decide who is chosen. A berth to the NCAA tournament should be a reward based on performance: who you beat and who beat you. "A win is a win" is no longer true if margin of victory plays *any* role in a metric that is used for selection.

Margin of victory does in fact help you better predict future performance, but that's not the point. NCAA selection should be a reward for past performance (measured only in wins and losses), not based on the prediction of future performance. An undefeated team that won every single game by 1 or 2 points absolutely deserves to be ranked above a team that has 3 1-point losses and 27 30-point blowouts.

hawkoooo 08-24-2018 11:48 AM

The real question is, will the RPI still be updated daily by this site? I hope so, because I visit it religiously.

jack72 08-24-2018 02:10 PM

You can be sure that they will tie it into some big revelation show on ESPN.

Chris R 08-24-2018 11:06 PM

May keep the RPI around just to track against the NET. But can't calculate the NET until the formula is published.

Lifelong Flyer Fan 09-04-2018 04:16 PM

https://www.mydaytondailynews.com/sp...tYHB2BoakgciN/

Neil quotes
“In general, I fully support modernizing the evaluation tools to use all the best available resources we have,” Sullivan said. “You think how far data and analytics have come since the RPI was introduced, I think, in the early 80s. It’s probably the right direction to head. I would say it’s a little bit challenging in that it’s the second year in a row the sorting mechanism has kind of changed after the schedules are complete.”

The fact that is true under any metric and what we understand is we have to win games against NCAA tournament caliber teams and compete for an A-10 championship,” Sullivan said. “No matter what the metric is, we know that’s true, and that fact will always remain. We accept that challenge and acknowledge that. I just want to make sure we have the opportunity to meet that challenge because when you look at the data, clearly the committee believes that all conferences are not created equal.”

CE80 09-04-2018 04:28 PM

Quote:

Originally Posted by Lifelong Flyer Fan (Post 557578)
clearly the committee believes that all conferences are not created equal

Isn't his what it's all about? Too much of the data is the result of incestuous conference competition and with 20 game conference schedules, it is getting worse.

FlyingArrow 09-05-2018 03:26 PM

If they used something that is close to being statistically valid scheduling difficulties really wouldn't matter that much. But when they do a raw count of "good wins" without counting how many chances a team gets to obtain those good wins, it's ridiculously biased.

TA111 09-05-2018 03:40 PM

[QUOTE=FlyingArrow;557620]If they used something that is close to being statistically valid scheduling difficulties really wouldn't matter that much. But when they do a raw count of "good wins" without counting how many chances a team gets to obtain those good wins, it's ridiculously biased.[/QUO

Agreed, you really need to look at the winning percentage of “good wins”. IMO a 3-1 record in games against “good teams” should be rewarded before a 5-9 record.

N2663R 10-29-2018 01:16 PM

Chris -

Are you able to post both the RPI link as you have in the past along with the new NET formula rankings so we can follow along and see how it affects the ranking for us and all teams??

Sea Bass 10-29-2018 02:43 PM

It won't mean anything unless they release the code. They are saying trust us its good but they don't deserve the trust. The RPI wasn't perfect but it was in the open. This .NET thing is behind closed doors.

I'm highly skeptical.

NJFlyr71 10-29-2018 03:55 PM

I cracked the code just a short time ago!!!!!!!!!!!!!!!

Actually it was fairly easy and not much to it. Here's how it works:

NCAA Tournament:

Conference Champion Winners - One each
At Large - 36 men's slot available are doled out in sequence

non-P5 one for you
P5 schools - 11 for you

non-P5 one for you
P5 schools - 11 for you

non--P5 one for you
P5 schools 11 for you

non-P5 one for ... Oh! Wait .. sorry there aren't any more ... so sorry, come back next year!

:outtahere:

Don 10-29-2018 05:10 PM

Comparison
 
Quote:

Originally Posted by cj (Post 557016)
It would be nice to compare the NET with the old RPI over the next few years to see where they differ and which one is the most "accurate" predictor.

I have been tabulating, for the past 4-5 years, whether the RPI rank or the seed numbers are a more accurate predictor of NCAA tournament success. The RPI rank is an objective and understood calculation. The tournament seeding is a subjective "gut feeling" of the selection committee. I have found that it's a tossup between which method is the better predictor. Some years the RPI is more accurate at predicting winners, some years less accurate. They are NEVER very far apart.

What I have noticed however, is that the P5 teams are consistently over-seeded and the non-P5 teams are consistently under-seeded. What this does is bias the results by more frequently pitting a non-P5 team against a higher seed (tougher opponent) while the P5 higher seeded but lower RPI-ranked teams get easier first round opponents. If this bias were eliminated, my guess is that the RPI would be a consistently better predictor of results.

I think the selection committee could be entirely done away with at no loss to tournament integrity and a definite gain in fairness.

Chris R 10-29-2018 09:51 PM

Quote:

Originally Posted by N2663R (Post 560383)
Chris -

Are you able to post both the RPI link as you have in the past along with the new NET formula rankings so we can follow along and see how it affects the ranking for us and all teams??

Right now we dont know when and where the new NET will be posted to interrogate, nor the formula to replicate it if we wanted to take on the task ourselves.

I did hear from a pretty reliable source that David Worlock(?) of the NCAA was anticipating disseminating the general parameters of the formula at some point, but no ETA when nor how specific those details might be.

flyerfanatic86 10-30-2018 08:49 AM

Quote:

Originally Posted by Don (Post 560398)
I have been tabulating, for the past 4-5 years, whether the RPI rank or the seed numbers are a more accurate predictor of NCAA tournament success. The RPI rank is an objective and understood calculation. The tournament seeding is a subjective "gut feeling" of the selection committee. I have found that it's a tossup between which method is the better predictor. Some years the RPI is more accurate at predicting winners, some years less accurate. They are NEVER very far apart.

What I have noticed however, is that the P5 teams are consistently over-seeded and the non-P5 teams are consistently under-seeded. What this does is bias the results by more frequently pitting a non-P5 team against a higher seed (tougher opponent) while the P5 higher seeded but lower RPI-ranked teams get easier first round opponents. If this bias were eliminated, my guess is that the RPI would be a consistently better predictor of results.

I think the selection committee could be entirely done away with at no loss to tournament integrity and a definite gain in fairness.

I think the committee has been transitioning away from which teams "earned" a bid to which teams are the "best." This gives them latitude to include P5 schools with a lot of "talent" that may not have won enough to really earn a spot in the tournament. It's also why you will see discussion of a lot of metrics that have nothing to do with W/L and who you beat/where you played. Stuff like offensive/defensive efficiency shouldn't matter if you aren't winning enough, but the committee can use that to justify a team's inclusion or exclusion.

Alberto Strasse 10-30-2018 09:15 AM

The NCAA
 
is protecting their business interests. Their business is football and the money invested by colleges in that sport. Those who invest more money get more favorable treatment on Selection Sunday.

CT Flyer 10-30-2018 10:41 AM

Quote:

Originally Posted by flyerfanatic86 (Post 560411)
I think the committee has been transitioning away from which teams "earned" a bid to which teams are the "best." This gives them latitude to include P5 schools with a lot of "talent" that may not have won enough to really earn a spot in the tournament. It's also why you will see discussion of a lot of metrics that have nothing to do with W/L and who you beat/where you played. Stuff like offensive/defensive efficiency shouldn't matter if you aren't winning enough, but the committee can use that to justify a team's inclusion or exclusion.

I've been saying this for a long time now that it won't be long before a P5 team makes it with a sub 500 record because "they are a good team but played a tough schedule". Winning games should still matter!!!

Figgie123 10-30-2018 12:03 PM

Quote:

Originally Posted by CT Flyer (Post 560424)
I've been saying this for a long time now that it won't be long before a P5 team makes it with a sub 500 record because "they are a good team but played a tough schedule". Winning games should still matter!!!

You can only get into the tournament with a sub-.500 record if you win a conference tournament. Rules are already in place that state you cannot get an at-large bid with lower than a .500 record.

Gazoo 10-30-2018 12:49 PM

Quote:

Originally Posted by Figgie123 (Post 560429)
You can only get into the tournament with a sub-.500 record if you win a conference tournament. Rules are already in place that state you cannot get an at-large bid with lower than a .500 record.

All the animals are equal. But some are more equal than others.

Figgie123 10-30-2018 01:33 PM

Quote:

Originally Posted by Gazoo (Post 560431)
All the animals are equal. But some are more equal than others.

I never argued this wasn't the case, just refuting the sub-.500 comment.

CT Flyer 10-30-2018 01:44 PM

Quote:

Originally Posted by Figgie123 (Post 560429)
You can only get into the tournament with a sub-.500 record if you win a conference tournament. Rules are already in place that state you cannot get an at-large bid with lower than a .500 record.

OK .500 record then....and I was being a little facetious anyway.

Gazoo 10-30-2018 04:15 PM

Quote:

Originally Posted by Figgie123 (Post 560432)
I never argued this wasn't the case, just refuting the sub-.500 comment.


Uh-huh. Based on today's rules.

It's not like it takes a supermajority of state senates to ratify a change. Remember when athletes who got paid were a major infraction?

Then the P5 said (in their Office Space voice): yeah, if we could just change that rule so we could start paying players, that would be great.
And now players get free tuition plus a stipend.


All the animals are equal.

Figgie123 11-24-2018 01:40 PM

Very early on, but if my math is right, using the graphic someone posted for net offensive efficiency, UD is ranked 6th in the A10 in net offensive efficiency at 0.0780.
  • Fordham (4-1) RPI 152 = 0.1723
  • Rhode Island (2-1) RPI 157 = 0.1307
  • Massachusetts (4-3) RPI 279 = 0.1249
  • VCU (4-1) RPI 86 = 0.0941
  • Saint Louis (4-1) RPI 70 = 0.0839
  • Dayton (4-2) RPI 111 = 0.0780
  • Saint Josephs (3-2) RPI 172 = 0.0618
  • Davidson (4-1) RPI 61 = 0.0365
  • Duquesne (3-1) RPI 33 = 0.0132
  • Richmond (2-3) RPI 234 = 0.0023
  • George Mason (2-4) RPI 266 = -0.0145
  • Saint Bonaventure (1-5) RPI 348 (ugh!) = -0.0520
  • La Salle (0-6) RPI 309 = -0.1775
  • George Washington (0-5) RPI 265 = -0.2555

This is based on stats straight from NCAA.org this morning, and it includes non-Div1 games. For instance, Fordham beat CCNY in their first game, and those stats are included in the team totals that NCAA lists. I'd hope NCAA finds a way to get the non-Div 1 games out of there.

Figgie

TX Flyer 11-24-2018 02:22 PM

If they don’t then ud just needs to load the non conf with d2 and d3 teams?

When does the first rankings get released? I googled but couldnt find anything

ruechalgrin 11-24-2018 02:55 PM

Quote:

Originally Posted by Figgie123 (Post 563548)
Very early on, but if my math is right, using the graphic someone posted for net offensive efficiency, UD is ranked 6th in the A10 in net offensive efficiency at 0.0780.
  • Fordham (4-1) RPI 152 = 0.1723
  • Rhode Island (2-1) RPI 157 = 0.1307
  • Massachusetts (4-3) RPI 279 = 0.1249
  • VCU (4-1) RPI 86 = 0.0941
  • Saint Louis (4-1) RPI 70 = 0.0839
  • Dayton (4-2) RPI 111 = 0.0780
  • Saint Josephs (3-2) RPI 172 = 0.0618
  • Davidson (4-1) RPI 61 = 0.0365
  • Duquesne (3-1) RPI 33 = 0.0132
  • Richmond (2-3) RPI 234 = 0.0023
  • George Mason (2-4) RPI 266 = -0.0145
  • Saint Bonaventure (1-5) RPI 348 (ugh!) = -0.0520
  • La Salle (0-6) RPI 309 = -0.1775
  • George Washington (0-5) RPI 265 = -0.2555

This is based on stats straight from NCAA.org this morning, and it includes non-Div1 games. For instance, Fordham beat CCNY in their first game, and those stats are included in the team totals that NCAA lists. I'd hope NCAA finds a way to get the non-Div 1 games out of there.

Figgie

Tied for #2 with Davidson in offensive efficiency behind only St Joe's in kenpom. Please tell me the NCAA adjusts for competition. Kenpom not totally normalized yet (relies on assumptions until normalized). But I bet the issue is with the NCAA. The whole NET thing is an absolute fiasco. Zero transparency on how calculate the 5 factors, zero transparency on weighting of the 5 factors, and then they throw in "machine learning." We have over 100+ artificial intelligence (think machine learning but I am simplifying) engineers and how the NCAA constructs the algorithm, what data they use to learn, etc, etc, etc all matter massively. GuRanteee nobody on the committee has a clue.
Posted via Mobile Device


All times are GMT -4. The time now is 11:47 AM.

Powered by vBulletin® Version 3.8.6
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.

Powered by vBadvanced CMPS v3.2.1

All times are GMT -4. The time now is 11:47 AM.


Powered by vBulletin® Version 3.8.6
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Advertisement System V2.6 By   Branden

     
 
Copyright 1996-2012 UDPride.com. All Rights Reserved.