Big East OOC tracker - December

[SoFla Steve]

The RPI was straight forward. Your record, your opponents' record, and your opponents' opponents' record, with an adjustment for home and away. Everyone understood how it worked, and no one could complain when they finished the season with a low RPI. The only real complaint, that some high majors gamed the system by playing a lot of mid-majors is way overblown. Playing decent mid major teams is a risk, because those teams occasionally beat the high majors. If a team is willing to take the risk and play a MAC or MAAC team instead of Central, they deserve the reward.

NET rewards running up the score on bad teams. Leagues like the Big 12 have already cut back on major conference opponents, because why bother?

From the article . . .

NCAA announces changes to NET ranking system​

By RILEY GATES May 11, 2020

On Monday afternoon, the NCAA announced changes to the NET ranking system. The ranking will now lean on just two factors: Team Value Index and adjusted net efficiency rating. TVI rewards teams for beating quality opponents, particularly in road games or at neutral sites. Adjusted efficiency takes strength of opponent and location into account when analyzing just how valuable a win was.

The old NET ranking system included those two factors, while also taking into consideration: winning percentage, adjusted winning percentage and scoring margin. But after consulting with Google Cloud Professional Services, those factors were dropped.

 

[nelsonmuntz]

From the article . . .

NCAA announces changes to NET ranking system​

By RILEY GATES May 11, 2020

On Monday afternoon, the NCAA announced changes to the NET ranking system. The ranking will now lean on just two factors: Team Value Index and adjusted net efficiency rating. TVI rewards teams for beating quality opponents, particularly in road games or at neutral sites. Adjusted efficiency takes strength of opponent and location into account when analyzing just how valuable a win was.

The old NET ranking system included those two factors, while also taking into consideration: winning percentage, adjusted winning percentage and scoring margin. But after consulting with Google Cloud Professional Services, those factors were dropped.

If they say so. Since the formula is a black box, and the Big 12 is ranked as the #1 league in NET despite few quality OOC wins, I am going to go with my description.

 

[nelsonmuntz]

It's way more than that. it is indeed straight foward... too much so. 75% of the RPI was out of team's control, the only influence being who they scheduled against. Gaming could be done lots of ways and none of which actually measured how good or not a team actually is. That last part is key, because of how the NET is used in determining the quadrant system and not for seeding. You want to really know the skill level of a team, not using it to hang banners. No one is meant to hang their hat on being NET #1, it's just a tool used to figure out how good of an opponent a team played.

75% was not out of a team's control. 50% was the opponents' winning percentage. Schedule cupcakes and get punished, schedule tough opponents, and get rewarded. The last 25% was opponents' opponents' record, but all that meant was playing the best team in a bad league didn't warp the RPI that much.

People can say it was easy to game, but if everyone knows how to game it, then anyone can do it and it is no longer really gaming the system. The NET is a black box that some leagues or schools may or may not have access too. It does not get more out of a team's control than that.

 

[MU in MKE]

You can't say 75% of the RPI was out of a teams control and also say it was easy to manipulate. You gotta pick one or the other plz

I probably could have worded it better. What I mean by that is 75% of a teams control over their RPI is not affected by what the team does on the court. A far greater part of that, at least 75%, had more to do with scheduling than a team actually playing a game.

If it were to me I would gladly bring the RPI back. Besides at the end of the day what you are trying to do is pick 68 teams. Then do it in the most transparent way possible.

This is the point I'm trying to make, but it is difficult through printed word. Neither the RPI nor the NET are used to pick the 68 teams directly. They are just a sorting tool to help the committee read the team resumes. That's why you want the best tool to actually tell you how good a team is, which is one in which you're using analytics to measure performance and the win/loss column of individual games. The W/L record is what gets you picked... the RPI/NET are just tools to determine how good or bad your W or L was. That's why the RPI is wholly inappropriate for the usage.

When it came out it was the best solution at the time, before most personal computers and spreadsheets existed. However, we have tools available to us now that make an analytical approach easy. We have for a long time, but as with most things, the NCAA is not quick to change or improve.

 

[JonnyRI]

Saragin rankings vs Top 50

BE 15-16
B12 10-6
B10 19-28
ACC 10-26

BE looking good. Pac12 so bad not worth posting.

 

[auror]

RPI was just not a very accurate measure of EITHER strength of record OR predictive team strength. Yeah it was more transparent and thus predictable, but that doesn't mean it was accurately measuring anything important.

NET is pretty good at both, but intentionally geared to do both so not perfect at either. That being said, by tournament time it's totally fine and a good enough metric for committee to use for organizing.

 

[ecasadoSBU]

This is the point I'm trying to make, but it is difficult through printed word. Neither the RPI nor the NET are used to pick the 68 teams directly. They are just a sorting tool to help the committee read the team resumes. That's why you want the best tool to actually tell you how good a team is, which is one in which you're using analytics to measure performance and the win/loss column of individual games. The W/L record is what gets you picked... the RPI/NET are just tools to determine how good or bad your W or L was. That's why the RPI is wholly inappropriate for the usage.

When it came out it was the best solution at the time, before most personal computers and spreadsheets existed. However, we have tools available to us now that make an analytical approach easy. We have for a long time, but as with most things, the NCAA is not quick to change or improve.

Got it. That's not bad I guess. I can live with that. Your W-L is most important and then they use the NET to evaluate the quality of those Wins and Losses on your record. Fair enough

But am I the only weirdo that thinks the NCAA should not be the gatekeepers to this formula?

Can we at least agree that the formula should be public and not kept in secret?

It just really rubs me the wrong way that the NCAA is the only one that has access to NET. Make it public and allow everyone to know what is the tool used to evaluate the quality of your wins and losses.

And are you really keeping it secret or are there a few insiders out there that already know the formula? Also, with how powerful computing power are nowadays isn't it technically possible to brute force (the few terms I remember from my CompSci days. lol) it and end up reverse engineering the formula anyway?


I can probably input the data points of the past two weeks into an AWS analytics service and run it until develops a formula that closely resembles the results of published NET

 

[Big East Fa]

Tuesday 12/14/21
Arizona State @ Creighton, 8:00, FS1 - Kevin Kugler, Bill Raftery
Line: Creighton -7.5/136.5

DePaul @ UIC, 8:00, ESPN+ - Jonathan Hood (PBP), Kenny Williams
Line: DePaul -10.5/142.5

 

[HooperScooper]

Anyone else watching this Creighton-Arizona St. game? I'm glad to know UConn isn't the only team in the BE that sometimes struggles mightily from three. Creighton is 5-25 from three with less than 4 minutes left and down by 2 points.

 

[shizzle787]

89-24. Bad loss for Creighton. 22-17 against the P5.

 

[shizzle787]

Creighton just made it difficult for the rest of the league. Need to go 7-1 to get to 96 wins and I don't see Butler beating Purdue so the league would need to go 7-0 in the remaining games.

I also think we can justifiably cross Creighton off the list of potential NCAA teams.

 

[HooperScooper]

Creighton just made it difficult for the rest of the league. Need to go 7-1 to get to 96 wins and I don't see Butler beating Purdue so the league would need to go 7-0 in the remaining games.

I also think we can justifiably cross Creighton off the list of potential NCAA teams.

Kind of early for that but they did look terrible in this game.

 

[Kolumbo]

But DePaul brought the haul. OK, a win is a win.

 

[Tiamat7]

Creighton just made it difficult for the rest of the league. Need to go 7-1 to get to 96 wins and I don't see Butler beating Purdue so the league would need to go 7-0 in the remaining games.

I also think we can justifiably cross Creighton off the list of potential NCAA teams.

Take it easy on them...they're young pups. They will surprise people -- in both good and bad ways this season. NIT bound.

Big East performed admirably out of conference this year. The remaining handful of games is icing on the cake.

 

[Big East Fa]

Wednesday
Howard @ Georgetown, 6:30, FS1 - Scott Graham, John Giannini
Line: Georgetown -11/156.5

Morehead State @ Xavier, 8:30, FS1 - Matt Schumacker, Donny Marshall
Line: Xavier -15/135.5

 

[nelsonmuntz]

RPI was just not a very accurate measure of EITHER strength of record OR predictive team strength. Yeah it was more transparent and thus predictable, but that doesn't mean it was accurately measuring anything important.

NET is pretty good at both, but intentionally geared to do both so not perfect at either. That being said, by tournament time it's totally fine and a good enough metric for committee to use for organizing.

NET is very good at justifying the outcome the power brokers in the NCAA want to get.

 

[nelsonmuntz]

Kind of early for that but they did look terrible in this game.

I agree. It was a stinker of a loss, but to be honest, I expected more of these losses by Big East teams by this point, and I expected some losses that were really ugly. I was expecting a few Big East equivalents of Navy over Virginia, a bad loss by one of our top half teams. So far, so good on that note.

 

[auror]

NET is very good at justifying the outcome the power brokers in the NCAA want to get.

By March, it's not very different from other independent advanced metrics. Prove to me with data that it favors a certain class of teams in a way that RPI didn't and I'll listen, instead of just spouting random conspiracy theories.

 

[MJ1]

Georgetown and Xavier win, keep the line moving.

 

[shizzle787]

91-24. Need to go 5-1 the rest of the way.

 

[nelsonmuntz]

Purdue stomping on Butler. Butler was lucky to beat Oklahoma. They are so slow, that I don't get why Purdue is shooting 3's. The Boilers can get right to the rim any time they want.

Butler should be two easy wins for the Huskies.

 

[Blades]

Well, was gonna check out Depaul today and their 9-1 start (2-0 vs. the P-5) ....but :

DePaul-Northwestern men’s basketball game canceled due to COVID-19 results with the Blue Demons

CHICAGO – It was expected to be an area match-up of two teams that have enjoyed a successful start to the 2021-2022 season. But like a few other contests in college basketball on tap this wee…

wgntv.com

 

[storrsroars]

Pitt giving Johnnies a game in MSG. 34-31 SJU at the half. Hugley sat out several minutes with 2 fouls.

 

[auror]

Pitt giving Johnnies a game in MSG. 34-31 SJU at the half. Hugley sat out several minutes with 2 fouls.

Champagnie is out today I think.

 

[storrsroars]

Champagnie is out today I think.

True. Covid.