(03-15-2021 06:48 PM)Wedge Wrote:  It's not just "schedule the easiest possible non-con games and win them all", it's "schedule non-con games against teams that are weak but have great W-L records, so that the win percentage of your opponents looks great to RPI".

Oh, one more note on that. I was friends with a college coach who just missed the NCAA tournament. She said "You know this RPI stuff, how do we get an RPI where they have to take us? Teach me."

So I pulled out the whiteboard and told her EXACTLY what you just said. Good SOS doesn't mean "tough opponents," it means "high win pct."

If you play at Notre Dame, you're probably going to lose and they're going to be an 15-15 team on your SOS. If you play small conference champions, like Winthrop, you can beat them and they'll be 22-1 on your SOS.

She put together just a brilliant schedule and was a lock for a bid even if they lost their conference tourney.

(03-15-2021 10:06 AM)IWokeUpLikeThis Wrote:  Colgate is #9 because they played the same 3 teams repeatedly. NET wasn’t built to accommodate for that.

I think the bigger flaw wasn't seeing the same small sample of schools over and over again, but playing in a conference where most of the non-conference schedule was eliminated, leaving pretty much a Patriot-only schedule for some teams, and perhaps breaking the SOS components.

I really hate the hypocrisy of it, honestly. The same metric that "Colgate broke" also had some kind of "pure result" where most of the Big Ten found their way into the top 100 (and a sub-.500 team around the top 40)? If they broke the .500 barrier, they were fair game for consideration. And they got seeded approximately close to their respective NET ranking.

And then there's Loyola, who should have it so much better than they did. It's an insult they couldn't even get the 7/10 to avoid the 1-seed its first weekend. I really hate this metric; hate it more because the committee clearly doesn't value it when it comes to the non-majors.

I've been in tournaments where I knew I had a team in the 4th to 6th best but found myself the 8 seed or 9 seed and wound up facing the top team in the 1st or 2nd round, really the only team we didn't match up with. It is what it is.

Loyola Chicago is slotted 30th overall seed, an 8 seed in bracket. Honestly that is probably where they legitimately fall. They start in the churn (7-10 seeds, 25-40 seeds overall) where anyone can beat anyone, but your reward is to face a 1 or 2 seed. They don't get any special treatment, nor should they or anyone else from any conference. They have a winnable first game and that is all you can ask.

The NET is a refinement on RPI, because it takes into consideration road, home and neutral site, among other adjustments. It is tested to predict results and has been tweaked to improve it's predictive capability.

https://www.si.com/college/2018/11/04/co...em-explain
https://www.ncaa.com/news/basketball-men...-explained

But it's a tool not an end all, which Ken wants. There never will be an end all. There is a human judgement element. NET merely lumps teams into rough groupings. And the committee gave it very strong value. Every eligible school in the top 48 (Arizona at #44 is serving a self-imposed post season ban ahead of NCAA sanctions, #42 Penn State had a losing record making them ineligible) is in the tournament, accounting for 35 of the 37 at-large bids, except Saint Louis at #43.

Now it is worth noting that the actual NET score is not published for the public, merely the ranking position. This gives a false impression of discrete values and of equal difference between each ranking. That is not true. What I heard repeatedly from the pundits who know the actual numbers was that there is a very large gap between the first 4 schools, the 1 seeds and the 2 seeds. It is a reasonable extrapolation, given the mixed up shuffling of rankings and seeds after the first 8 (Colgate being an outlier due to the covid scheduling), that the actual NET scores are much more bunched after that. There probably is not much difference in the actual NET scores from say Utah State at #39 to say NC State at #73. Even so the committee only picked two schools from deeper.

Do I think politics played a role? Possibly, but most likely not for the M6 (P5 + Big East) rather for the UMM5 (WCC, MVC, MWC, AAC, A10) to get each of them a bid (Drake, Wichita State and Utah State all 11 seeds).

The case for Michigan State is 5 quad-1 and 4 quad-2 wins with no quad-3 or quad-4 losses, and for Wichita State the regular season winner of the American, 2 quad-1 and 2 quad-2 wins with only 1 quad-3 loss. Saint Louis also had 2 quad-1 and 2 quad-2 wins but did not win the A10 regular season, and suffered 2 quad-3 defeats. Also not all quad-1 and quad-2 games are worth the same. Michigan State beat Ohio State and Illinois, while Wichita State beat Houston. The best Saint Louis had was a win over St. Bony, a 9 seed, which doesn't match up to those 1 and 2 seed wins. The committee decision is understandable given that.

One can split hairs over Syracuse instead of Louisville and Colorado State, but blame that on bid stealers Oregon State and Georgetown. Even Syracuse did better in the ACC tournament, so arguably played their way in while Louisville and Colorado State played their way out. But hey we are at the fringe of the Tournament selections and there are always marginal cases.

We do see the NIT committee payed great attention to NET in their selection. Every eligible school up to 82nd, except Furman #75 and Stanford #78 was offered a slot (5 turned them down: Louisville, Duke, Seton Hall, Xavier, St. John's). Note, the NIT didn't give automatic qualifiers. There would have been 14 or 15 (not sure on NC A&T as they pulled out due to covid, may have just shut down like Duke), of which 2 they invited (Western Kentucky, Toledo). They probably offered the same teams they would have in a regular year, just a few more rejections than normal (Covid-19, so no home games to collect gate from for M6 schools).

The evidence is pretty strong that NET was relied upon heavily by the committees. But it is not an end all, they seeded as they saw fit.

NET worked well for UC Santa Barbara. Began conference play in the 120s, finished at 54 on Selection Sunday. Despite Zero quad 1 games, a 2-4 record in quad 2. A 12 seed ain't bad

(03-16-2021 01:36 PM)jdgaucho Wrote:  NET worked well for UC Santa Barbara. Began conference play in the 120s, finished at 54 on Selection Sunday. Despite Zero quad 1 games, a 2-4 record in quad 2. A 12 seed ain't bad

For most mid majors, #12 seeds tend to be the ceiling. Historically #12 seeds outperform #13 seeds in the NCAA Tournament. From 1985 to 2019, #12 seeds have won 72 games in the NCAA Tournament, #13 seeds have won 35, less than half. In the first round, #12 seeds have won 50 games vs. 29 for #13 seeds. Great for UCSB to get that #12.

More interesting seed trivia: https://www.bloggingthebracket.com/2021/...ce-by-seed

(03-16-2021 01:24 PM)Stugray2 Wrote:  I've been in tournaments where I knew I had a team in the 4th to 6th best but found myself the 8 seed or 9 seed and wound up facing the top team in the 1st or 2nd round, really the only team we didn't match up with. It is what it is.

Loyola Chicago is slotted 30th overall seed, an 8 seed in bracket. Honestly that is probably where they legitimately fall. They start in the churn (7-10 seeds, 25-40 seeds overall) where anyone can beat anyone, but your reward is to face a 1 or 2 seed. They don't get any special treatment, nor should they or anyone else from any conference. They have a winnable first game and that is all you can ask.

The NET is a refinement on RPI, because it takes into consideration road, home and neutral site, among other adjustments. It is tested to predict results and has been tweaked to improve it's predictive capability.

https://www.si.com/college/2018/11/04/co...em-explain
https://www.ncaa.com/news/basketball-men...-explained

But it's a tool not an end all, which Ken wants. There never will be an end all. There is a human judgement element. NET merely lumps teams into rough groupings. And the committee gave it very strong value. Every eligible school in the top 48 (Arizona at #44 is serving a self-imposed post season ban ahead of NCAA sanctions, #42 Penn State had a losing record making them ineligible) is in the tournament, accounting for 35 of the 37 at-large bids, except Saint Louis at #43.

Now it is worth noting that the actual NET score is not published for the public, merely the ranking position. This gives a false impression of discrete values and of equal difference between each ranking. That is not true. What I heard repeatedly from the pundits who know the actual numbers was that there is a very large gap between the first 4 schools, the 1 seeds and the 2 seeds. It is a reasonable extrapolation, given the mixed up shuffling of rankings and seeds after the first 8 (Colgate being an outlier due to the covid scheduling), that the actual NET scores are much more bunched after that. There probably is not much difference in the actual NET scores from say Utah State at #39 to say NC State at #73. Even so the committee only picked two schools from deeper.

Do I think politics played a role? Possibly, but most likely not for the M6 (P5 + Big East) rather for the UMM5 (WCC, MVC, MWC, AAC, A10) to get each of them a bid (Drake, Wichita State and Utah State all 11 seeds).

The case for Michigan State is 5 quad-1 and 4 quad-2 wins with no quad-3 or quad-4 losses, and for Wichita State the regular season winner of the American, 2 quad-1 and 2 quad-2 wins with only 1 quad-3 loss. Saint Louis also had 2 quad-1 and 2 quad-2 wins but did not win the A10 regular season, and suffered 2 quad-3 defeats. Also not all quad-1 and quad-2 games are worth the same. Michigan State beat Ohio State and Illinois, while Wichita State beat Houston. The best Saint Louis had was a win over St. Bony, a 9 seed, which doesn't match up to those 1 and 2 seed wins. The committee decision is understandable given that.

One can split hairs over Syracuse instead of Louisville and Colorado State, but blame that on bid stealers Oregon State and Georgetown. Even Syracuse did better in the ACC tournament, so arguably played their way in while Louisville and Colorado State played their way out. But hey we are at the fringe of the Tournament selections and there are always marginal cases.

We do see the NIT committee payed great attention to NET in their selection. Every eligible school up to 82nd, except Furman #75 and Stanford #78 was offered a slot (5 turned them down: Louisville, Duke, Seton Hall, Xavier, St. John's). Note, the NIT didn't give automatic qualifiers. There would have been 14 or 15 (not sure on NC A&T as they pulled out due to covid, may have just shut down like Duke), of which 2 they invited (Western Kentucky, Toledo). They probably offered the same teams they would have in a regular year, just a few more rejections than normal (Covid-19, so no home games to collect gate from for M6 schools).

The evidence is pretty strong that NET was relied upon heavily by the committees. But it is not an end all, they seeded as they saw fit.

I don't want an "end all". There isn't one. All I'm saying is that the NET is just a more complicated flawed metric that really didn't make the selection process any better than it was before. It's not that it's bad, or even worse than others. It just wasn't needed and wasn't an improvement.

(03-16-2021 01:24 PM)Stugray2 Wrote:  Loyola Chicago is slotted 30th overall seed, an 8 seed in bracket. Honestly that is probably where they legitimately fall. They start in the churn (7-10 seeds, 25-40 seeds overall) where anyone can beat anyone, but your reward is to face a 1 or 2 seed. They don't get any special treatment, nor should they or anyone else from any conference. They have a winnable first game and that is all you can ask.

But only Loyola and Colgate saw the sort of respective seed drop from their NET position. And Loyola didn't play just a conference schedule like Colgate and most of Patriot. To go from NET #10, aligning with the 3-line, to NET #29-32 and the 8-line is inexcusable if nobody else is seeing that kind of drop or shift. And given the gaps that occur from lines 1 and 2 (because they are there), not even giving Loyola the benefit of the 7-line is what annoys me more.

And I said this earlier and elsewhere, but, it's not just for Loyola that I feel for in this case. It's Illinois, too. It's unfair for Illinois because the seeding concept should be to regionally protect the top seed for the first weekend (it's why the 8/9's may be the worst lines in the whole tournament, since they are to have NO advantage over the 1-seed of the region). That's a wash here.

Honestly, the potential matchup screams of CBS/Viacom using their "network partner" privileges to advise on a potentially strong regional matchup, but also a quick national check on whether Loyola "stands up" against the nation's best. Potentially thrilling television, but, not in the spirit of competition. It's almost like the reverse of what UNC or Duke would experience when they would draw a game in Greensboro and their opponents would be from a different time zone or a few states away north or south.

This was avoidable. And it definitely sets a bad tone about NET going forward, especially if you're a non-major who does well but doesn't benefit from strong conference surroundings. The committee WILL hold it against you, but not the others. Even though it's the same metric with the same flaws.

(03-17-2021 01:06 PM)The Cutter of Bish Wrote:  And it definitely sets a bad tone about NET going forward

No, it doesn't. The committee didn't take an oath solemnly swearing to use NET as the one and only factor in seeding teams, or in deciding which teams are in or out. In fact, the NCAA said when NET was introduced that it was only one tool to be used.

It's obvious that the committee has additional factors, including the number of quality wins a team has -- that factor helped Michigan State get in, with a NET ranking of #70, well below many teams that were not invited, and quality wins most notably helped Oklahoma State get a #4 regional seed even though their #29 NET ranking would correspond to a #8 seed if NET was the only factor.

I am not endorsing which factors they use or don't use -- for example, I think it's absurd that they don't reward road wins and don't penalize teams who can't win on the road; there are a handful of at-large teams that won 3 or fewer road games all season -- but the point is, they have never said that NET was anything more than one factor among several.

8/9 is the death slot too

(03-17-2021 03:31 PM)Wedge Wrote:  

(03-17-2021 01:06 PM)The Cutter of Bish Wrote:  And it definitely sets a bad tone about NET going forward

No, it doesn't. The committee didn't take an oath solemnly swearing to use NET as the one and only factor in seeding teams, or in deciding which teams are in or out. In fact, the NCAA said when NET was introduced that it was only one tool to be used.

It's obvious that the committee has additional factors, including the number of quality wins a team has -- that factor helped Michigan State get in, with a NET ranking of #70, well below many teams that were not invited, and quality wins most notably helped Oklahoma State get a #4 regional seed even though their #29 NET ranking would correspond to a #8 seed if NET was the only factor.

I am not endorsing which factors they use or don't use -- for example, I think it's absurd that they don't reward road wins and don't penalize teams who can't win on the road; there are a handful of at-large teams that won 3 or fewer road games all season -- but the point is, they have never said that NET was anything more than one factor among several.

Wasn't part of the framework of the NET that it rewarded quality wins by using "quads"? Or "bad losses" as well? So wouldn't Michigan State benefit twice from winning games against the better teams in its conference - once in their NET rank, and a second time in the committee's "eye test"?

My point is that if we are going to have a selection committee of "experts" (and I'm not sure that's really what we have) why not just let them apply their expertise without opaque formulas that they can just ignore when they don't fit their personal biases or tout as the reason for their decisions when they do?

Or, if we are going to have a committee structured so that various interest groups (like conferences) are represented so as to have offsetting biases, why not just recognize that the makeup of the final field will always be political and decide in advance how many bids each conference should have?

(03-17-2021 07:15 PM)ken d Wrote:  

(03-17-2021 03:31 PM)Wedge Wrote:  

(03-17-2021 01:06 PM)The Cutter of Bish Wrote:  And it definitely sets a bad tone about NET going forward

No, it doesn't. The committee didn't take an oath solemnly swearing to use NET as the one and only factor in seeding teams, or in deciding which teams are in or out. In fact, the NCAA said when NET was introduced that it was only one tool to be used.

It's obvious that the committee has additional factors, including the number of quality wins a team has -- that factor helped Michigan State get in, with a NET ranking of #70, well below many teams that were not invited, and quality wins most notably helped Oklahoma State get a #4 regional seed even though their #29 NET ranking would correspond to a #8 seed if NET was the only factor.

I am not endorsing which factors they use or don't use -- for example, I think it's absurd that they don't reward road wins and don't penalize teams who can't win on the road; there are a handful of at-large teams that won 3 or fewer road games all season -- but the point is, they have never said that NET was anything more than one factor among several.

Wasn't part of the framework of the NET that it rewarded quality wins by using "quads"? Or "bad losses" as well? So wouldn't Michigan State benefit twice from winning games against the better teams in its conference - once in their NET rank, and a second time in the committee's "eye test"?

My point is that if we are going to have a selection committee of "experts" (and I'm not sure that's really what we have) why not just let them apply their expertise without opaque formulas that they can just ignore when they don't fit their personal biases or tout as the reason for their decisions when they do?

Or, if we are going to have a committee structured so that various interest groups (like conferences) are represented so as to have offsetting biases, why not just recognize that the makeup of the final field will always be political and decide in advance how many bids each conference should have?

The committee structure is very flawed, no doubt. When the committee (this year's members listed below) gets anything right, it's in spite of those flaws.

The biases that you mention don't offset, because most conferences are not represented. And the way the media covers the sport, with the compulsive fixation on conferences rather than teams (a fixation that is well-represented on this message board, of course), each committee member has a vested interest to not only not help teams from other conferences, but to screw them, so that their own conference looks better in terms of number of bids as compared to whatever conference(s) they compare themselves to.

If your league isn't represented on this committee, the biases can only hurt you, not help you.

Quote:Mitch Barnhart, Division I Men’s Basketball Committee chair and director of athletics at University of Kentucky
Mike Bobinski, vice president and director of athletics, Purdue University
Tom Burnett, commissioner, Southland Conference
Lawrence R. Cunningham, director of athletics, University of North Carolina
Charles McClelland, commissioner, Southwestern Athletic Conference
Bernadette McGlade, commissioner, Atlantic 10 Conference
Michael O’Brien, vice president and director of athletics, University of Toledo
Jamie Pollard, director of athletics, Iowa State University
Chris Reynolds, vice president for intercollegiate athletics, Bradley University
Craig Thompson, commissioner, Mountain West Conference

(03-15-2021 06:48 PM)Wedge Wrote:  

(03-15-2021 06:22 PM)JSchmack Wrote:  Whatever your conference does OOC becomes your Conference Game SOS; so all that matters is having the highest OOC win percentage possible.

There is one more layer to the RPI than that. The RPI "fixers" who sold themselves as consultants pushed the idea that while non-conference win percentage is important, the win percentage of your non-conference opponents (and even your conference mates' non-conference opponents) is also important. It's not just "schedule the easiest possible non-con games and win them all", it's "schedule non-con games against teams that are weak but have great W-L records, so that the win percentage of your opponents looks great to RPI".

Thus, when a team looks to buy a non-con opponent, the perfect "easy win" non-con opponent for RPI is not the worst team you can find, but the best team in one of the worst conferences, eg the team that ends up winning the 27th strongest D-I conference out of 32. That's because your team will (hopefully) get not only a sure win but also a win over a team that will (hopefully) finish with 17 to 24 wins by dominating its conference.

It was possible for "non-power" conferences, as well as power conferences, to game the RPI this way because buying non-con basketball games is not very expensive. It's peanuts compared to buying a non-con football game. Even today, it's usually $50,000 or less, and 15-20 years ago when some folks started griping about RPI manipulation it was probably less than that.

I know what I know and I know what I don't know. And this goes in the latter category. Thanks

The committee can cite an extremely high NET ranking as a reason a major conference team almost HAS to be in the tournament, while ignoring it for anyone they deem unworthy. They can use it say they used it, or ignore it and say they ignored it. Its just a sound bite that can justify anything they choose to do. Total scam.

It either has a specific function or it doesn't.

(03-15-2021 09:06 AM)ken d Wrote:  if you aren't going to use it to decide who gets invited to The Dance?

Drake, UCLA, Missouri and Virginia Tech are rated #45-48 and they got into the field. But #43 St. Louis and #44 Arizona did not.

#70 Michigan State and #72 Wichita State also were invited ahead of Duke, Boise State, Colorado State, Memphis, Ole Miss, Louisville and Seton Hall, who were rated between #49 and #57.

Loyola-Chicago was ranked #10 in the NET, but given an 8 seed (where you would expect to see teams ranked 29-32.

And Colgate, #9 in the NET, is seeded on the #14 line.

I'm not suggesting those seeds were inappropriate. I'm suggesting that those NET rankings mean the NET itself is a lousy way to rank teams.

So why does the selection committee sometimes use it to justify who to put on the right side of the bubble, and other times ignore it completely? Why not just do away with it? How is it any better than the many other rankings used before it? The use of "quads" to calculate a ranking just makes it more confusing to fans without adding any value.

Is the NET just a solution in search of a problem?

Because they need excuses for the piss poor decisions they make. NET is a convenient talking point when needed. Just like the eye test and the last 5 games excuses that are used when needed. We can spout off all the metrics we want, but the committee will continue to reward the p5 and screw everyone else.

Sent from my SM-G975U using CSNbbs mobile app

The trick is to minimize bad teams in a conference. CUSA has 14 schools, and if 6 of them fail to understand the game plan then the other 8 are going to have problems making the tournament.

Tulane, my alma mater, has been that problem school in recent memory (Dunleavy era).

The AAC at 11 teams can only absorb 1 bad team - and maybe a second if the big dogs eat in nonconference play.

5 days later...

(03-15-2021 12:36 PM)Scoochpooch1 Wrote:  So let's give Loyola Chicago 3 seeds and watch them lose by 40?

(8) Loyola 71
(1) Illinois 58

(03-17-2021 03:31 PM)Wedge Wrote:  

(03-17-2021 01:06 PM)The Cutter of Bish Wrote:  And it definitely sets a bad tone about NET going forward

No, it doesn't. The committee didn't take an oath solemnly swearing to use NET as the one and only factor in seeding teams, or in deciding which teams are in or out. In fact, the NCAA said when NET was introduced that it was only one tool to be used.

It's obvious that the committee has additional factors, including the number of quality wins a team has -- that factor helped Michigan State get in, with a NET ranking of #70, well below many teams that were not invited, and quality wins most notably helped Oklahoma State get a #4 regional seed even though their #29 NET ranking would correspond to a #8 seed if NET was the only factor.

I am not endorsing which factors they use or don't use -- for example, I think it's absurd that they don't reward road wins and don't penalize teams who can't win on the road; there are a handful of at-large teams that won 3 or fewer road games all season -- but the point is, they have never said that NET was anything more than one factor among several.

Walking this one back a bit after the initial rounds
From the NCAA's website:

Quote:The NET, which stands for the NCAA Evaluation Tool, replaced the RPI after the 2017-18 season and it is used as the primary sorting tool for selection and seeding for the NCAA tournament.

That being noted, the two components aren't mutually exclusive. You can have both a bad metric and problematic committee. What makes the validity of NET troublesome, even with what lack of accountability or transparency the committee provides to its processes, is when NET rankings almost mirror the seedings (with an average of +/-1 line deviance) as the end result. Is it guilt by association?

Like you, it's not about all the committee uses or doesn't with any endorsements beyond NET (I wonder whether RPI may still be very much alive, since RPI might explain Oklahoma State's big bump up the seed line). We may never know how much, if any, weight is given solely to NET. But, it IS the metric the NCAA is primarily using to select and seed programs. And the committee seems to be using it pretty close when selecting and seeding the majority of their teams.

On its own feet, NET's just a metric. It appears biased toward major conferences somehow (SOS it seems), but the double-down the committee takes on seeding majors almost aligned to NET but to drop it on mid-majors DOES reflect poorly on the stat. The committee is rejecting its primary tool when it places non-major schools who score well overall, but are observed for other purposes, like W/L's in the quads, away from the near adherence it gives toward majors. This year, the committee got it wrong, WAY wrong on Loyola. Conversely, we're supposed to believe that NET got it wrong, WAY wrong on Colgate. Both can be true. Both make the NET look bad.

And I get it that this season was unique. But, I still look at the Colgate's and Penn State's of this metric, where someone can rocket into the top 10 without playing even close to the average number of games all other schools played, or watch a sub-.500 team fall into safe/bubble territory of the ranking component. I don't want to make this about hating a conference like the Big Ten, but, NET and the committee look like they got them wrong. I don't want to make it about saving the poor little guy, because, well, Gonzaga and Loyola are doing just fine, while others clearly had flaws (including the likes of Winthrop and the snubbed Belmont). And if you swapped out Drake and Wichita for Colorado State and Saint Louis, or Memphis, or Old Miss, or Louisville...were any of those teams equipped to do battle with USC? HIGHLY doubtful. Still...the metric attempts to tell a narrative on the season, with the committee the interpreter.

It's not a good look or pairing.