MBB: Bracketology 2019-20

[NYBri]

Pitt and UNC playing tonight in the NIT play-in game. Loser plays the Washington Generals.

Cheats v. The Generals

[OldPhiKap]

Cheats v. The Generals

Cheats < The Generals. Take the G’s and give the points.

[Bluedog]

ESPN's John Gasaway has a piece up on his blog -- https://johngasaway.com/ -- advocating replacing the Committee for purposes of selection/seeding with some form of wins-based statistical ranking method; he proposes either Torvik's Wins-Above-Bubble ("WAB") or Strength of Record.

If the NCAA didn't want to pick just one such method, presumably they could (like with the NET, but better) create some kind of composite formulated from several different of such methodologies, that could have the benefit of being (1) transparent and (2) remove the vagaries and conspiracy theories surrounding the Committee and (3) no less "accurate" in its rankings than the Committee's choices.

Here's the crux of his argument for the benefits of such a change (beyond just "this isn't something we should be doing by a Committee"):

“[Replacing selection/seeding by a committee with WAB or SOR-type method] would make its impact felt on every aspect of the season … . Selection would become real-time, and each game’s true impact on the bubble or on the race for the four No. 1 seeds would be known day by day. Conference tournaments in particular would be turbo-charged as we’d know the exact stakes in play and as bids change hands hour by hour.

Then, with the field already selected, Selection Sunday could transcend the mere reading of a static list and instead be as entertaining and unpredictable as an NBA draft. The bracket could build itself before our eyes according to choices made in real time by 68 programs. Is it better for No. 3 seed Michigan to play in a bracket of death close to home or to head out West for a more favorable No. 6 seed? Why is a third party making this calculation? Let Juwan Howard decide, and let him do so live on CBS and on the clock.”

I know we've debated over the years the plusses and minuses of having the bracketing done via the kind of "draft" Gasaway proposes. Personally, I'd love it -- as it would greatly increase the drama and theatre of the the Selection Sunday process and provide much better debate fodder over the dozens of decisions the schools/coaches make about matchups and placement vs. carping about, e.g., the Committee hosing Duke by putting Michigan St. in our region, etc.

But, his point about the effect his proposed approach would have on the last couple of weeks of the regular season and conference tournament play is pretty legit too -- people would be able to see in real time and based on real data (not merely what Joe Lunardi or Jerry Palm or the Bracket Matrix predicts the Committee will do) of every win and loss by every team on the composition of the NCAA tournament field.

From an entertainment perspective, it'd be great. The major downside is basically teams would then be required to scout the rest of the NCAA teams to make those decisions as to where they want to be placed. Do we really want dedicated "scout coaches" across the country and schools dedicating that much time, effort, and money to figuring out the most favorable matchups? That sort of expertise is probably not something that should be at play in determining a champion even though it sounds like a lot of fun.

[Acymetric]

I have read that the NCAA wants a system that doesn't reward aggressively running up the score. The computer models generally reward that sort of behavior, but NET doesn't.

-jk

I thought at least some of the advanced stats sites capped things at a certain point for blowout losses, but maybe I'm mistaken. Certainly, that could be done. I've also never been entirely sure why this is such a problem anyway.

[tbyers11]

I thought at least some of the advanced stats sites capped things at a certain point for blowout losses, but maybe I'm mistaken. Certainly, that could be done. I've also never been entirely sure why this is such a problem anyway.

I think Ken Pom does cap margin of victory. But I think it is around 25 pts and it was because he didn’t see increased predictive value in a 35 pt win vs a 25 pt win.

I agree that I have no idea why the NET would want a cap around 10. To me, there is a big difference between winning by 20 vs 10. It’s not “running up the score” at that level.

[OldPhiKap]

I thought at least some of the advanced stats sites capped things at a certain point for blowout losses, but maybe I'm mistaken. Certainly, that could be done. I've also never been entirely sure why this is such a problem anyway.

When Steve Spurrier was at Florida, they ran up the score against UGA. The Georgia coach at the time (Ray Goff) was furious, and after the game Goff confronted Spurrier at mid-field and demanded to know why Steve had run up the score.

"Damn, Ray," responded Spurrier, "I didn't see in the rule book where it was my job to stop my team from scoring."

[Mtn.Devil.91.92.01.10.15]

From an entertainment perspective, it'd be great. The major downside is basically teams would then be required to scout the rest of the NCAA teams to make those decisions as to where they want to be placed. Do we really want dedicated "scout coaches" across the country and schools dedicating that much time, effort, and money to figuring out the most favorable matchups? That sort of expertise is probably not something that should be at play in determining a champion even though it sounds like a lot of fun.

I don't want tournament brackets dictated by "what system is most dramatic or entertaining."

No thank you.

[SCMatt33]

I think Ken Pom does cap margin of victory. But I think it is around 25 pts and it was because he didn’t see increased predictive value in a 35 pt win vs a 25 pt win.

I agree that I have no idea why the NET would want a cap around 10. To me, there is a big difference between winning by 20 vs 10. It’s not “running up the score” at that level.

Also remember that while the MOV component of NET is capped at 10 points per game, there is an entirely separate net efficiency component. This is based on raw efficiency (not adjusted for opponent strength) and there’s no cap on efficiency margin, so this is essentially an uncapped margin component, just adjusted for tempo.

[tbyers11]

Also remember that while the MOV component of NET is capped at 10 points per game, there is an entirely separate net efficiency component. This is based on raw efficiency (not adjusted for opponent strength) and there’s no cap on efficiency margin, so this is essentially an uncapped margin component, just adjusted for tempo.

Thanks for the info. I haven't dived that deeply into the NET. If I understand you correctly though, the MOV component is partially double-counting the some of the same data from the raw efficiency component. That is odd

[devildeac]

That therefore predicts UNC finishing the season from here 6-11, with a final conference record of 7-13.

Okay.

Still too many W for me.

9F.

[AGDukesky]

It is hard to believe the ACC is only a 4-5 bid conference after the preseason expectations. Unfortunately the middle/bottom of the conference has really earned it. Honestly there might only be three deserving teams in the league, but Virginia’s coaching and experience combined with how awful the rest of the teams look should be enough to get enough wins. Hopefully 1-2 teams can take advantage like State and maybe BC to make good resumes...

[Hingeknocker]

I don't want tournament brackets dictated by "what system is most dramatic or entertaining."

No thank you.

I totally agree. And one of the *problems* right now with the tournament bracketing process is that the committee *does* care about being more entertaining. Once they opened Pandora's Box and started geographically placing teams past the 1-line, they tainted the whole thing.

I would be in favor of continuing to have a committee that selects the teams, but seeding should be done via a straight 1-68 ranking of the teams selected, based on whatever your favorite metric/s would be. (And not NET.)

[SCMatt33]

Thanks for the info. I haven't dived that deeply into the NET. If I understand you correctly though, the MOV component is partially double-counting the some of the same data from the raw efficiency component. That is odd

It’s not that odd. There’s a lot of cross pollination between categories incorporated into the NET, especially on the results based side. The reality is that the NCAA is not very transparent on the exact calculation, so it’s hard to know how much each factor is weighted. Of the five categories they list, 3 are more about “did you win?” and the other two are about “How much did you win or lose by?”. Within that, each individual factor favors certain things over others. So having one that caps and one that doesn’t isn’t necessarily better or worse, but because one of the categories is specifically called “margin of victory” and that category is capped, it can lead to confusion that winning by more than the cap can’t help your NET, which isn’t true because it factors into another category. The link below is a primer article the NCAA released last year with some basic details.

https://www.ncaa.com/news/basketball...etball-ranking

[CameronBornAndBred]

While I don't disagree with Lunardi having only 5 ACC teams in the bracket, what's up him cramming them all into one region?
Duke is in the East, FSU in the South, and everyone else is plucked into the Mid-West, including play-in NCSU. I can't see the selection committee doing that. Two teams, sure, but not three and leaving one region ACC free.
Also, Lunardi has set up a potential Duke - SFA rematch.
http://www.espn.com/mens-college-bas...l/bracketology

[Nugget]

I totally agree. And one of the *problems* right now with the tournament bracketing process is that the committee *does* care about being more entertaining. Once they opened Pandora's Box and started geographically placing teams past the 1-line, they tainted the whole thing.

I would be in favor of continuing to have a committee that selects the teams, but seeding should be done via a straight 1-68 ranking of the teams selected, based on whatever your favorite metric/s would be. (And not NET.)

While that would be, objectively, fairer than the way they do it now, I seriously doubt it could happen because it would basically render moot the "selection show."

So, if the choice is between the current system of the bracketing being done by a committee of ADs -- who (non-transparently) sort of do it by the 1-68 rankings, but sort of do it by geography (and their guess as to which region would be perceived to be better for team x or y other than the overall #1 seed who gets to choose) and sort of do it by their rules about trying to avoid rematches (but not always) since the only rules they really have to adhere to are about keeping teams from the same conference out of the same region in seeds 1-4 and separated for a certain number of rounds depending on how often they played -- and one where each school, in the order of the straight 1-68 rankings, picks its spot in the bracket (presumably subject to the same rules about conference teams not being bracketed together), then I don't see any way it ends up worse than what we have now.

And to respond to the poster above about that somehow being undesirable because it would require coaches to spend time planning for which spot to pick, I don't see that as much of a downside. As with scouting generally, there would be some variance -- some would approach it analytically/methodically, others might apply some short of emotional decision (aiming for a particular matchup, like Wichita choosing to be an 8 seed in the Midwest to get a shot at Kansas rather than a 7 someplace else, or to avoid a particular matchup); but, I'm sure most coaches would simply pick based on the highest bracket spot they can get + geography, so wouldn't spend much time on the issue at all on the premise of "I don't care who we play, I'm worried about my team."

But, I'd much rather be debating the handful of interesting decisions that would occur -- "did Izzo choose to avoid Duke or be bracketed with Duke" than the current tiresome debate over who the Committee screwed.

[Acymetric]

While I don't disagree with Lunardi having only 5 ACC teams in the bracket, what's up him cramming them all into one region?
Duke is in the East, FSU in the South, and everyone else is plucked into the Mid-West, including play-in NCSU. I can't see the selection committee doing that. Two teams, sure, but not three and leaving one region ACC free.
Also, Lunardi has set up a potential Duke - SFA rematch.
http://www.espn.com/mens-college-bas...l/bracketology

They may not actually do it, but it falls within the rules about how early conference teams can meet based on how many times they played in the regular season/tournament (it happens to work out that it wouldn't matter if any of those teams met in the ACC tournament the way they're seeded).

[DavidBenAkiva]

With Kansas and now Michigan State losing this weekend, there's a new pecking order for 1 seeds.

Baylor, Duke, and Gonzaga appear to be clear-cut favorites as of now for 1 seeds in the South, East, and West, respectively. I would argue that Butler is in a pretty good spot for a 1 seed in the Midwest at the moment with San Diego State and Auburn worth considering as well. I guess the important factor here is that none of these teams are in a position to unseat Duke in the East region for the 1 seed.

Looking at the teams contending for a top 4 seed, Duke is in a really good position should the Blue Devils finish the season with the best record in the ACC. There just isn't another team in the northeast or Mid Atlantic that is there to challenge for the New York City regional. Maryland is the closest contender, but they would have to do a lot to truly contend. The same is true for Louisville.

A lot can and will change over the next 6 weeks, but I feel pretty good about how things look for Duke at the moment.

[duke2x]

Looking at the teams contending for a top 4 seed, Duke is in a really good position should the Blue Devils finish the season with the best record in the ACC. There just isn't another team in the northeast or Mid Atlantic that is there to challenge for the New York City regional. Maryland is the closest contender, but they would have to do a lot to truly contend. The same is true for Louisville. A lot can and will change over the next 6 weeks, but I feel pretty good about how things look for Duke at the moment.

Louisville would prefer Indianapolis by a large margin. It's the equivalent of UNC drawing Greensboro. The under the radar schools that would want NYC are Seton Hall and Villanova. Both are capable of winning the Big East and making a claim for a #1 seed at this point.

Winning this ACC isn't going to guarantee you a #1 seed, particularly if KS and MSU continue to struggle. I would not complain too much if we were the #2 seed in NYC.

[DavidBenAkiva]

Louisville would prefer Indianapolis by a large margin. It's the equivalent of UNC drawing Greensboro. The under the radar schools that would want NYC are Seton Hall and Villanova. Both are capable of winning the Big East and making a claim for a #1 seed at this point.

Winning this ACC isn't going to guarantee you a #1 seed, particularly if KS and MSU continue to struggle. I would not complain too much if we were the #2 seed in NYC.

I think you are right about Villanova and Seton Hall being options for the #1 seed in NYC, although that would require a lot of things to happen, such as Butler collapsing. Speaking of Butler, the reason I don't see Louisville being competitive for the Indy region #1 is that there are so many teams in the Midwest that have an inside track to that Midwest #1 Seed, including Butler, Michigan State, Kansas, and others. I think being in the Midwest region is an option for Louisville, but not as the #1 seed.

[budwom]

When Steve Spurrier was at Florida, they ran up the score against UGA. The Georgia coach at the time (Ray Goff) was furious, and after the game Goff confronted Spurrier at mid-field and demanded to know why Steve had run up the score.

"Damn, Ray," responded Spurrier, "I didn't see in the rule book where it was my job to stop my team from scoring."

Reminiscent of a Duke - UVA football game in which (IIRC which is perilous) Duke trailed 28-0 in the first quarter and lost by a huge margin. Crusty (but competent) UVA coach George Welsh continued to pour it on, and after the game said "I can't help it if they can't tackle." He was right, of course.