Page 2 of 2 FirstFirst 12
Results 21 to 40 of 40
  1. #21
    Join Date
    Feb 2007
    Location
    NC
    Quote Originally Posted by SkyBrickey View Post
    Here's what's broken to me. The committee is giving equal weighting to wins in Nov/Dec vs wins in Feb/Mar. They aren't allowing any consideration for teams improving over the season. This hurts Duke because we are so young and under Scheyer our teams have been playing our best ball late in the season. It also led to UVA being selected over Pitt this year. The best at large teams are not getting the at large bids.
    I think the problem here is that there isn't any sufficient evidence that results in more recent games are predictive of tourney success. We can point anecdotally to examples either for or against, but nothing I have seen suggests there is a stronger correlation between late-season and tourney results than there is between full season and tourney results.

    Additionally, if you weighted the later games, you would further devalue the only interconference data we have.

    I think adding interconference games later in the season would be the more appropriate approach.

  2. #22
    Join Date
    Feb 2007
    Location
    NC
    I guess the other big comment I have is that, aside from the fact that we're basing all of this on essentially 3 anomalies (2022 UNC, 2022 Miami, 2024 NC St), there is the logical fallacy that tournament results mean more or fewer teams should have made the tournament. Even if we take the position that the results of the 2022 and 2024 tournaments (note: I'm intentionally ignoring 2023 as I don't think it shows any evidence that the conference was underrated) show that the ACC was underrated in those years, it doesn't then inherently suggest that the ACC deserved more teams. It would just mean that those teams that were given a bid may have deserved a higher seed.

    Now, it's certainly possible that underrating a conference could mean that more teams should have gotten in. But only if the omitted teams were fairly close to the edge of getting in. If the committee felt that the next best ACC team was 10 spots outside of the field, then who is to say that that team deserved to be in the field? For an extreme example, back in the late-90s/early-00s when Gonzaga was destroying the WCC and making surprising tourney runs, that didn't mean the WCC deserved more bids. It was still a 1-bid quality league at the time.

    So, again, I agree that accurately seeding the field is a fool's errand. I just don't think the proposal of looking back at prior years' tourney results is the right solution. I think rather than using an arguably more flawed approach, I would agree that that increasing the number of interconference games in January and February would be the best solution to the problem. I'd also support the recommendation here of a ".500+ record in conference" requirement for at large bids. But relying on data from prior seasons doesn't seem very appropriate to me, and would in my opinion lead to the committee chasing its tail moving forward. See 2022 vs 2021 for example. If we relied on 2021 to inform seeding of the 2022 tournament, the PAC-12 would have gotten more teams in and the ACC fewer teams. So if you felt that the 2022 ACC was underrepresented, using the historical data perspective would then exacerbate the problem.

    If you add more interconference games in January/February, you can reduce the quality challenges of comparing across conferences. If you set the minimum of a .500 record in conference, you reduce the risk of a 9th/10th team getting in from one conference. Both of those changes would theoretically improve the process while still making the tournament all about an assessment of THAT SEASON'S teams.

  3. #23
    Quote Originally Posted by CDu View Post
    I guess the other big comment I have is that, aside from the fact that we're basing all of this on essentially 3 anomalies (2022 UNC, 2022 Miami, 2024 NC St), there is the logical fallacy that tournament results mean more or fewer teams should have made the tournament. Even if we take the position that the results of the 2022 and 2024 tournaments (note: I'm intentionally ignoring 2023 as I don't think it shows any evidence that the conference was underrated) show that the ACC was underrated in those years, it doesn't then inherently suggest that the ACC deserved more teams. It would just mean that those teams that were given a bid may have deserved a higher seed.

    Now, it's certainly possible that underrating a conference could mean that more teams should have gotten in. But only if the omitted teams were fairly close to the edge of getting in. If the committee felt that the next best ACC team was 10 spots outside of the field, then who is to say that that team deserved to be in the field? For an extreme example, back in the late-90s/early-00s when Gonzaga was destroying the WCC and making surprising tourney runs, that didn't mean the WCC deserved more bids. It was still a 1-bid quality league at the time.

    So, again, I agree that accurately seeding the field is a fool's errand. I just don't think the proposal of looking back at prior years' tourney results is the right solution. I think rather than using an arguably more flawed approach, I would agree that that increasing the number of interconference games in January and February would be the best solution to the problem. I'd also support the recommendation here of a ".500+ record in conference" requirement for at large bids. But relying on data from prior seasons doesn't seem very appropriate to me, and would in my opinion lead to the committee chasing its tail moving forward. See 2022 vs 2021 for example. If we relied on 2021 to inform seeding of the 2022 tournament, the PAC-12 would have gotten more teams in and the ACC fewer teams. So if you felt that the 2022 ACC was underrepresented, using the historical data perspective would then exacerbate the problem.

    If you add more interconference games in January/February, you can reduce the quality challenges of comparing across conferences. If you set the minimum of a .500 record in conference, you reduce the risk of a 9th/10th team getting in from one conference. Both of those changes would theoretically improve the process while still making the tournament all about an assessment of THAT SEASON'S teams.
    Every time I see one of these threads talking about how the ACC being underrated and overperforming in the tournament, I just wait for CDu to show up and talk about sample size and random events.

    You absolutely aren't wrong. But it's also hard to counter how these things "feel" to most of us.

  4. #24
    Join Date
    Feb 2007
    Location
    NC
    Quote Originally Posted by Mtn.Devil.91.92.01.10.15 View Post
    Every time I see one of these threads talking about how the ACC being underrated and overperforming in the tournament, I just wait for CDu to show up and talk about sample size and random events.

    You absolutely aren't wrong. But it's also hard to counter how these things "feel" to most of us.
    LOL, guilty as charged.

  5. #25
    Quote Originally Posted by CDu View Post
    LOL, guilty as charged.
    To be clear - very much appreciate your sobering comments that tone things down. Lol.

  6. #26
    Quote Originally Posted by CDu View Post
    I think the problem here is that there isn't any sufficient evidence that results in more recent games are predictive of tourney success. We can point anecdotally to examples either for or against, but nothing I have seen suggests there is a stronger correlation between late-season and tourney results than there is between full season and tourney results.

    Additionally, if you weighted the later games, you would further devalue the only interconference data we have.

    I think adding interconference games later in the season would be the more appropriate approach.
    I've never seen the analysis either but I'd be shocked if it doesn't reveal that, yes, some teams improve more than others during the season. Quantifying it and rewarding it is admittedly a whole other challenge...

  7. #27
    Join Date
    Feb 2007
    Location
    Los Angeles
    Quote Originally Posted by House P View Post
    While I don't think the Nov/Dec information is as useless as you seem to think, it is certainly far less than ideal. Based on results that you point out from the last 3 years, I am also coming around to the idea that the ACC may be undervalued by the committee. I am biased, but Duke seemed better than a 5 and 4 seed the past two seasons.

    On the other hand, the ACC was underwhelming in 2021 (4 total wins from 7 teams which were expected to win 6.6 total games based on seeding) and the ACC hasn't exactly excelled in the NIT in recent years.
    Seems a little hard to make the argument that we were better than a 5 seed in 2023 when we were soundly beaten in the second round by the 4 seed.

    Bigger picture on this thread I agree that using past years' results makes no sense in choosing teams or seeding them in subsequent years. It would have made little sense in the pre-transfer crazy world but makes even less sense now. Teams are different year to year; conferences are different year to year. There does not appear to be any reliable evidence that any team or conference consistently either overperforms its evaluation by the committee or underperforms it. There is a large degree of randomness to it, and that's part of what makes the whole thing so fun.

    More late season interconference games? Definitely. Can only help. I'd go so far as to require P6 teams to play at least two post-Jan 15 games vs other P6 teams in order to be eligible to receive an at large bid. No loading up against the dregs of mid or low major conferences, and tying it to quadrant placement is too far out of the hands of the team in question.

  8. #28
    I think the best solution is to start conference play a week earlier, then in the middle of the season, sometime in FEB, have a week where no conference games are allowed, as a window for everyone to play 2 out of conference games.

  9. #29
    Join Date
    Feb 2007
    Location
    NC
    Quote Originally Posted by SkyBrickey View Post
    I've never seen the analysis either but I'd be shocked if it doesn't reveal that, yes, some teams improve more than others during the season. Quantifying it and rewarding it is admittedly a whole other challenge...
    Just to be clear, I wasn't saying that teams don't get better (or worse). I'm saying that I don't think there's data to suggest that getting better results as the season goes on has any significant impact on tournament performance. And I guess adding that determining what is truly getting better vs just random fluctuation is really hard to figure out.

  10. #30
    Quote Originally Posted by CDu View Post
    I think the problem here is that there isn't any sufficient evidence that results in more recent games are predictive of tourney success. We can point anecdotally to examples either for or against, but nothing I have seen suggests there is a stronger correlation between late-season and tourney results than there is between full season and tourney results.

    Additionally, if you weighted the later games, you would further devalue the only interconference data we have.

    I think adding interconference games later in the season would be the more appropriate approach.
    Also, we can't really know whether the committee "is giving equal weight to games in Nov/Dec. vs. wins in Feb/March." It's certainly correct that the computer rankings they look at weight games equally, and that there is no express instruction they give greater weight to more recent games (having removed the "last 12 games" line item from the team charts). But, that doesn't mean individual committee members in their assessments/votes don't choose as part of their "eye test" to give greater weight in their assessments of the teams to who is hot and who isn't.

  11. #31
    Quote Originally Posted by CDu View Post
    Just to be clear, I wasn't saying that teams don't get better (or worse). I'm saying that I don't think there's data to suggest that getting better results as the season goes on has any significant impact on tournament performance. And I guess adding that determining what is truly getting better vs just random fluctuation is really hard to figure out.
    I think your key word here is "significant". Because the definition of better means they would do better than how the NCAA seeds them based on equally weighting early season games.

    Two examples close to home this year. NCSU was a different team at the end of the year and clearly better than an 11 seed. And Pitt was pretty clearly a better team than UVA and would have had a better shot at winning that play in game.

    I bet if someone did an analysis using the Torvik trending data on how the seed performs relative to seeding you'd see a pretty strong correlation. The Vegas lines on these games definitely give credit to "hot teams" like NCSU and penalize "cold teams" like UVA.

  12. #32
    Join Date
    Feb 2013
    Location
    Cambridge, MA
    Quote Originally Posted by CDu View Post
    I think the problem here is that there isn't any sufficient evidence that results in more recent games are predictive of tourney success. We can point anecdotally to examples either for or against, but nothing I have seen suggests there is a stronger correlation between late-season and tourney results than there is between full season and tourney results.

    Additionally, if you weighted the later games, you would further devalue the only interconference data we have.
    One of the last 37 times* this topic was brought up, I took a look at how well KenPom ratings predicted NCAA/NIT tourney results vs how well they predicted conference tourney results. The hypothesis was that KenPom should be better predicting conference tourney results as there are a whole bunch of recent games comparing the teams within a conference, while there are fewer comparisons between conferences (and essentially none since December).

    In the end, KenPom didn't seem to do any better at predicting conference tourney games than inter-conference post season games. That suggests to me that, while imperfect, there is meaningful information from the Nov/Dec games.

    That being said, I agree with MtnDevil's comment that it still "feels" like the ACC got a raw deal the past couple years. As such, I wouldn't be entirely opposed to an objective criteria which rewards conferences that performed well in previous tournaments vs those who underperformed. In the end, the difference between the 40th best team and the 75th best is probably less than the measurement error in the current metrics, so I don't think giving the ACC the benefit of the doubt next year is objectively "wrong".

    I also think the NIT champ should get an automatic bid to the next NCAA tourney. Sure, this would mean that an otherwise "undeserving" team would regularly make the field. I'd argue that winning the previous NIT is more worthy of being rewarded with a tourney bid than finishing 6th in the Mountain West. This would also raise the profile of the NIT and fewer teams would opt out.



    *To be clear, I am not complaining that this topic keeps getting brought up. I'm pretty sure I have responded each time and it's the offseason, so what else are we gonna talk about.

  13. #33
    Join Date
    Feb 2013
    Location
    Cambridge, MA
    Quote Originally Posted by tommy View Post
    Seems a little hard to make the argument that we were better than a 5 seed in 2023 when we were soundly beaten in the second round by the 4 seed.
    I said I was biased.

    Duke's round of 32 loss doesn't bolster its case, but Duke was very strong last year once Lively got rolling. Per Torvik, Duke played like a top 10 team over the final 2 months of the regular season last year and played like a top 5 team over the final month. Duke entered the tourney ranked 12 in the AP poll. Losing a game without a starter to a 4 seed doesn't necessarily discount it anymore than Houston losing to Duke without Shead for the final 24 minutes discounts their 1 seed.

    Quote Originally Posted by tommy View Post
    More late season interconference games? Definitely. Can only help. I'd go so far as to require P6 teams to play at least two post-Jan 15 games vs other P6 teams in order to be eligible to receive an at large bid. No loading up against the dregs of mid or low major conferences, and tying it to quadrant placement is too far out of the hands of the team in question.
    I agree that there is some risk in requiring a Quad 3 game, but it shouldn't be all that difficult to find a team which is going to finish in the top 160 (home game) or top 240 (road game) if you have 2+ choices. In reality, the would probably cause most teams to schedule Quad 1 and Quad 2 games. I wonder how many teams in KenPom's preseason top 75 ended up finishing outside the top 240 in the NET?

    If you want to remove all doubt, you could use the previous season's NET ratings (but I'd probably require a Quad 2 game in that case).

  14. #34
    Quote Originally Posted by CDu View Post
    If you add more interconference games in January/February, you can reduce the quality challenges of comparing across conferences. If you set the minimum of a .500 record in conference, you reduce the risk of a 9th/10th team getting in from one conference. Both of those changes would theoretically improve the process while still making the tournament all about an assessment of THAT SEASON'S teams.
    I agree that both of these would improve the process. One other possibility (consistent with CDu's first suggestion for more late-season non-con games, but may render the .500 conference record limit unnecessary) would be to try John Gasaway's proposal to eliminate the Committee entirely and go to a system based solely on:

    (i) keeping auto bids for conference champs;

    (ii) "selection" of the at-large bids based on Wins Above Bubble or Strength of Record, or "a simple win proxy built atop the NET rankings" (which he asserts could be made "agnostic on a given game's margin of victor and instead assign a weight to each win based on opponent strength and game location," or any number of other objective, metrics-based approaches that the NCAA/NABC could develop -- the important part being they would be transparent and disclosed in advance of the season so everyone would know how it would work and relative rankings would be trackable every day of the season;

    (iii) seeding of both groups based entirely on whatever that agreed metric is; and

    (iv) bracketing determined at the time of the current selection show via teams selecting their place in the bracket in order of their 1-68 rankings.

    https://johngasaway.com/2019/12/18/r...ketball-games/

    https://johngasaway.com/2024/03/16/why-selection/

    And, if there was enough of a consensus that the agreed metric should give greater weight to more recent results to capture "improvement" over the course of the season, presumably it would be easy to build in something that lessens the weighting each result slightly and incrementally as it recedes in time-- for example, weighting the most recent result at 1, the prior game at 0.3% less, the prior game at 0.6% less, etc., so that by the end of the season the first 10 games would be weighted at around .87-.90 and the most recent 10 at .97-1.

  15. #35
    Join Date
    Feb 2007
    Location
    NC
    Quote Originally Posted by SkyBrickey View Post
    NCSU was a different team at the end of the year and clearly better than an 11 seed.
    But there was no way of knowing before the tournament whether they were better than an 11 seed. Yes, they won the ACC tournament. But prior to that they had lost 4 in a row. A 5-4 record going into the NCAA tournament isn't exactly strong evidence that they should be seeded higher than 11. The only evidence that they were better was a 5-day stretch in the ACC Tournament. And we know that conference tourney success doesn't inherently translate to NCAA Tournament success.

  16. #36
    Quote Originally Posted by SkyBrickey View Post
    I think your key word here is "significant". Because the definition of better means they would do better than how the NCAA seeds them based on equally weighting early season games.

    Two examples close to home this year. NCSU was a different team at the end of the year and clearly better than an 11 seed. And Pitt was pretty clearly a better team than UVA and would have had a better shot at winning that play in game.

    I bet if someone did an analysis using the Torvik trending data on how the seed performs relative to seeding you'd see a pretty strong correlation. The Vegas lines on these games definitely give credit to "hot teams" like NCSU and penalize "cold teams" like UVA.
    Bingo, you figured it out. Let Vegas pick the teams with a comittee of the largest sports betting houses.
    They know all.

  17. #37
    Quote Originally Posted by CDu View Post
    I think the problem here is that there isn't any sufficient evidence that results in more recent games are predictive of tourney success. We can point anecdotally to examples either for or against, but nothing I have seen suggests there is a stronger correlation between late-season and tourney results than there is between full season and tourney results.

    Additionally, if you weighted the later games, you would further devalue the only interconference data we have.

    I think adding interconference games later in the season would be the more appropriate approach.
    I believe there was some of that with a couple of the mid-majors having a "challenge" in February Mid-American - Sun Belt Challenge. The top teams played the corresponding top teams in the other conference all down the line.

    Maybe the ACC-SEC challenge should be pushed to February?
    Last edited by Lurkingdukedog; 04-18-2024 at 04:58 PM.

  18. #38
    Join Date
    Feb 2013
    Location
    Cambridge, MA
    Quote Originally Posted by Lurkingdukedog View Post
    I believe there was some of that with a couple of the mid-majors having a "challenge" in February Mid-American - Sun Belt Challenge. The top teams played the corresponding top teams in the other conference all down the line.

    Maybe the ACC-SEC challenge should be pushed to February?
    As far as I can tell, there were 20 games involving teams from difference conferences from Jan 7 through March 17. Eleven of those games involved the Sun Belt-MAC challenge and six involved independent Chicago State. The only game involving teams potential at-large teams was the Gonzaga-UK game on Feb 10.

    In other words 30% of the information the committee had to compare the relative strengths of conferences after early January came from games involving Chicago State.

    By the way, as the only D1 Independent, Chicago State had an odd schedule. They managed to cobble together 8 games between Jan 8 and Feb 18*. At this point their season went on pause. They somehow finagled a spot in the CBI tournament despite a 9-18 record vs D1 teams and resumed games on March 23 where the upset 21-11 UC San Diego before losing their next game on March 25.



    *The played 5 games vs D1 teams and 3 games vs D2 teams.

  19. #39
    Join Date
    Feb 2007
    Location
    Steamboat Springs, CO

    Now Adjusting for Seeds by Conference

    Here's some more...

    It's not sufficient to look at just "wins by conference." We need to look at whether conference partiicpants performed up to their seed. As you remember there are projected "wins by seed" calculated by Bracket Odds at the U. of Illinois. So I went through the exercise of measuring wins by each member of the P-5 conferences against their projected wins based on their seed.

    Anyway, the SEC, Big Ten, and Big 12 -- the "Bid Hogs," as I am calling them -- underperformed their seed, while the ACC and Big East performed better than their seeds. Here are the net wins for each of the five conferences for 2022 through 2024.

    NCAA "Bid Hogs"
    Code:
    	
    SEC	
    2022	 (5.2)
    2023	 (0.7)
    2024	 (2.6)
    Total	 (8.4)
    Big 12	
    2022	 2.3 
    2023	 (2.6)
    2024	 (5.6)
    Total	 (5.9)
    Big Ten	
    2022	 (1.4)
    2023	 (3.3)
    2024	 1.8 
    Total 	(3.0)
    Other P-5
    Code:
    ACC	
    2022	 8.9 
    2023	 1.7 
    2024	 4.7 
    Total	 15.2 
    Big East	
    2022	 0.0 
    2023	 4.6 
    2024	 2.5 
    Total	 7.1
    The Bid Hogs won 17 fewer games than projected in the last three tournaments.

    The ACC won 15 more and the Big East 7 more than projected based on seeds.

    I also did this with z-scores. But the point is, whether through accident or scheming or gaming the system, the ACC is getting screwed in tournament bids and almost certainly seeding levels vis a vis the Bid Hogs.

    Defense of the current system using only some perturbation of current-year results is clearly in the best interests of KenPom, Torvik and other who are making a living off of this woefully inadequate date. But I think it's leading to bad results and hurting the ACC and Duke.
    Sage Grouse

    ---------------------------------------
    'When I got on the bus for my first road game at Duke, I saw that every player was carrying textbooks or laptops. I coached in the SEC for 25 years, and I had never seen that before, not even once.' - David Cutcliffe to Duke alumni in Washington, DC, June 2013

  20. #40
    Quote Originally Posted by House P View Post
    As far as I can tell, there were 20 games involving teams from difference conferences from Jan 7 through March 17. Eleven of those games involved the Sun Belt-MAC challenge and six involved independent Chicago State. The only game involving teams potential at-large teams was the Gonzaga-UK game on Feb 10.

    In other words 30% of the information the committee had to compare the relative strengths of conferences after early January came from games involving Chicago State.

    By the way, as the only D1 Independent, Chicago State had an odd schedule. They managed to cobble together 8 games between Jan 8 and Feb 18*. At this point their season went on pause. They somehow finagled a spot in the CBI tournament despite a 9-18 record vs D1 teams and resumed games on March 23 where the upset 21-11 UC San Diego before losing their next game on March 25.



    *The played 5 games vs D1 teams and 3 games vs D2 teams.
    Wow!! That's quite a spreadsheet. Thanks for compiling that

Similar Threads

  1. Is this NYT April Fool’s Joke or is the NCAA a total joke
    By SoCalDukeFan in forum Elizabeth King Forum
    Replies: 13
    Last Post: 04-01-2024, 05:51 PM
  2. Bracketology: a Fool’s Errand
    By -jk in forum Elizabeth King Forum
    Replies: 2
    Last Post: 03-10-2020, 01:26 PM
  3. The NCAA Tournament Selection Committee is on a Fool's Errand
    By sagegrouse in forum Elizabeth King Forum
    Replies: 62
    Last Post: 03-20-2015, 12:38 PM
  4. The Tournament Selection Committee is on a Fool's Errand
    By sagegrouse in forum Elizabeth King Forum
    Replies: 24
    Last Post: 01-09-2012, 01:07 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •