Tar Heel Seed Comparison

By Adam Lucas | 26 Comment(s) | Posted

Ultimately, every team has to play good opponents to advance in the NCAA Tournament. But seeding is important in determining exactly when and where you have to play those teams. With that in mind, here's a look at the credentials at some of the teams handed certain seeds on Sunday night's selection show. This table (using data from WarrenNolan.com) compares the credentials of Carolina, an 8 seed, with the four 7 seeds. According to the official NCAA seed list, the committee ranked all four of the below teams ahead of Carolina:

TeamD1 recordSeedRPISOSvs top 100Last 10
Carolina 23-10 8 17 8 9-9 8-2
SDSU 20-10 7 30 15 8-10 5-5
Notre Dame 25-9 7 35 49 9-9 6-4
Illinois 21-12 7 40 9 7-10 6-4
Creighton 27-7 7 25 82 10-5 7-3
Memphis 30-4 6 12 73 9-4 9-1
Arizona 25-7 6 15 25 12-6 5-5
UCLA 25-9 6 26 20 13-6 7-3
Butler 25-8 6 22 32 10-8 6-4

 

When asked about the bracketing process on Sunday night's selection show, NCAA selection committee chairman Mike Bobinski said, "A lot of mechanical things happen," when the committee has to slide teams one line higher or lower. He went on to say, "We don't look at matchups." 

Looking at the above table, and looking at the fact that the NCAA coincidentally managed to potentially set up a Carolina-Kansas matchup in Kansas City, it's hard not to have a very clear understanding of exactly what some of those "mechanical things" might be--starting with the mechanics of cashing checks from the tournament's $10.8 billion contract with Turner and CBS, a contract that requires high ratings in order to be successful. 

It's quite a coincidence that Carolina and Kansas have been in the same region in three of the nine Tar Heel NCAA Tournaments (in which UNC has participated) during the Roy Williams era. Prior to that, the two programs hadn't been in the same region since 1985, which means 18 tournaments passed without the teams being paired, and suddenly they show up in the same region three times in nine years. There must have been a lot of "mechanical things" that caused that phenomenon.

Comments

  1. Hal garland's avatar
    Hal garland
    | Permalink
    It's all about Roy and Kansas .miami gets a cakewalk .money for TV is what is driving this .state got shafted as well .
  2. BFA's avatar
    BFA
    | Permalink
    Apparently, their "mechanics" don't jibe with the numbers.
  3. Jesse's avatar
    Jesse
    | Permalink
    3 in 9 is 33%. Random change would have UNC and Kansas be in the same bracket 25% of the time, so it's not suspicious at all.
  4. Steve's avatar
    Steve
    | Permalink
    Of course they got hosed. It clearly has nothing to do with the fact that they have 1 win all season against a top 25 team (home against UNLV). Compare that with Illinois' wins against Gonzaga, Indiana, Ohio St. But clearly since they had a better record in their last 10 UNC was the better team this year.
  5. Bill's avatar
    Bill
    | Permalink
    The NCAA is pretty hidebound and must still use a mechanical cash register.
  6. Matt's avatar
    Matt
    | Permalink
    What is suspicious is that UNC is legitimately a 6 seed, and possibly a 5 seed.

    An 8 seed is seriously out of line. It is fair to neither KU nor UNC to face each other in round 2.

    UNC should face off against a 3 or 4 seed in round 2, not a 1, and KU should get a team with a 30-40 RPI, not a top 20 RPI.

    Not fair to either squad.

    KU will get the luxury of playing in KC, while UNC gets a relatively easy 9 seed against a Villanova team that is a borderline top 50 RPI squad...but that just clarifies the committees intentions even more.

    Dollars reign.
  7. Dave S's avatar
    Dave S
    | Permalink
    The Lunacy of SOS is getting maddening. For example, How many teams that scheduled KY had their SOS artificially raised because it was way over ranked as a 3 even with Noels in the lineup?

    Pre-conference games are given far to much weight as well.

    Almost all of Dukes signature wins are home pre conference .. go on the road and get bad beats by unranked teams and BLOWN OUT by Miami

    yet they have an SOS of 1

    To much manipulation is being done with the scheduling and pre season rankings
  8. Dean's avatar
    Dean
    | Permalink
    This is team that went like 2-9 against tourney teams. Even if you include Virginia and Maryland thats only 6-14. Our best road win was what, Maryland? Notre Dame beat Louisville,Illinois beat Gonzaga, SDSU beat New Mexico, heck Creighton beat Wisconsin. Our best win was UNLV at home? State at home? Not that much worse than SDSU's or Creighton's but not that much better. Our body of work is just not that impressive, though the eye test would make one think that we should be higher certainly. Yea I think we could have been a 6 or a 7, but an 8 is not unreasonable, just bad luck.
  9. Tk's avatar
    Tk
    | Permalink
    I personally do believe that UNC was under seeded. In my mind they were worst case on the #7 line unless they beat Miami.

    The committee blatantly dropped Carolina a line or two(depending on who you want to believe) for a potential "Roy Bowl" match-up in the 2nd round. Really unfair to North Carolina, and also to Kansas.
  10. jjackflash's avatar
    jjackflash
    | Permalink
    Adam,
    Your article is dead on the money. This is a hosejob of the highest order. Personally, I wanted to see a UNC- Kansas rematch, as I thought Kansas would be THE WEAKEST TWO SEED!! (And, I suspected the committee would screw UNC by making them a #7, when they deserved a #5 or #6.) However, it's just laughable that the committee felt like they could screw up the seedings this badly & send UNC out to Kansas City to play KU in a 1-8 matchup.

    Oh, and for accuracy's sake, it should be noted that UNC's record is 24-10, NOT 23-10.
  11. Matt's avatar
    Matt
    | Permalink
    Jesse, you suck at probabilities.

    There is a 25% chance that that two teams will be placed in the same region in any given year.

    However, there is a 0.83% chance that two teams will be placed in the same region 3 times in 9 years.

    (9/3) * (1/4)^3 * (3/4)^6

    http://www.sparknotes.com/testprep/books/sat2/math1c/chapter4section8.rhtml

    If this is just dumb luck, then the committee needs to start buying powerball tickets.
  12. Eric's avatar
    Eric
    | Permalink
    Wow, Matt, you're the one that sucks at probabilities.

    There is a 23.4% chance of being in the same bracket 3 out of 9 years.

    The correct formula is (9 choose 3) * (1/4)^3 * (3/4)^6.

    In fact, 2/9 is the only more probably outcome than 3/9.

    0/9 = 7.5%
    1/9 = 22.5%
    2/9 = 30.0%
    3/9 = 23.4%
    4/9 = 11.7%
  13. Will's avatar
    Will
    | Permalink
    The Kansas thing is the kind of paranoid stuff I'd expect to see on a State message board.

    As noted by others, there's nothing particularly suspicious about being in Kansas' region 3 times in 9 years. We've actually been in the same region as Oklahoma 3 times just since 2008. OMG SCANDAL!!!

    As far as UNC's seeding, I thought we'd be a little higher. But the chart is misleading. For one thing, last 10 games is not a criteria that the committee officially considers (though we can debate whether that makes sense). Clearly Carolina would have benefited if more weight were given to finishing strong.

    But more importantly, you list top 100 record, but not top 50 or top 25. Of course, Carolina's record in those ranges is not nearly as impressive. "Quite a coincidence," as you might say.
  14. Will's avatar
    Will
    | Permalink
    Also, Dave S. - while there are plenty of legitimate complaints about SoS and the RPI in general, you can at least rest assured that preseason rankings (or, for that matter, any other poll rankings) have nothing to do with it. They are not a factor in determining SoS.
  15. Carolyn foster's avatar
    Carolyn foster
    | Permalink
    Carolina was done so wrong,in my thoughts they should have been 6.how could Duke been 2 I'll never understand.
  16. Carolyn foster's avatar
    Carolyn foster
    | Permalink
    Carolina was done so wrong,in my thoughts they should have been 6.how could Duke been 2 I'll never understand.
  17. JohnHopkins's avatar
    JohnHopkins
    | Permalink
    Do not forget that San Diego St is a 7 seed despite a 9-7 MWC record and no non-conf. win over a ranked team. 31 RPI Lost both times to UNLV
  18. Michael Hosking's avatar
    Michael Hosking
    | Permalink
    Adam's writings are frequently fun and insightful, but this one is a gross miscarriage of journalistic integrity. The stats he uses are misguided and only a tiny subset of a nitty-gritty report. SOS only is useful if you beat some of the good teams you play. I can make a decent argument for us to be a 7 seed, but 1 and 7 versus the top 25, 2 and 8 versus the top 50, does that sound like a top 25 team (approximately required for a 6 seed or better)?
  19. Michael Hosking's avatar
    Michael Hosking
    | Permalink
    90% of statistics you see on a message board are wrong; this is not such a post. I have a MS in Stat/OR from UNC and am finishing a PhD in Operations Research, my credentials are fine. Regarding the posts of Jesse, Matt, and Eric:

    Oddly despite his view being the most simplistic, Jesse is the closest to being correct in analyzing the situation. What Jesse is doing is taking the empirical statistic (3/9) and comparing it to the theoretical ideal (1/4). Matt an Eric both took the more complex route modeling the process as a Binomial distribution. Matt used the formula incorrectly whereas Eric used it properly, hence the differences in their results. However all the analysis so far is lacking.

    The main weakness so far is that in years where UNC and KU have the same seed they cannot be in the same region. Particularly, there are 2 years (07 and 08) where both teams got a 1 seed. This means the analysis should proceed using the 7 years UNC and KU were both in the tourney with different seeds. That is critical if the question we desire to answer is “Does the committee abnormally place UNC and KU together when possible.”

    The probability of UNC being in the same region as KU X out of Y years is:
    (Y choose X) * ((1/4)^X) * ((3/4)^(Y-X))
    This formula gives us the following for Y = 7 (the correct value, not 9)
    X Prob
    0 .1335
    1 .3115
    2 .3115
    3 .1730
    4 .0577
    5 .0115
    6 .0013
    7 <.0001<br />
    From this we can see that the chance of seeing as result as unlikely as 3 of 7 or more by chance is more than 20%. This has no statistical significance. While this data is a cute anecdote, it cannot be safely said to be a trend, it is currently a blip.

    There is even more advanced analysis which could consider how far apart UNC and KU are placed in the bracket. I have already wasted far too much time on this, so I will leave that for someone else to pursue.

    Step your math game up…don’t bring an abacus to a graphing calculator fight.
  20. Will's avatar
    Will
    | Permalink
    LOL - good stuff Michael.

    I would also say that Adam's point about UNC not being in Kansas' region before Roy came here actually works against his argument. That is, if we're to believe the committee is conspiring to create TV-friendly storylines, do you not think Dean Smith vs. his former assistant Roy Williams would have qualified? Or Dean vs. his former player Larry Brown? What about Roy vs. the caretaker Guthridge, with the UNC opening looming and Roy the obvious first choice? Both UNC and KU were 8 seeds in 2000 - wouldn't matching them up in an 8/9 game have been a good way to spice up an otherwise nondescript part of the bracket?
  21. Michael Hosking's avatar
    Michael Hosking
    | Permalink
    Will, I agree. My conspiracy embracing and cynical side feels that setting up storylines is good for business and something I would want to do if the long term value of the tourney property was in my best interests (as is the case for everyone on the seeding committee). However, considering the difficulty of setting up this tourney essentially in real time means that such concerns may be too difficult to enforce. I wish I could have a tape of the conversation in the room…

    Another factor which I did not consider in my previous analysis is that the seeding rules make it somewhat more likely to play someone outside your conference due to the rules against playing team in your conference early on. This would further drive up the odds of UNC and KU being in the same region.
  22. Michael Hosking's avatar
    Michael Hosking
    | Permalink
    There is one other significant flaw I see in my analysis; I am taking the seeding numbers given as gospel. Who gives those numbers? The exact same people who put teams in regions. If one were looking to argue that the committee sets up matchups they could "tweak a seeding to help allow putting two teams on the same seed line into the same region by bumping one team up or down (exactly like Will mentioned) This is another level of analysis which would be very time consuming and probably ultimately something which could only every be subject to the eye of the beholder. I would be interested in anyone who could show us maybe other anecdotes of this kind of behavior.
    Examples:
    -moving UK near Duke after the Asshole-who-shall-not-be-named’s shot
    -UCLA and Valpo
    -Rick Pitino/John Calipari/Tubby Smith's old team vs. current team

    As many statisticians like to say: "what do you call a bunch of anecdotes? Data."
  23. Michael Hosking's avatar
    Michael Hosking
    | Permalink
    Whoa, I should have looked that post over, lots of errors. Corrected version (I hope):

    There is one other significant flaw I see in my analysis; I am taking the seeding numbers given as gospel. Who gives those numbers? The exact same people who put teams in regions. If one were looking to argue that the committee sets up matchups for storylines they could "tweak” a seeding to allow putting two teams which should be on the same seed line into the same region by bumping one team up or down (exactly like Will mentioned). This is another level of analysis which would be very time consuming and probably ultimately subject to the eye of the beholder. I would be interested in anyone who could show us other anecdotes of this kind of behavior.
    Examples:
    -moving UK near Duke after the Asshole-who-shall-not-be-named’s shot
    -UCLA and Valpo
    -Rick Pitino/John Calipari/Tubby Smith's old team vs. current team

    As many statisticians like to say: "what do you call a bunch of anecdotes? Data."
  24. Adam Lucas's avatar
    Adam Lucas
    | Permalink
    Wow, lot of good comments here. Wish I had checked in sooner.

    I don't particularly mind that the NCAA clearly sets up matchups and has for years (unless we all believe it was coincidence that UNC's opponent in Dean Smith's record-breaking 877th win was supposed to be Bob Knight, until his Hoosiers struggled against Colorado). They put together compelling March programming and their TV partner is a huge part of that process. It's not wrong to run a good business.

    But I do mind them lying about it. All they have to say is, "We want to create the best tournament we possibly can, and that includes potentially creating games that will attract the attention of both hardcore basketball fans and the casual fan." I can't argue with that at all--and that's the truth, whether they say it out loud or not.

    As for the seeding, I don't think Carolina should have been a 6. I do think they should've been a 7. But hey, let's say the Tar Heels beat Villanova and then beat Kansas--we'll all be thanking the committee for creating one of the most memorable non-title postseason wins in the Roy Williams era.
  25. Will's avatar
    Will
    | Permalink
    Thanks for checking in Adam.

    To be clear, I agree with you that Carolina should have been a 7. I just don't think it's a travesty to be off by 1 line either way. A travesty is what they did to Oregon, who is about 5 lines too low.

    As far as storylines go, let me give you an alternate view - in a field of 68 teams, there are countless connections among the various teams, coaches, schools and players. It would be all but impossible to arrange the bracket in such a way that none of the potential games before the Final Four would have any sort of compelling personal angle - and even if you could, why would you?

    So, it's inevitable that there are going to be potentially compelling matchups in the early rounds. If you're inclined to see conspiracies, then you see those matchups as evidence of such a conspiracy.

    If you're not, then you recognize that every year there are going to be big-name programs who have slightly down years that make them candidates for the 8/9 game. And pretty much every year, most of the 1 seeds are also going to be big-name programs. So unless you want the 8/9 games to be entirely populated by mid-major teams, you're always going to have potential "made-for-TV" 2nd round games in that spot.
  26. Chris N's avatar
    Chris N
    | Permalink
    Looking at this analysis, it would have been easy to flip South #8 UNC with South #7 San Diego State. That would have put us in the same regional, but seeded appropriately relative to SDSU, and would have us in the Philly pod, with a first round game against #10 Oklahoma and a likely 2nd round game versus Georgetown. At the same time, it’s not clear that we’re obviously better than the other #7 seeds, and it’s tough to make a case for us as a #6 (our recent numbers match up well with Butler, but we lost to them badly in Maui).

    Tellingly, the committee has said that they didn’t consider “last 10 games” as a seeding factor this year, which I think is a huge mistake. Frankly, I don’t think the committee watched a minute of basketball last week, because if they had, they might have given us more credit.

Leave a Comment