Page 1 of 2 12 LastLast
Results 1 to 20 of 29

Hybrid View

Previous Post Previous Post   Next Post Next Post
  1. #1

    #1 vs. #2 seeds -- does it matter?

    This discussion has been going on in the Kyrie thread, and that's not really the right place for it, so I thought I'd start a new thread about it here.

    Since the tournament expanded to 64 teams in 1985, there have been 26 tournaments. In that time, there have obviously been 104 #1 seeds and 104 #2 seeds. Of those, 45 #1 seeds have made the Final Four (43%), vs. 23 #2 seeds (22%), almost twice as many #1s as #2s. There have been 16 #1 seeds who became champion in those 25 years, against 4 #2 seeds (excactly four times more #1s than #2s).

    Some people have stated the opinion that the reason the #1s have succeeded better than the #2s is because the #1s are better. I'm sure that's part of it. But I believe a much bigger part of it is the road of the #2 is much harder.

    Can we prove it, either way? Not definitively, but I have dug a little deeper and here's my data:

    Historically, #1 seeds are 104-0 against #16s. #2 seeds are 100-4 against #15s. Is the difference because #1s are better? Possibly, but probably not. There's such a huge difference in skill between the #1/#2 and the #15/#16 that it's more likely that the the #15s are better than the #16s. Either that, or random chance, which might be the most probable conclusion.

    #1 seeds have beaten #9 seeds to the tune of 58-6 and #8 seeds by 53-13, a combined total of 111-19. #2 seeds have beaten #10 seeds by 28-20 and #7 seeds by 56-21, for a combined total of 84-41. Here, I'm sure some of the difference is due to #1s being better than #2s, but my guess is it's much more likely that 7/10s are a lot better than 8/9s, either because they are more likely to be underseeded (the 10s) or because they're just plain better (the 7s).

    Either way, apparently only 84 #2 seeds have reached the Sweet 16 in 26 years, while 111 #1 seeds have done so. That difference of 27 is actually more than the overall difference of #1s over #2s in making the Final Four (22). And if that's really where the difference is, I would argue that the road is more of an explanation than the inherent talent differential between #1 and #2.

    To go a little further, in the next round, #1 seeds are 39-17 against #4s and 38-7 against #5s, while #2 seeds are 34-21 against #3s and 22-10 against #6s. That's 77-24 (76.2%) against 56-31 (64.4%), a fairly large differential, and in my mind more than you'd expect for the best four teams vs. the next best four teams, again supporting the idea that the competition is stiffer for the #2s. The #1 seeds also got to play twice as many double-digit seeds as #2s have (because #4 & #5 are more likely to lose early than #3 or #6). This piece of luck has absolutely nothing to do with how good the higher seed is, and in those bonus games, #1 seeds have gone 20-0 while #2 seeds have gone 9-1, giving 10 (really 11, I suppose) additional #1s a chance to reach the Final Four.

    In the last three rounds, #1s beat #2s, 34-28. Here we finally see some evidence that #1s are better, but it's a reasonable 55%, rather than double or quadruple.

    I found the full breakdown here.

    My conclusion is the road matters. Being a #1 seed gives you a big advantage in reaching the Final Four over being a #2 seed.
    Last edited by Kedsy; 03-12-2011 at 01:49 PM.

  2. #2
    Quote Originally Posted by Kedsy View Post
    This discussion has been going on in the Kyrie thread, and that's not really the right place for it, so I thought I'd start a new thread about it here.

    Since the tournament expanded to 64 teams in 1985, there have been 26 tournaments. In that time, there have obviously been 104 #1 seeds and 104 #2 seeds. Of those, 45 #1 seeds have made the Final Four (43%), vs. 23 #2 seeds (22%), almost twice as many #1s as #2s. There have been 16 #1 seeds who became champion in those 25 years, against 4 #2 seeds (excactly four times more #1s than #2s).

    Some people have stated the opinion that the reason the #1s have succeeded better than the #2s is because the #1s are better. I'm sure that's part of it. But I believe a much bigger part of it is the road of the #2 is much harder.

    Can we prove it, either way? Not definitively, but I have dug a little deeper and here's my data:

    Historically, #1 seeds are 104-0 against #16s. #2 seeds are 100-4 against #15s. Is the difference because #1s are better? Possibly, but probably not. There's such a huge difference in skill between the #1/#2 and the #15/#16 that it's more likely that the the #15s are better than the #16s. Either that, or random chance, which might be the most probable conclusion.

    #1 seeds have beaten #9 seeds to the tune of 58-6 and #8 seeds by 53-13, a combined total of 111-19. #2 seeds have beaten #10 seeds by 28-20 and #7 seeds by 56-21, for a combined total of 84-41. Here, I'm sure some of the difference is due to #1s being better than #2s, but my guess is it's much more likely that 7/10s are a lot better than 8/9s, either because they are more likely to be underseeded (the 10s) or because they're just plain better (the 7s).

    Either way, apparently only 84 #2 seeds have reached the Sweet 16 in 26 years, while 111 #1 seeds have done so. That difference of 27 is actually more than the overall difference of #1s over #2s in making the Final Four (22). And if that's really where the difference is, I would argue that the road is more of an explanation than the inherent talent differential between #1 and #2.

    To go a little further, in the next round, #1 seeds are 39-17 against #4s and 38-7 against #5s, while #2 seeds are 34-21 against #3s and 22-10 against #6s. That's 77-24 (76.2%) against 56-31 (64.4%), a fairly large differential, and in my mind more than you'd expect for the best four teams vs. the next best four teams, again supporting the idea that the competition is stiffer for the #2s. The #1 seeds also got to play twice as many double-digit seeds as #2s have (because #4 & #5 are more likely to lose early than #3 or #6). This piece of luck has absolutely nothing to do with how good the higher seed is, and in those bonus games, #1 seeds have gone 20-0 while #2 seeds have gone 9-1, giving 10 (really 11, I suppose) additional #1s a chance to reach the Final Four.

    In the last three rounds, #1s beat #2s, 34-28. Here we finally see some evidence that #1s are better, but it's a reasonable 55%, rather than double or quadruple.

    I found the full breakdown here.

    My conclusion is the road matters. Being a #1 seed gives you a big advantage in reaching the Final Four over being a #2 seed.
    It certainly did- I was curious how this plays out over the past 10 years. Duke lost its first player early to the NBA in 1999 and . I just wonder if there is more parity now and if that would change the numbers.

    Here is the last decade

    2000, 8, 5, 1, 8
    2001 1, 3, 1, 2
    2002 1, 1, 5, 2
    2003 3, 1, 3, 2
    2004 2, 3, 1, 2
    2005 1, 4, 1, 5
    2006 4, 2, 3, 11
    2007 1, 2, 2, 1
    2008 1,1,1,1
    2009 2, 1,1, 3
    2010 5, 5, 2, 1

    18/44 1 seed 41%
    10/44 2 seed 22%
    6/44 3
    2/44 4
    5/44 5
    2/44 8
    1/44 11

    So it seems that there is still roughly the same advantage to being a 1 seed over a 2 even it the era where teams are not keeping the bulk of their players for 3 or more years. Curiously being a 5 seed is about as good as being a 3.

  3. #3
    Quote Originally Posted by Kedsy View Post
    I found the full breakdown here.

    My conclusion is the road matters. Being a #1 seed gives you a big advantage in reaching the Final Four over being a #2 seed.
    I wouldn't mind running a quick study on this. Does anyone know where I can get pre-tournament Pomeroy or Sagarin numbers from?

  4. #4
    Quote Originally Posted by ns7 View Post
    I wouldn't mind running a quick study on this. Does anyone know where I can get pre-tournament Pomeroy or Sagarin numbers from?
    I have found pre-tournament Sagarin and RPI going back to 1996. I will attempt to put them into a consistent format and post them in this thread, but I might not finish that task today. I'll do my best to get it done in the next couple days.

  5. #5
    Join Date
    Nov 2007
    Location
    Delaware
    I don't think that there is even a question. For those that follow Pete Tiernan's Bracketscience.com, you know that 1 seeds on average win 1 more game per tournament than 2 seeds (~3.5 to ~2.5). Are 1 seeds better than 2 seeds, yes, but not by a full game.

    I think the biggest difference in the road for those teams is the first game. Statistics show that by any measure (RPI, Kenpom, etc.), the biggest difference between any two seeds is 15 and 16 seeds. They are generally both going to be low major teams, but the 16 seeds are often won low major conference tournaments in an upset, and the 15 seeds were the ones expected to win. Does it make too much of a difference in results, no, as 15 seeds are 4-100 as opposed to 0-104, but there are generally at least 1 or 2 2-seeds every year that have to work for 35-40 minutes to win that game while 1 seeds can normally turn it off at halftime at the latest. Rarely does a 1-seed have to work past the ten minute mark of the second half.

    This is a big reason why second round upsets are much more common for two seeds than 1 seeds. There isn't much difference between 7, 8, 9, and 10 seeds. In fact, 10 seeds perform much better against 2 seeds than 7's do, so the quality of the second round opponent isn't much of a factor. 1-seeds are 91-13 in the second round, while 2 seeds are just 67-33. That's an 87.5 win percentage compared to just a 67.0 win percentage. The gap actually closes once you get to the Sweet 16 with 2 seeds going 48-19 (71.9%) vs. 75-16 for 1-seeds (82.4%). The 1 seeds are still better, but not by nearly as much.

    If you're a 2 seed, and you can cruise through the sub-regional with TWO easy wins, you probably don't have any disadvantage in the regional round compared to a 1-seed, but if you skate by in either of the early games (which is much more common for 2's than 1's), you're in trouble.

    It's also interesting (and not very fun to note) that a full 25% of 1 seed losses in the sweet 16 (4/16) have been by Duke.

    EDIT: just saw the question about Pomeroy and Sagarin. Unfortunately, those two don't archive their pre-tourney numbers, so unless a third party does it, they don't exist.
    Pratt '09
    GO DUKE!

  6. #6
    Join Date
    Oct 2009
    Location
    Durham
    Quote Originally Posted by SCMatt33 View Post
    I don't think that there is even a question. For those that follow Pete Tiernan's Bracketscience.com, you know that 1 seeds on average win 1 more game per tournament than 2 seeds (~3.5 to ~2.5). Are 1 seeds better than 2 seeds, yes, but not by a full game.
    Actually this makes perfect sense...the difference comes from 1 seeds beating 2 seeds...

    I think to do this the most objectively, you have to look at whether the 1 or 2 seeds make it to the elite 8, at that point they play each other, and difficulty of road is no longer applicable and strength of teams is....so can we get a % of 1 and 2 seeds in the elite 8?
    1200. DDMF.

  7. #7
    Join Date
    Feb 2007
    Location
    Steamboat Springs, CO

    Talking The Tale is in the Tail

    There are 300+ Div I teams. The top four teams represent the extreme tail of the distribution (density fcn., for the picky stat guys). There is the potential for the top one or two teams to be so much better that the difference between these teams and #2 seeds can be greater than the difference between #2 seeds and #4 seeds.

    For me that's the main difference between #1 and #2. Admittedly #2 has the potential to play tougher teams all the way through, but -- especially in seeds #6 through #11 -- I think that this is just a mass of teams of about equal quality. Of course, the fact that #9 seeds have a better record head-to-head than the #8 seeds supports this position.

    sagegrouse

  8. #8
    Join Date
    Nov 2007
    Location
    Delaware
    Quote Originally Posted by uh_no View Post
    Actually this makes perfect sense...the difference comes from 1 seeds beating 2 seeds...

    I think to do this the most objectively, you have to look at whether the 1 or 2 seeds make it to the elite 8, at that point they play each other, and difficulty of road is no longer applicable and strength of teams is....so can we get a % of 1 and 2 seeds in the elite 8?
    I should have put this in my first post, but CBS just published Pete Tiernan's yearly "Guide to Every Seed Matchup" The 1's and 2's are actually 18-18 against each other, further supporting the road to the Elite 8 as a factor, though 1-seeds do hold a +1.2ppg scoring edge.
    Pratt '09
    GO DUKE!

  9. #9
    Quote Originally Posted by SCMatt33 View Post
    I should have put this in my first post, but CBS just published Pete Tiernan's yearly "Guide to Every Seed Matchup" The 1's and 2's are actually 18-18 against each other, further supporting the road to the Elite 8 as a factor, though 1-seeds do hold a +1.2ppg scoring edge.
    Thanks for this link. I apologize for my data (in the initial post) being wrong -- I should have checked my link (which contains faulty data) better before I quoted from it.

    The basics are unchanged, however. #1 seeds get out of the 1st round at a clip of 104-0, out of the second round by 91-13, and out of the third round at 75-16. #2 seeds survive the first round by 100-4, the second round by 67-33, and the third round at 48-19.

    As you say, apparently in Elite Eight matchups, the #1s and #2s split 18 and 18. But 24 more #1s get out of the second round alive, going up to 27 more #1s surviving past the Sweet 16 round. This goes down to 22 more #1s reaching the Final Four.

    So, basically the reason #2 doesn't reach the Final Four as much as #1 is because the #2 seeds lose a lot more to the 7/10s than #1s lose to the 8/9s (20 times more, to be precise). Can this be explained by the four best teams being better than the next four best teams? I doubt it. Can it be explained by the idea that the top 1 or 2 teams are so good they never lose and thus skew the data? Not only do I doubt that would explain it, I don't think the premise is true (witness Kansas last year). So, although my initial data was a little off (and once again I apologize for that), I stick by my conclusion that the #2 has a much tougher road.

    P.S.: As far as having pre-tournament numbers, I think I have pre-tourney Sagarin back to 2002. I only have 2009 and 2010 of pre-tournament Pomeroy.

  10. #10
    Join Date
    Nov 2007
    Location
    Delaware
    Quote Originally Posted by Kedsy View Post
    P.S.: As far as having pre-tournament numbers, I think I have pre-tourney Sagarin back to 2002. I only have 2009 and 2010 of pre-tournament Pomeroy.
    Do you have the exact date of your computer numbers, because one of the problems with computer numbers is that they change daily. Last year, Duke and Kansas traded places once or twice (I can't remember exactly) between Selection Sunday and tip-off on Thursday due to NIT results. Of coarse, teams at the top aren't normally that close, but using pythag numbers to make your picks could result in different picks if you turn in your bracket Monday morning or Thursday morning. Of coarse, this point will be moot in the future if bracket pools require First Four selections, as Monday's Pomeroy and Sagarin numbers will be the only relevant ones.
    Pratt '09
    GO DUKE!

  11. #11
    This argument just has so many variables. One day, some statistician will go through it all with a supercomputer and analyze the whole situation. But the whole 15/16 seed is a valid argument and probably explains why there are more 7/10 upsets than 8/9. But I call b/s that somehow the 10 seeds are better than the 8/9's b/c the 10 seeds are underseeded. That just doesn't make sense and seems to just be a simplification to make this argument work.
    I would say the path has something to do with it but I would also say the fact that the #3-#6 teams on the S-curve are probably even but the difference between 1+2 vs. 7+8 probably makes the greater difference. Just look at it this year. The difference between say Duke, UNC, Pitt, and ND is arguable at best. But then you start to compare BYU or SD st. against OSU and KU and that explains the reason why the 1 seeds have a distinct advantage.
    I also think that the 2 seeds tradtionally are either top teams going through slumps (Villanova) or you get some overrated teams who got hot at the right time, perhaps OSU comes to mind last year. However, this year you have 5-7 teams depending upon whether Pitt and Texas are going to get their act together who can win a national title. So that means you will have 1-3 2-seeds who by all indications are sitting dead in the water when it comes to the national scene.
    So I'd say its a combo of the 2 seeds not being as good and the 2 seed having to play a real first game. But the 7-10 seeds, I don't see a huge difference between. But the idea that 1 seeds are more likely to see a weak opponent in the S16 could explain some things as well since obviously the 4-5's are more likely to go down than say the 3.
    Last edited by sporthenry; 03-12-2011 at 08:18 PM.

  12. #12
    Join Date
    Feb 2008
    Location
    20 Minutes From The Heaven That Is Cameron Indoor
    Great work Kedsy. This one is a no brainer to me. It is better to be a 1 Seed. The Sweet 16 game is a much more favorable matchup in most cases for the 1 seed than the 2 seed. The gap between the 1 seed and the 4 seed is normally significantly different than the gap between the 2 and the 3 seed.

    Often times as a 2 seed in the Sweet 16 game you a facing a team that is either just as good as you or even better.

    Tomorrow's game is huge for Duke. Win, and we likely get the 1 seed. Lose and we end up on the 2 line with a tougher road to hoe. We need to get it done tomorrow.

  13. #13
    Quote Originally Posted by Newton_14 View Post
    Great work Kedsy. This one is a no brainer to me. It is better to be a 1 Seed. The Sweet 16 game is a much more favorable matchup in most cases for the 1 seed than the 2 seed. The gap between the 1 seed and the 4 seed is normally significantly different than the gap between the 2 and the 3 seed.

    Often times as a 2 seed in the Sweet 16 game you a facing a team that is either just as good as you or even better.

    Tomorrow's game is huge for Duke. Win, and we likely get the 1 seed. Lose and we end up on the 2 line with a tougher road to hoe. We need to get it done tomorrow.
    Well 1's win 70% of the time vs #4's while 2's win 62% of the time vs #3's. 5 more wins/4 less losses. So the huge gap doesn't come from those matchups. If anything, you would say that those 2 stats aren't nearly as favorable to the 1 seeds as they are playing 13-16.

  14. #14
    The only way to do this analysis would be to use the S-curve and find out where these teams end up. Last year, I'd say the S-cuve was KU, UK, Duke, Cuse, WVU, Oh. State, K-state, and Nova.
    The worst #2 ended up losing in the 2nd round and could have very well lost in the first round. Of course, the 1st overall team got upset. But the #2 team ended up losing to the #5 team, so if Duke is #5, I'm not that worried about it.

  15. #15
    Quote Originally Posted by Newton_14 View Post
    Great work Kedsy. This one is a no brainer to me. It is better to be a 1 Seed. The Sweet 16 game is a much more favorable matchup in most cases for the 1 seed than the 2 seed. The gap between the 1 seed and the 4 seed is normally significantly different than the gap between the 2 and the 3 seed.

    Often times as a 2 seed in the Sweet 16 game you a facing a team that is either just as good as you or even better.

    Tomorrow's game is huge for Duke. Win, and we likely get the 1 seed. Lose and we end up on the 2 line with a tougher road to hoe. We need to get it done tomorrow.
    In the K era, Duke has been to the FF 11 times with the following seeds

    86 1 seed
    88 2 seed
    89 2 seed
    90 3 seed
    91 2 seed
    92 1 seed
    94 2 seed
    99 1 seed
    01 1 seed
    04 1 seed
    10 1 seed

    5 of the 11 they have been a 2 or 3. 6 of the 11 they have been a 1. So for Duke, they have gotten to the FF as a 2 or lower seed almost as often as they have as a 1 seed.

  16. #16
    Join Date
    Feb 2008
    Location
    20 Minutes From The Heaven That Is Cameron Indoor
    Quote Originally Posted by dukelifer View Post
    In the K era, Duke has been to the FF 11 times with the following seeds

    86 1 seed
    88 2 seed
    89 2 seed
    90 3 seed
    91 2 seed
    92 1 seed
    94 2 seed
    99 1 seed
    01 1 seed
    04 1 seed
    10 1 seed

    5 of the 11 they have been a 2 or 3. 6 of the 11 they have been a 1. So for Duke, they have gotten to the FF as a 2 or lower seed almost as often as they have as a 1 seed.
    Fair point, but it does not change the fact that the path is easier for the 1 seed. This Duke team can make the FF either way, but I would much prefer to be a 1.

  17. #17
    Quote Originally Posted by Newton_14 View Post
    Fair point, but it does not change the fact that the path is easier for the 1 seed. This Duke team can make the FF either way, but I would much prefer to be a 1.
    No question- Duke has won the NC 4 times and 3 of 4 as the 1 seed. So for all the marbles- it seems to matter.

  18. #18
    I agree that it's more interesting to compare the 3-4-5-6 seeds as opposed to the 1-2-7-8 seeds. Rather than s-curve it, some sort of different comparison would worl out. For instance, one-seeds that were 3 in the AP poll at selection time or worse versus two-seeds that were 6th or better.

  19. #19
    Join Date
    Nov 2007
    Location
    Delaware
    Quote Originally Posted by dukelifer View Post
    5 of the 11 they have been a 2 or 3. 6 of the 11 they have been a 1. So for Duke, they have gotten to the FF as a 2 or lower seed almost as often as they have as a 1 seed.
    Here's some notable stats on this. In the K era, Duke is 6 for 11 making the FF as a 1 seed, 4 for 7 as a 2 (4 for 8 overall if you count pre-K), and 1 for 5 as a 3. So making the Final Four has been about the same for Duke as a 1 and 2 seed (54.5% and 57.1% respectively). As was mentioned earlier, though, the tough road that the 2 seeds faced seemed to hurt once they got to the Final Four (though you can't really tell with these numbers). In the Final Four as a 1-seed, Duke is 5-1 in semi-final games and 3-2 in title games. As a 2-seed, Duke 2-2 in the semis and 1-1 in the finals. Duke lost to UNLV in the finals in its only 3-seed appearance.

    As much as you'd like to exonerate or blame the seeding for all this, it is really too small of a sample size to use Duke's results to analyze 1 vs. 2 seeds. You have to go on the larger sets of data.
    Pratt '09
    GO DUKE!

  20. #20
    Quote Originally Posted by SCMatt33 View Post
    Do you have the exact date of your computer numbers, because one of the problems with computer numbers is that they change daily. Last year, Duke and Kansas traded places once or twice (I can't remember exactly) between Selection Sunday and tip-off on Thursday due to NIT results. Of coarse, teams at the top aren't normally that close, but using pythag numbers to make your picks could result in different picks if you turn in your bracket Monday morning or Thursday morning. Of coarse, this point will be moot in the future if bracket pools require First Four selections, as Monday's Pomeroy and Sagarin numbers will be the only relevant ones.
    All my numbers are either from Selection Sunday night or during the day on Monday. They are unsullied by play-in or NIT results.

Similar Threads

  1. Seth Davis re #1 seeds
    By TNDukeFan in forum Elizabeth King Forum
    Replies: 30
    Last Post: 02-28-2011, 08:13 AM
  2. 2010 Tournament prediction seeds
    By rotogod00 in forum Elizabeth King Forum
    Replies: 16
    Last Post: 04-15-2009, 12:39 PM
  3. Lunardi 08-09 Bracketology #1 seeds: UNC & Duke
    By MrBisonDevil in forum Elizabeth King Forum
    Replies: 20
    Last Post: 05-03-2008, 10:57 AM
  4. Who are the #1 seeds now?!?!
    By JasonEvans in forum Elizabeth King Forum
    Replies: 12
    Last Post: 02-28-2007, 09:33 PM
  5. ACC Tourney seeds.
    By geeveebee in forum Elizabeth King Forum
    Replies: 9
    Last Post: 02-28-2007, 10:42 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
  •