Page 1 of 1

Bracketing question

Posted: Fri Oct 27, 2017 9:21 am
by Joshua Rutsky
So I find myself in the unusual situation of having a potential field of 42 teams for the HIT this year. I'm all in favor of rebrackets and round robins, but I think that a sizable portion of my tourney field might balk at a 12 rounds for everyone tourney. At 36 teams, we were perfect--six pools of six before lunch, then a rebracket to the same numbers in the afternoon. I'd love to be able to stay with this format of 6 team pools, but with seven pools of six, one team that wins a pool will not get the chance to play for the title, and I hate that.

What I'm thinking is that I take the top five teams automatically by taking each pool's winners and comparing either PPG or PPB average (which would be better?) We then have a single "play-in" match between teams 6 and 7 to determine which team goes to the top flight and which team goes into the second tier. Does this seem like a reasonable approach? Which is preferable as a determining factor in your minds, PPG or PPB average? Alternatively, is there a better format I could use for a 42 team field to keep us (largely) at 10 games max?

Thanks for the input.

Re: Bracketing question

Posted: Fri Oct 27, 2017 9:28 am
by dtaylor4
7 pools of 6 prelims
7 pool winners plus top 5 runners-up based on record, then PPB. Snake-seed by record/PPB into 2 pools of 6, RR within each pool, then winners play off for the title, runners-up play for 3rd/4th
For the other 30 teams, seed by record/PPB, and do 5 consolation brackets of 6.

Re: Bracketing question

Posted: Fri Oct 27, 2017 10:51 am
by Irreligion in Bangladesh
I agree with Donald on championship tier = 12 teams, with parallel pools up top to feed into a final match.

To determine all your prelim-into-playoff seeds -- both the "which 5 pool runners up get the wild cards?" and the "how do I seed the 12 teams into the two parallel pools," [EDIT: as well as the "seeding your consolation pools," I'm barely awake] PPB has an advantage over PPG because it's opponent independent. (Thus, if one pool is surprisingly weaker than the other pools, the pool winner there isn't accidentally overseeded because they got to play weaker opponents all morning.) PPG has an advantage in that it does incorporate more data - PPB ignores all tossup performance, of course - but for inter-pool comparisons like these, PPB is considered to be more fair.

If you've got a tie in a pool for 1st and/or 2nd, PPG is the best statistical tiebreaker there because they all played the same schedule. We don't need the opponent independence that PPB provides, so we happily pick up the extra data that PPG provides.

Re: Bracketing question

Posted: Fri Oct 27, 2017 10:54 am
by dtaylor4
Irreligion in Bangladesh wrote:I agree with Donald on championship tier = 12 teams, with parallel pools up top to feed into a final match.

To determine all your prelim-into-playoff seeds -- both the "which 5 pool runners up get the wild cards?" and the "how do I seed the 12 teams into the two parallel pools," PPB has an advantage over PPG because it's opponent independent. (Thus, if one pool is surprisingly weaker than the other pools, the pool winner there isn't accidentally overseeded because they got to play weaker opponents all morning.) PPG has an advantage in that it does incorporate more data - PPB ignores all tossup performance, of course - but for inter-pool comparisons like these, PPB is considered to be more fair.

If you've got a tie in a pool for 1st and/or 2nd, PPG is the best statistical tiebreaker there because they all played the same schedule. We don't need the opponent independence that PPB provides, so we happily pick up the extra data that PPG provides.
My references to using PPB were meant in using them for determining the wild card spots, and for seeding teams 13-42 for the playoffs.

Re: Bracketing question

Posted: Fri Oct 27, 2017 11:00 am
by Irreligion in Bangladesh
dtaylor4 wrote:
Irreligion in Bangladesh wrote:I agree with Donald on championship tier = 12 teams, with parallel pools up top to feed into a final match.

To determine all your prelim-into-playoff seeds -- both the "which 5 pool runners up get the wild cards?" and the "how do I seed the 12 teams into the two parallel pools," PPB has an advantage over PPG because it's opponent independent. (Thus, if one pool is surprisingly weaker than the other pools, the pool winner there isn't accidentally overseeded because they got to play weaker opponents all morning.) PPG has an advantage in that it does incorporate more data - PPB ignores all tossup performance, of course - but for inter-pool comparisons like these, PPB is considered to be more fair.

If you've got a tie in a pool for 1st and/or 2nd, PPG is the best statistical tiebreaker there because they all played the same schedule. We don't need the opponent independence that PPB provides, so we happily pick up the extra data that PPG provides.
My references to using PPB were meant in using them for determining the wild card spots, and for seeding teams 13-42 for the playoffs.
Right indeed - I forgot to mention that the same theory applies to the consolation seeds as well. (Without that, it looks more like I'm trying to correct you than agreeing with you and explaining the theory behind making the PPG vs PPB choice.) Thanks!

Re: Bracketing question

Posted: Fri Oct 27, 2017 12:26 pm
by 1992 in spaceflight
Similar question: What if you want 16 teams to be eligible for the championship bracket, instead of 12?

Re: Bracketing question

Posted: Fri Oct 27, 2017 12:45 pm
by dwd500
I came up with something for that, once. I like making sure that the top 2 from each bracket proceed out of prelims, just in case I seeded it all wrong.

Prelims: 7 groups of 6.

Playoffs: Top 2 of each bracket, plus 2 wild cards (however you determine those) - snake seeded into 4 groups of 4 for rounds 6 through 9.
Placement for teams #17-24, same as playoffs, rounds 6-9.
Consolation for teams #25-42 - 3 tiers of 6 teams, play a round robin to finish after round 10.

The problem was that Consolation teams were getting more games than the #17-24 teams. I fixed it with carrying over a match and re-bracketing the two placement pools (top 2 in one, bottom two in one) giving them 2 more games for rounds 10-11.

Superplayoffs was the playoff winners in a round robin (rounds 10-12), ditto for the #2's, #3's, and #4's to finish their day.

Left 2 games for the final.

I have it written out (more-or-less) here.

Re: Bracketing question

Posted: Fri Oct 27, 2017 1:27 pm
by Monstruos de Bolsillo
Joshua Rutsky wrote: I'd love to be able to stay with this format of 6 team pools, but with seven pools of six, one team that wins a pool will not get the chance to play for the title, and I hate that.

What I'm thinking is that I take the top five teams automatically by taking each pool's winners and comparing either PPG or PPB average (which would be better?) We then have a single "play-in" match between teams 6 and 7 to determine which team goes to the top flight and which team goes into the second tier. Does this seem like a reasonable approach? Which is preferable as a determining factor in your minds, PPG or PPB average? Alternatively, is there a better format I could use for a 42 team field to keep us (largely) at 10 games max?

Thanks for the input.
I've always felt it is better to take too many teams into the playoffs, rather than too few. With a tournament of that size, most of the 2nd place teams in the morning should absolutely belong in the playoff brackets. The seeding above for the wild cards for PPB across the brackets works well, and having the two equal playoff brackets means you only need one more game the rebracketed round robin to playoff each place.

For Henderson's HS tournament last year, we had 30 teams, with the usual 5 brackets of 6. I felt that taking 6 teams was too few, and it really can screw over the 4-1s, so we took 12 into playoffs. Top 2 from each bracket, plus the two next best 3rd place teams. Taking just the top 2 would have also worked, but then the numbers aren't as nice and the scheduling is unbalanced (or you have to be more creative.) 12 was probably a bit much, but I feel this was more fair, even if it does make the playoff brackets a shade less competitive.