HSNCT seeding discussion

[Edward Powers]

William,

Any team that can defeat Alpharetta, Centennial and Walton in one tournament deserves to climb in the rankings, minimally into the Top 25. Of course Chattahoochie's play at nationals in a couple of weeks could make all such talk moot, but for seeding purposes at nationals it certainly deserves Top 25 consideration, don't you think?

[AlphaQuizBowler]

William,

Any team that can defeat Alpharetta, Centennial and Walton in one tournament deserves to climb in the rankings, minimally into the Top 25. Of course Chattahoochie's play at nationals in a couple of weeks could make all such talk moot, but for seeding purposes at nationals it certainly deserves Top 25 consideration, don't you think?

I think so, but Chattahoochee's overall play, in my opinion, is at a higher level than what their PPB indicates (and they have at previous tournaments as well as this one beaten teams with higher PPBs). If seeding is based entirely on PPB, then they're unfortunately destined to be underseeded.

[bt_green_warbler]

NAQT does not normally release seeding data to the public, but I can confirm that:

1) HSNCT seeding incorporates measures of both tossup and bonus performance.

2) Yes, Chattahoochee A is among the current top 25 seeds.

[RountreeCHS]

NAQT does not normally release seeding data to the public, but I can confirm that:

1) HSNCT seeding incorporates measures of both tossup and bonus performance.

2) Yes, Chattahoochee A is among the current top 25 seeds.

This is nice to hear.

Jeff, out of curiosity, has NAQT ever considered releasing its top 25 HSNCT seeds to the public? Of course, I realize the implications in doing so, but I was just wondering...

[bt_green_warbler]

We have not; in short, we think the potential cost in hurt feelings/public argument will outweigh the potential small benefit in improved rankings. (Note for this purpose that the precise rankings involved in seeding HSNCT are much less significant than those for a tournament with defined playoff brackets; HSNCT seeds are only intended to provide a rough balance for the first four rounds of play on Saturday.)

[Edward Powers]

Jeff,

Without naming any schools, could you give an example of how this works? For example, for this year's field of 224 teams will you create 14 brackets of 16 teams, then roughly seed in an "S" curve and have the sixteen teams constantly divide into winners & losers brackets for 4 rounds, etc? So, for example, might the # 1 seed be in a bracket with the 28th, 29th, 56th, 57th, 84th, 85th, 112th, 113th, 140th, 141st, 168th, 169th, 196th, 197th and the 224th seeded teams? And if seeds hold, is the theory that after 4 rounds there will be one undefeated team left in each of the 14 brackets, the Top 14 seeded teams, at least in theory? Or, without getting too complicated, is it done somewhat differenty?

[bt_green_warbler]

Jeff,

Without naming any schools, could you give an example of how this works? For example, for this year's field of 224 teams will you create 14 brackets of 16 teams, then roughly seed in an "S" curve and have the sixteen teams constantly divide into winners & losers brackets for 4 rounds, etc? So, for example, might the # 1 seed be in a bracket with the 28th, 29th, 56th, 57th, 84th, 85th, 112th, 113th, 140th, 141st, 168th, 169th, 196th, 197th and the 224th seeded teams? And if seeds hold, is the theory that after 4 rounds there will be one undefeated team left in each of the 14 brackets, the Top 14 seeded teams, at least in theory?

This is exactly correct.

[etchdulac]

Hey Jeff. Within the seeding process, is any effort made to keep apart teams from the same state/region? Or is that considered a low priority by NAQT?

[bt_green_warbler]

We do try and separate teams from the same state/region (as well as teams from the same school). Usually this takes the form of switching two adjacent teams on the seeding list, or adjusting the order in a set of three teams with similar stats.

The exact details differ a bit from year to year, depending on the field. For instance, this year teams will probably play teams from their own states somewhat sooner than they did last year. (Math as follows: there are 14 brackets of 16 teams each; but with the 224-team field, some states, like Texas and Ohio, have 15 or more teams in the field, setting up the possibility of a round 4 same-state game.)

[Eccles cake]

I'm a bit confused, are we not doing Swiss pairs anymore?

[bt_green_warbler]

We're using the same system as last year; in general, teams will play teams with similar records. This thread is about: how do we set up the initial rounds of play to provide balanced input into the system? (That is: a purely random draw would produce suboptimal results. Such as "once every two centuries the #1 and #2 seeds play each other in round one" and "there's a 2% chance of one LASA team playing another in the first round." We want to avoid that sort of thing.)

[jonpin]

I'm a bit confused, are we not doing Swiss pairs anymore?

By "brackets", he means units which will be playing within themselves for the first few rounds of the swiss pairing. So in round two, the winners of two specific games will play and the losers of those games will play. In round three, there will be four games consisting of (a team from 4-team block A) vs (a team from 4-team block B). In round four, there will be eight games consisting of (a team from 8-team block X) vs (a team from 8-team block Y). After that, there may be 32-team brackets from the merging of two 16-team blocks, but that might be made difficult by the existence of byes.

[Eccles cake]

Oh, I see. Thanks guys. This makes sense now.

[ether a go-go]

It does go to seven blocks with 32 teams each for each team's fifth match. It then goes to one block with 96 teams and two blocks with 64 teams each for each team's sixth match.

[Lion81]

So are teams ranked all the way from 1 to 224? Or is it just top 25? I'm not gonna lie, the way you guys made the system to work is pretty sweet. Especially last year when you somehow coordinated the switching between buildings and byes.

[bt_green_warbler]

Yes, we produce a complete list from 1 to 224. (actually, 1 to whatever the last standby team is)

[Joe Romersa]

In each 16 team bracket, in the first round, does the highest seeded team play against the lowest seeded team or the second highest seeded team?

[Edward Powers]

Jeff,

Is each team's original card number entering Round 1 identical to its original seeding in the overall field? For example, would the overall number one seed start with card 1 and play the team seeded at 224 in the opening round, seed 2 playing seed 223, seed 3 playing seed 222, all the way up to seed 112 playing seed 113 if it is a field of 224 teams?

[Madagascar Serpent Eagle]

I think the actual card numbers are randomized; I know I started with card 5 and there was no way anyone in their right mind would have seeded me anywhere close to #5.

[Andrew's a Freshman]

I think the actual card numbers are randomized; I know I started with card 5 and there was no way anyone in their right mind would have seeded me anywhere close to #5.

We had #1 so I'm pretty sure they were in seeding order.

[Matthew D]

I don't think it was a seeding order because we got card to start the day...

[jonah]

The card numbers are indeed scrambled such that the seedings cannot be reverse-engineered.

[Remember-the-Alamo-Remember-Goliad]

For Cistercian Prep, the obvious 'advantage' to the system used at NAQT/nationals is that we didn't have to play Texas teams. After all, we've been playing LASA A, Seen Lakes A and St. Mark's A all year long. We were tired of this! So, having the opportunity NOT to play our state neighbors was a real PLUS as far as I'm concerned. In fact, when the complete stats are posted, I will need to check and see if we played ANY Texas state teams on Saturday.

[ether a go-go]

If you didn't play any Texas teams, there was some luck involved. Both HSNCT and NSC prevent you from playing a team from your state on Saturday morning, but after that the pairings depend more and more on performance at the tournament rather than initial placement, and there are some matches between teams from the same state.

[etchdulac]

In fact, when the complete stats are posted, I will need to check and see if we played ANY Texas state teams on Saturday.

Cistercian A played LASA B Saturday when both teams were 6-1, if memory serves.

[BlueDevil95]

I remember that Norcross played Woodward Academy, both teams from Georgia.

[dtaylor4]

If you didn't play any Texas teams, there was some luck involved. Both HSNCT and NSC prevent you from playing a team from your state on Saturday morning, but after that the pairings depend more and more on performance at the tournament rather than initial placement, and there are some matches between teams from the same state.

Depending on the set-up, this is impossible for the NSC. If that were the case, we'd need 15 brackets this year to accommodate the IL teams.

[bt_green_warbler]

I believe this year's HSNCT only prohibited same-state matches in the first three rounds of the tournament.

[Down and out in Quintana Roo]

I believe this year's HSNCT only prohibited same-state matches in the first three rounds of the tournament.

That's still really impressive though, to make that sort of commitment to teams playing other teams from different states. I'll bet that takes a lot of work for NAQT to set that up.

[Kenneth Widmerpool]

I believe this year's HSNCT only prohibited same-state matches in the first three rounds of the tournament.

Not to question the work that NAQT puts in, but, is this really necessary? With the sheer geographical diversity of schools in the tournament, every team is already facing opponents from out of their state for several, if not most, of their matches

[ether a go-go]

It's not really necessary, but it probably is for the better. For one thing, it's pretty easy to do. It involves a small number of teams being seeded one or two places higher or lower, and in the end it probably makes the results more accurate since it is easy to overrate or underrate schools from the same state uniformly because they play each other and common opponents, so spreading those teams out gives them a chance to sink or swim against competition representative of other states. As a coach, if I was going to send my team 1000 miles to play, I would prefer to have them play teams they haven't already played several times. NAQT and PACE can't control that after the first few rounds, but they can control it for the first few rounds, and I'm glad they do.

[matt979]

That's still really impressive though, to make that sort of commitment to teams playing other teams from different states. I'll bet that takes a lot of work for NAQT to set that up.

The process used to take more time (and be more manual) than I'd care to admit, but now the seed adjustment and card assignment randomization happen via perl script.