Sunday, September 25, 2005

Seeding

There has been a lot of talk on RSD about seeding. Some of it is about how folks feel certain tournaments should be seeded. Other points are about the criteria and parameters for seeding. Interesting stuff. Here are some things I discuss with my coordinators when it comes time to seed UPA sectionals and Regionals.

1) Seeding is reactive, not predictive.

Subtle point. I am terrible at NCAA type brackets. Folks who care, and think that seeding is predictive, think I suck at seeding. I probably do, but at least this gives me an out. In my two years of seeding Nationals, I have successfully created a legitimate "hell pool" each year. In 03, it was pool A (seeds 1, 8, 9, 16). They finished 2nd, 9th, tied for 3rd, and 6th, and the pool featured two double game point upset wins. In 04, it was pool B (seeds 2, 7, 10, 15). They finished 8th, 5th, 2nd, 10th. Pool play resulted in several double game point victories and a 3-way tie first. This is the kind of stuff that makes one proud.

Seeding is a relative strength ranking of the teams based on past data. The UPA SRT has been great in helping to provide a database of results, but the data is not at all comprehensive. Many teams and tournaments do not report results. When it comes to seeding Regionals, many teams have only their Sectional results to report. In the face of lack of data, seeding becomes subjective, and this is where data like past history, for example, can provide a guide.

Perhaps I will seed Nationals based solely on the RRI this year. That would create some great drama as the results from Regionals came in...and would provide incentive for reporting results next year.

2) Rank the teams, as best you can, in order of relative strength. I think it is useful to generate such a list without adjustments for Regional/Sectional rematches, but with respecting Sectional finishes.

3) Consistency. I view this as a mandate of my position. I try, when discussing seeding with my coodinators, to apply the "rules" of seeding equally and consistently across the sections and regions.

4) Tweak to avoid Sectional/Regional rematches.

The importance of this varies wildly from person to person. Regional rematches in pool play at Nationals can impact Wildcards. I have talked with players who insist that rematches are to be avoided, and, for example, that the teams from a Section or Region must be evenly distributed across the pools (so, if a section sends 4 teams to regionals, then 1 team should be in each pool).

My view is that the purpose of Sectionals and Regionals is to qualify the best teams to the next stage. The Formats Manual is a study in consistency in trying to achieve this purpose. I think seeding should be handled with equal care and respect. If, after ranking the teams, the seeding can be tweaked in a manner that does not create obvious over- or under-seeding issues...then make the change.

5 comments:

Tarr said...

I agree with you on the need to construct an "ideal" order of teams, without concerns for sectional rematches (or even sectional order). To expound further on your "tweak when it doesn't distort" point -

It's really a question of uncertainty. If there's a strong general agreement that team X is the 8 seed and team Y is the 9 seed, then seed them 8 and 9, even if it creates a sectional rematch. But if (IF!) there's a decent amount of uncertainty there (e.g. maybe team Y is really 10 or 11), then pick the seeding set, among the reasonable possibilities, that avoids rematches the best.

parinella said...

Regarding reactive vs predictive, and the creation of hell pools: while it's fair for the team that underperforms in the prelims to be given a bad seed, it's not really fair for the teams that end up in its pool, while teams that didn't do as well end up with easier pools. In your '03 example, seed 8 would have been more likely to advance further had they been seeded 10th. Or in '04, seed 2 earned a top seed, only to find itself with two very good teams while say seed 3 had it easy.

I'd also be leery of your possibly half-joking suggestion of seeding this year based solely on RRI. You would be implementing a policy retroactively. I still seethe remembering how the UPA decided after 1996 Nationals that there was a hidden secret sportsmanship clause in determining which teams from Nationals qualified for Worlds.

degs said...

I think the Mixed div in Sarasota should have the extra crossover on Friday (3 seeds from power pools playing 2 seeds from lower pools). I would say that since the division is notoriously difficult to seed, and inconsistent, this additional game might better mitigate the problems of bad seeding. Not that seeding is ever bad, per se (George), it's just an impossible task. Like rebuilding Iraq.

gcooke said...

Jim,

To be fair...to me, in 03, the hell pool was created by the 16th seed playing out of their minds and finishing 6th overall. I think their seeding was fair as no Central #3 has finished higher than maybe 15th or so.

In 04, the mistake was underseeding Chad Larson at 10th (they made it to the finals). I don't think that anyone predicted, though, that they would do so well. I think my mistake there was not respecting their win at CHC enough. I do have an out, though, as the feedback I received from a CLZ source was in line with their initial seeding.

I do agree with you, though, that for the reactive model to work well, there needs to be good data.

Which brings me to.....

the seeding Nats by RRI was not serious.

gcooke said...

degs,

I have heard that suggestion before.

I think what is going to be interesting this year is that will be much more parity top to bottom than in previous years.

I really felt that the tournament was about 10 teams deep last year, and, if you look at it, Pool B was hell, but Pool A had 2 semifinalists in it.

My point is....i think there is the potential for all the pools to be very hard.

I think the #4 NW will be good.

I think that the #2 South team will be very good.

I think that the #3 team out of the Central will show up this year.

My main concern is entering in all the teams into the SRT Seeding Tool and sitting down on Oct 10 to see very little head to head data.

A couple of other points:

I make fun of myself, but I believe folks (maybe even from B+) know that it is a hard task to seed Nationals and that I give it my best shot.

I do think that the format gives teams ample time to get to where they deserve to be.

For example, I played on a team in the A pool in 03. I seeded us 8th. We went up on the 16 team 8-4, and then 12-8...lost 15-14. We then played B+. Took half, then lost 12-11. We ended up finishing 9th as we did not have the opportunity to play into qtrs.

Should we have had that opportunity, or, do we suck it up and take responsibility for our inability to close out games?