Thursday, October 13, 2005

Seeding Rules

Is it time to get rid of the requirement that the team that finishes higher at regionals has to be seeded higher at nationals? It seems like this rule consistently forces seeding that doesn’t gel with the common sense perception of strength. I think it’s particularly silly in the scenario where the 2 and 3 seed don’t even play. In open this year, I don’t think there were any huge surprises at regionals, but it sounds like their might have been some craziness in coed. Anyway, there are examples of other sports that don’t force this requirement – The NCAA Basketball Tourney for instance – I was just wondering what people thought about this rule.


Flo said...

Regionals is often the first time in the year teams play at full strength, so at least one should weigh these results very heavily for seeding at nationals (similarly from sectionals to regionals in competitive sections).
But I agree, by strictly using these results in case of an upset that everyone agrees on was an anomality, you do two things:
1. Reward the team that pulled the upset at the earlier stage and punish the team that lost by giving them easier/harder opponents.
2. Create unbalanced pools (or brackets), affecting 6 (or more) teams that had nothing to do with the upset.

1 is a good thing, 2 is bad. Often times people only see 1, and argue "you get the seed that yo earned, win when it counts." They completely forget about the other teams affected.

I think the current seeding rule is too strict.I would go so far to say that results from the previous round should count very heavily, but not to the point that the order is manditory.

The case AJ is talking about: Central mixed.
The format was 16/3 out of pools. One team (Method) lost an early pool play game against Madhaus and then won out, finishing first in the region.
Another team (CLX) won all their games but front door semis (lost to Meth), finishing third.
A third team (woodchuck) lost to both CLX (in pool play) and Meth (front door finals) and finished second.

The format is set up in a way with pools and cross overs, so that it gets the best two teams in the finals, no 2/3 necessary (it would be game no.5 on Sunday)---assuming consistent play. Meth fucked up in pool play, ending up on the side of the bracket of the arguably second best team CLX, kicking them down to third.

There is no argument to place wood over CLX: they both lost to Meth, and CLX even has the head-to-head.
What else is strange: If they had played the other official format with brackets, it would have most likely been 1. CLX, 2.Meth, 3.Wood (if wood even qualifies). After the upset, Meth would not have had a chance to play for first. So in the end, the choice of format made a 3-seed out of central out of a 1-seed, in other words, a 15-seed at nationals out of a 5-seed or so.

Both formats have advantages and disadvantages to them, neither one is clearly better than the other. I personally like pools better for 16/3...

But a little seeding policy change should happen at least: If the results of the previous stage don't determine the order of two teams, then the order determined by the format should not be mandatory for seeding the next stage.

What it means for result to determine the order needs to be defined, but that should be doable.

Anonymous said...

I just can't understand why the bracketed version was not used. In the MA, we have always used the bracketed format with no pool play. It just makes no sense unless they were just trying to have teams play more games (ie. not eliminated on saturday).


Jon said...

Brackets were proposed by the RC, but the response from teams was lukewarm and the central has typically not had 16 teams, so the tradition has been pool play. Plus, our #16 seed was pretty shaky (only 4 players showed up in time for the first round) and you need 16 teams for brackets.

Also, I think there was some consideration for eliminating teams as late as possible. That is one of the guidelines in the tournament format manual, and about half of the teams in the region had a legitimate shot at advancing.

meth(od) #3

Flo said...

it's just as easy to construct a scenario similar to the one on the weekend with the bracket version. It's not the format.
The problem came from the fact that Madhaus upset Method, a clear inconsistency in the tournament compared to all other results. Similar inconsistencies will lead to questionable rankings in the bracket format as well.

The fact that a lot of people are used to bracket formats at regionals doesn't make them more fair.

The only format relatively safe of such results is a complete round-robin---and there is just no time for this in one weekend, aside from the other difficulties you would run into.

IdaHO said...

why not play it out? who can plan for upsets? deal with them. who's being punished here? in the open division it's not a matter of who you might play, but when you play them. if you can't beat them at various points in the tourney then your team is not as solid or prepared as you think. put away your slide-rules and stats and stick to the format. don't like it? don't lose games at regionals. don't like that 'team x' lost and you're being punished? take the week and a half to prepare for the game (defensive stunts, film, matchups...etc.) or go to a different division where you might be more dominant.


sometallskinnykid said...

Does not going by regional placement give a little bit too much subjectivity?

The thing about March madness is the search committee who has supposedly watched every minute of every game and "determines" who is better than who (a very big generalization for the greatest time of the year).

So, do we get a search committee? Who do we put on it? I think not going by regional placement can really screw with pools even more. Captains *could* try to argue for a better pool just because they want what they think is a better game. To allow the captains free reign over putting teams anywhere, much tougher and much more subjective (i dislike subjectivity).

I like the pressure at regionals that, in a way, you are also playing for seed. I felt this past weekend if we won, we would get the 10 seed. If not, probably 12/13/14. We will see what happens, but I am fairly certain we will not be lower than 11.

It gives something more to the regional winner. Something they earned by playing well over the most important weekend they played that year....

Maybe if I were on the seed committee, I would approve.

aj said...


I think, yeah, you'd have a seeding committee. Basically it would be composed of some group of knowledgeable non-natties participants. I think the SC would take in-put from captains and then make their decision.

I don't think it's necessary to have watched every game to make an informed decision. You can compare teams based on their scores against each other and common opponents.

Based on the way you seeded over your blog, I'd vote for you to be on the sc...after you retire of course.

As Flo pointed out - it makes sense to reward the team that had the succesful regional upset, but it's doesn't make sense to create unbalanced pools thereby punishing three teams who weren't involved in the upset.

parinella said...

I'd be in favor of making the rule conditional on team A actually having beaten team B at Regionals in knockout play (or A beat C who beat B) and applying it only to the non-#1 finishers.

And regarding consistent play: does someone want to take a look at all rematches within Regionals to see what the records are? In the NE (all divs), I know of two splits (Godiva/Brute Squad and Chinstrap/7 Slow) and two sweeps (Metal/Goat and Capitals/Ambush) involving teams that made it (there may be others).

I vote that Idaho's team has to play against the best available team each round while my team gets the worst.

Frito said...

Why not just go straight with RRI? I looked at the seedings for open, and it goes something like this:
1) Sockeye
2) Jam
3) Condors
4) Furious George
5) Johnny Bravo
6) Ring of Fire
7) Death or Glory
8) Sub-Zero
9) Pike
10) Chain Lightning
11) Doublewide
12) Twisted Metal
13) BAT
14) Potomac
15) PBR Streetgang
16) Viscious Cycle

I don't see what's wrong with that. If you apply the same idea to coed you get:
1) Hang Time
2) Charles Larson Experience
3) 6 Trained Monkeys
4) Method
5) Hot and Sweaty
6) Shazam
7) Brass Monkey
8) Slow White
9) Gorilla Foot
10) Donkey Bomb
11) Ror$hack
12) Drive Through Liquor
13) Bad Larry
14) Woodchuck
15) Olio
16) Mischief

Not knowing many of the club mixed teams this year, I'm not exactly sure how "fair" this is. But who really cares about fairness. First of all, the mixed division experiences the most turnover out of any of the UPA divisions simply because many of the top mixed players make their way to open/women's after a year or two, or the people that are playing mixed are older and nearing the end of their ultimate careers. I think going strictly by RRI would force teams to reconsider their strategies on what tournaments they go to. Going to one or two good tournaments where you play good teams with be much more beneficial then going to several sub-par tournaments where you will destroy the field. Of course this is a difficult measure for the mixed division since there is so much turnover, but perhaps enforcing RRI as the standard for seeding would reduce some of that turnover (not that I don't want to see the best men/women playing open/women).

I don't know...just thoughts.

bakkenra said...

Going with straight RRI for Mixed this year would be a disaster. The reason being a huge lack of inter-region play. Hangtime, the #1 RRI hasn't played more than like one game out of Texas, sorry two out-of-region games.

The formula for the RRI (the little look I have given it) means that you could have essentially seperate RRI groups if there is not enough team mixing. If you had multiple teams on an island, and one team was just dominating the others, they would have a high RRI, but it wouldn't really be an accurate measure of strength.

Reid B

bil said...

What's our count on the 16.3 pool play version failing to correctly seed the 2&3 out of a region, vs. our count of 16.3 pool play giving us palatable results?
I know that we have had funny seedings in Mixed two years in a row with the no 2-3 game being played. As has been noted, this year's Central results SEEM incorrect and last year's NW regionals had RFBF taking 2nd without ever playing Brass Monkey, which in turn resulted in funny nationals seedings.
I'm having trouble reconstructing previous regionals because of a lack of reporting information, and since I don't know when the current 16.3 pool play tournament format was instituted. I think I remember 2002 having a different 16.3 pool play format, and different rules since Shazam didn't play our 2/3 game because we lost to that team in semis.
Do we have any information that bracket play do better? And if we wanted to make 16.3 pool play the more difficult option to use, how would we go about that?

bil said...

whoops. i meant 16.3 bracket play the second time there.

as in:
>vs. our count of 16.3 bracket play giving us palatable results?

homrbush said...

The problem from Centrals stemmed from only 2 of the top seeds in pool play went 3-0, then 1 of those 2 lost in the crossovers.

So the top 4 seeds going into the tourney ended up in the same half of the quarterfinal bracket. Whoever lost in the semis was going to have this kind of gripe as whoever came out of the other half of the bracket (either the 5, 6, 8, or 10) was going to be a worse team.

There really isn't a major reason why you have to seed teams based on regionals results. You play at regionals to make it to Nationals. Whether you get there with the first bid or the third, the point is to get there.

UPA should be able to seed teams appropriately. It's not like there are going to be many (if any) teams that no one knows anything about.

Flo said...

The analysis of pool versus brackets is a complicated story, the main argument boils down to the following:
look at four teams and compare pool versus doubel e, and asume some inconsistency (with full consistency, all sound formats are fine).
A beats B beats C, but C beats A. A, B and C beat D. Order comes down to point differential.
double e:
seeding A,B,C,D. A beats D, B beats C, then A beats B, C beats D, then B beats C (again). 1.A, 2.B, 3.C, 4.D
seeding B,A,C,D. B beats D, C beats A, then B beats C, A beats D, then C beats A (again). 1.B, 2.C, 3.A, 4.D
seeding C,A,B,D. C beats D, A beats B, then C beats A, B beats D, then A beats B (again). 1.C, 2.A, 3.B, 4.D
So the finish depends on the seeding.

What do you rather have, a finish depending on point dif or a finish largely depending on seeding? You might say the first, but then look at the results again: you don't see anything wrong with the results after bracket play (since the upset game was never played), so the results feel right, where in pool play you always have the sour aftertaste of point difs with all the attached problems. Difficult call.

The same philosophy differentiates the two 16.3 formats a bit. With pool play, a few more games are played, leading to a few more chances for an inconsistency or intransitivity (a 3-way) to be discovered. Whenever this happens, the results may be better, but feel wrong. Or they are most likely wrong as it was the case with the CLX-Woodchuck distinction.

Why I like the 16.3 from pools:
It leads to a game-to-go between two games coming out of the same type of game the round before---a true do-or-die semis against comparable opponents. None of the old problematic backdoor with one team who just got beat against a team that just won---against two teams of potentially very different strength.
The format is still sound, and with consistent play it finds the right order.

In a perfect world, you want to play the 2/3 game in the end for seeding, but the time for this is just not there.

CoachingBig said...

I'll advocate for tossing seeding out the window. Make it all random draw and expect teams to win their games to make it to the top.

Anonymous said...

Yes,I agree with you....



Clinical Depression