by Howard » Wed Mar 10, 2004 7:32 pm
I'm with laszlow on this one. The random thing makes it all the more likely that the top teams will be fighting among themselves for a chance at the top bracket. I'm also not sure what you mean by dividing into a top bracket and a bottom and having the bottom half of each bracket play each other. Does this mean that top 1/4 plays top 1/4, third 1/4 plays third 1/4 and second 1/4 plays fourth 1/4?
I recently ran a 30 team tournament with preliminary rounds in swiss format. While I was generally happy with my choice, here are my assessments.
Positive points:
1. I preliminarily seeded the teams, but this was only of importance to guarantee that the best teams would not be playing each other. Admittedly, the seedings were not terribly accurate, nor did I intend them to be. They were meant to simply be good enough. Later pairings were based on game score and point score.
2. I did not have to do any major overhaul if teams did not show up.
3. It was easy to simply take out teams that withdrew. (Again, no overhauls, no byes).
4. As long as a sixteen team score group was paired 1vs9, 2vs10, ..., 8vs16, it avoided the messiness of having teams of nearly the same ability play each other.
The negative points:
1. It was a 30 team tournament, so the wait between rounds was about 15 minutes while I did pairings by hand. (A computer program may or may not help to fix this).
2. There were 6 rounds, so in round six, the top 4-1 team was punished with playing the 5-0 team. A significantly sized playoff field should reduce the impact of this. Our playoff field was designed to be 16 since that would be about half the number of teams. Other than playoffs, the way to correct this issue in a swiss tournament is to have approximately log base 2 of the number of teams as your number of rounds.
3. Approximately one pairing a round was significantly inaccurate because scores were reversed on the scoresheets. I've concluded that our scoresheet design led to the ability of the readers/scorers to make this error.
I can't stress enough my philosophy that 1 plays 9, 2 plays 10, ..., 8 plays 16. I've seen many tournaments where within a scoregroup, they pair 1vs2, 3vs4, ..., 15vs16 or 1vs16, 2vs15, ..., 8vs9. The 1vs2 example I believe is just horrible because it makes the result of just about every match almost random. The 1vs16 example is much improved, but still at least two of the games are pretty much random. In my system, modeled after the United States Chess Federation's swiss pair system, nearly every game has an intended result. The best teams should not wind up playing each other until near the end of the tournament or preliminary rounds.
The "punishment" described above that the top 4-1 team received by having to play the 5-0 team resulted in them receiving a very low point total in their last round and moved them from the second team to the seventh team of sixteen going into the playoffs. I've been meaning to see how many of the 4-2 teams at the end of the tournament played the top team and what percentage of those finished ahead of the team in question. That will help me determine whether the team in question was moved closer or farther away from the position in which they should ideally have finished. Admittedly, half of the other 4-1 teams had the opportunity to finish 5-1, while the team in question feels it did not have that opportunity.
Although there's a lot to digest here, I've come to the conclusion that if reasonable procedures are followed, swiss pairing is the way to go if wait times between rounds can be reduced to a more acceptable level. I know there is a significant number of coaches who disagree, and I'd welcome any input on either side.