Variants on the Swiss Format
Kevin Carter addresses some of the variants on the Swiss format.
For Swiss events with up to 24 players the optimum number of rounds is n+2, where 2n is the next highest power of two above the number of participants. There is about a 90% chance of an outright winner. Above 24 participants the optimum moves gradually in favour of n+3. (I posted a detailed analysis of this to the newsgroup on 27.5.98)
We generally overcome this by using a mutation, usually known as a 'Flexible Swiss'. Players in each round are matched in accordance with their availability, rather than their strict position in the table. This eliminates long waits and also accommodates players wanting different numbers of games overall.
Further, I have run Swiss events with 48 players on 10 lawns; I ask for 8 volunteers each morning to start late, and generally have no difficulty finding enough people wanting a lie-in. During the day having 8 people (or 1/6th) waiting for their next game is not seen as a great problem. If the average game is 2 hours this means the average wait is just 20 minutes.
I think this is unfair. I refer you to David Appleton's seminal work published in 'The Statistician'. He compares the probability that the 'best player' wins, given different ways of managing a tournament. Best of all are double round robin and seeded draw & process; towards the bottom are blocks and a simple KO; Swiss and single round robin lie in between.
Regards, Kevin Carter; 5.1.00
All rights reserved © 2000