Thread: An Alternative Tournament Structure: The Rolling Rumble
View Single Post
  #7  
Old April 16th, 2009, 07:27 AM
ollie's Avatar
ollie ollie is offline
Is a Quadradical
  
Join Date: March 19, 2007
Location: VT
Posts: 9,544
Images: 43
Blog Entries: 22
ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth ollie is a man of the cloth
Re: An Alternative Tournament Structure: A Recipe for Chaos?
Quote:
Originally Posted by Onacara View Post
I think you play the next available player whether you played them before or not with the 1 exception that you do not play the guy you just played.

With the Rumble format their was an extra guy on 1 team which almost always assured you faced a different player each time.

Perhaps using this format with only with an odd number of players would work best.
I think I still prefer the idea that you wait for a new player wherever possible for a couple of reasons. First, as Dok says, more mixing is good when determining ranking (pretty much however you do it). Secondly, as a player, I'd rather play lots of different people. This system will give, I think, more 'scaping time for (almost) every player than a regular tournament anyway. I don't think the odd number of players is noticeably different. The standard to compare to is not continuous playing for everyone; it's how much 'scape gets played at a regular event. There is always a game, and usually several, that are over in 20mins or so. Those players are getting their waiting time cut from 40mins to under 10.

Thinking more about it, I think I prefer the second option I gave in my answer to Kobu regarding repeat plays: only when there are no free maps do you consider playing someone you've played before. I have a 20 person tournament as my mental model. My guess is that fifteen maps (so ten active and five empty to begin) is about right.

Quote:
Originally Posted by dok
I was actually just having a PM discussion about a better strength-of-schedule metric. I am a big ranking algorithm nerd, and I went into a long description of how a good ranking algorithm could work and could be used for determining place.

It occurred to me that if the algorithm is good enough, you really don't need to match people up too carefully (e.g. in a Swiss format) in order to get good results. You just need a fairly well-connected set of games between all the players. But a random scramble of games will achieve that the majority of the time.

So, I like the format, particularly for a more casual atmosphere, but if you care about determining a winner fairly, I would sharpen up those scoring metrics.
More details? I want something that people can immediately calculate themselves on their index card. That's why I mentioned dropping SoS altogether.

Rather than win percentage, how about number of wins minus number of losses? (Win difference? Does this have a name?) This means that someone who goes 6-1 beats someone that goes 4-0, which I think is how it should be. Number of wins (or, equivalently, number of games) would still be the second splitter (So 7-2 beats 6-1). D20 to split equal records at the prize table.

When thinking about this system, I think the trade-off between an obvious winner and more 'scape is the central point. The more I go to tournaments, the less I care about the actual standings and more I care about the games themselves. The extra 'scape and still getting a pretty good ranking out seems like an improvement to me over our regular structure, but I can see that others might not agree.
Reply With Quote