Your views on the Swiss System

[Magic Dave]

What is your view's on the Swiss system.

Please just keep it to the Swiss system for the moment.

[mubo]

Not sure there are any (viable) alternatives.

Problems with Swiss are:
-Doesn't scale well, 6 games is not enough for >60 coaches.
-If TDs used for sorting after w/l/d, you tend to get AG mixed with AG and bash with bash. Some similar issues have been discussed elsewhere eg a game with two good coaches is likely to be 1-0 or 2-1, meaning the winner gets matched vs another winner of a close game. Scoring lots of TDs not necessarily an indicator of good play.

If you have a field of well matched coaches, I think you'd have much greater license to play around with the system, but games should be as even as possible.

[Grumbledook]

a lot of those points are the scoring system problem rather than the mechanics of swiss pairings themselves

maybe dave meant that though?

[Magic Dave]

I was just looking for a general feeling of the Swiss system. The funny thing is that BB is a sports parody, but the way we organise tournaments isn't.

Has anyone ran a tourney with a "cut" at the end of the first day. Then the top half play the second day Swiss style and the bottom half keep playing as well. At the end of the day you have two winners, one from each group?

[Joemanji]

Not sure there are any (viable) alternatives.

Problems with Swiss are:
-Doesn't scale well, 6 games is not enough for >60 coaches.
-If TDs used for sorting after w/l/d, you tend to get AG mixed with AG and bash with bash. Some similar issues have been discussed elsewhere eg a game with two good coaches is likely to be 1-0 or 2-1, meaning the winner gets matched vs another winner of a close game. Scoring lots of TDs not necessarily an indicator of good play.

All good points.

My experience is that Swiss isn't necessarily the "fairest" system for determining a winner. The round one draw is all ... get a weak player / race, you'll get a big win. Then you'll play someone else who played against a weak player/race, often still a good draw. This can easily carry into game 3, and has done many times in practice. All the while the other "good" coaches are playing each other and denying each other points with draws or narrow wins. Then on day two you'll see games even out, but by then the coaches who got a good draw on day one have a serious head start. So from three games you need to play well once and get lucky once, and that's 5/0/1 or 5/1/0 and a tournament win (because of your bonus points from day one).

However ... is there a better alternative? Not IMO.

[rodders]

maybe seeding is the way to avoid this a la pearlies

[Joemanji]

But at Pearlies everyone just takes a low ranked race to get around it. Seeding based on coach rather than race maybe.

[mubo]

I don't think seedings would work, seedings have been designed to keep the best players apart until final rounds, mainly in KO formats. I like them as a twist though. You'd be better off sorting by WLD only and ignoring other tiebreakers.

As Joe says, Swiss isn't the best way to determine a winner, but it does mean all games are as well matched as possible. Which is more important. An alternative e.g. World Cup format wouldn't work with tourney BB, because draws are prevalent and not easily resolved.

[Grumbledook]

it is the best way of having a table top hobby tournament though where people are going to travel to have some fun

it ensures everyone gets the same number of games and doesn't just turn up and gets knocked out straight away

as this thread has gone off from swiss system to scoring, then perhaps organisers should do the following

use swiss as normal, so winners play winners, losers play losers etc

instead of having points each round seeding all those who won etc just randomise them but obv can't play the same coach twice (which can happen for those coaches who drew games)

then after all the rounds factor in the scoring points of touchdown differnce, cas diff or whatever to determine the winner

then you avoid the "problems" of someone getting an easy draw in round one and avoiding certain types of teams for the rest of the tournament

swiss will still play its part in that you will end up playing someone with a similar record to yourself but the hard fought wins are then given equal rating until right at the end of the tournament

iirc at the Chaos Cup galak uses reverse alphabetical by team name (well at least he did one year he may change it every year to stop people being sneaky), I only found this out by asking him why I was the last team out every round of teams on the same record when I had scored more etc

looking back that was akin to being random, though I think using actual random draws would be perhaps better, at least to mix up the race types, someone can obviously still get an "easier draw"

[Magic Dave]

I still think we could do more than just Swiss. I've always seen it as a knock out anyway, with each round there are less and less players that can win the event. Normally by the end of round five your prob down to about 4 that can win it. So by putting a cut in after day one you have two mini tourneys with two winners and by round 5, you could have 8 people that are playing for something.

[SillySod]

Swiss is absolutely fine.

Scoring systems sometimes screw things though. Lots of touchdowns or lots of casualties arent terribly important, I'm not convinced that they merit bonus points. Bonus points are ok when used as a tie breaker but often overshadow points gained from actually winning.

[GalakStarscraper]

Scoring systems sometimes screw things though. Lots of touchdowns or lots of casualties arent terribly important, I'm not convinced that they merit bonus points. Bonus points are ok when used as a tie breaker but often overshadow points gained from actually winning.

Agreed ... that why I just use the TDs and CASs as tie breakers only for the swiss pairings.

Galak

[Grumbledook]

It is using them for the swiss pairings that I think causes some of the issues that occur.

Such as at the Blood Bowl when agility teams will just play against agility teams and bash against bash. Though they do it on just TD scored rather than the combined difference of both.

anyone familiar with score know if it can just randomise swiss pairings based on just W D L?

[mubo]

Yep, looking at it seems you can.

1st sort on points, then randomly. There's an option not to include bonus pts in the total score, so I guess you'd use that. Seems like a better system to me.

[Grumbledook]

Dave if you wanted to mix things up a bit more though would really work best with an odd number of games is to have perhaps games 1+2+3 and create standings from that

then games 4+5+6 create another standings list from that, then in a game 7 play them off against each other

though I personally think using straight swiss all the way works best