Double Elimination

Remote-Controlled Dave · Fri Feb 18, 2011 10:26 pm

Gary, what happens when we get to the "finals" section of the competition? Those group sets would get us up to the top 8 robots, I believe, of which 4 will have 2 lives remaining and 4 will only have one.
I'm guessing it pretty much speaks for itself at that stage of the competition. I think I'm right here:
- the 4 with the 2 lives left fight each other in 2 fights, leaving 2 with 2 lives and 2 with 1
- the 4 with only one life fight each other in 2 fights, leaving 2 with one life and 2 out
- the 4 with 1 life left fight each other in 2 fights, leaving 2 with 1 life and 2 out.
This leaves 4 robots, 2 with 2 lives and 2 with 1.
- the 2 with 2 lives fight each other, leaving 1 with 2 lives and one with 1.
- the 2 with 1 life fight each other, leaving 1 with 1 and 1 out.
- the 2 with 1 life fight each other, leaving 1 with 1 and 1 out.
- the final is between the one remaining one-lifer and the 2-lifer.

Of course, this works much better in practice than written down

But I think I'm right, if you can follow it.

Post by **BeligerAnt** » Sat Feb 19, 2011 7:38 am

That's far too complicated to try to follow at this time of the morning! (Although it sounds plausible.) But never fear, Mr Waller has charts to cope with finals for 2, 4, 8 and even 16 groups!
I think the spreadsheet currently implements the more common 4- and 8-group versions.
When Modelworld is out of the way I will get the remaining new sheets sorted out and send them to Peter. I will also update the spreadsheet so that it follows the new charts properly. Then I think we'll be ready to go!

Andrew_Hibberd · Post by **Andrew_Hibberd** » Sat Feb 19, 2011 9:01 am

Sounds about right dave, but its still too early. I can't handle my beer like I used too...

This picture also looks ok, but remember the final can happen twice.

Remote-Controlled Dave · Sat Feb 19, 2011 11:18 am

Yeah, one robot having to beat the other one twice to win is the only thing I don't like about double elimination, but its the only fair way of doing it, so game on.

I think, when I get down to this final stage, with the winners and 2nd placers from each group, I will do another quick draw, just to keep things interesting.

Andrew_Hibberd · Post by **Andrew_Hibberd** » Sat Feb 19, 2011 5:23 pm

mixing up the winners braket and mixing up the loosers separatly?

Yeah the having to beat them twice is annoying, but if someone wins all the way through to the final to fight a robot that has already lost one it makes sense. I am too lazy to look up previous results, but has there been a case of the 'bot who has lost one' beating the 'bot who has not lost' first then lost in the second fight?

Remote-Controlled Dave · Sat Feb 19, 2011 6:04 pm

There's definitely been finals that have been played twice at AWS's. Not sure if the "one that only has one life left" has ever beaten the other one twice to go on and win it though.

Post by **BeligerAnt** » Sat Feb 19, 2011 7:22 pm

Dave, there's no need to re-draw at the final stage. Peter's charts carefully mix up the groups so that the 2 robots from one group don't meet each other until the end (if at all). I don't think there's any need to re-draw as everything gets shuffled up anyway. A pseudo-random draw wouldn't really add anything.

I will post all the charts next week when Modelworld is out of the way!

Remote-Controlled Dave · Sat Feb 19, 2011 8:31 pm

Ah ok. Thanks Gary, think I'm pretty much sorted in my head what to do. I've done it before, but not on such a large scale. lol

razerdave · Post by **razerdave** » Sat Feb 19, 2011 10:33 pm

At Snibston we had a mini competition, and I got to the final and hadn't lost, and got beaten twice by Not-so Flippant (who only had 1 life) in the final, so it does happen, all be it rarely

Post by **BeligerAnt** » Sun Feb 27, 2011 10:29 pm

New, improved charts and an updated AntLog:
http://www.robotwars101.org/phpBB/viewt ... 7352#37352