gcode-mirror / ai-contest

Automatically exported from code.google.com/p/ai-contest
0 stars 1 forks source link

Switch to swiss tournament instead of random pairings #23

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago
Here is how the matching system works right now: choose two different 
submissions at random.

Here is how it should work:
  1. Find active submission with the least games played.
  2. If the submission has no ranking, choose an opponent at random.
  3. If the submission does have a ranking, choose an opponent of similar rank (+/- 10 or so)

Original issue reported on code.google.com by cameron.jp@gmail.com on 12 Aug 2010 at 12:06

GoogleCodeExporter commented 9 years ago
I have one suggestion to add if I may:

Your initial match should be someone with an ELO/rank similar to the ELO/rank 
of your last submission. I think at the moment it is random and if you are 
started 1000+ ELO/rank away from where you last submission ended, it takes a 
looooooong time to play your way back - usually it's just better to keep 
resubmitting until you get a decent initial opponent. This would also help keep 
down the number of submissions.

Thanks!

Original comment by Teeb...@gmail.com on 6 Oct 2010 at 10:04

GoogleCodeExporter commented 9 years ago
I would like to propose a slight adaption to this suggestion which would result 
into faster convergence of true ranks:

  1. Find active submission with the least games played.
  2. If the submission has no ranking, choose a bot ranked around the center of the rank spectrum with the most played matches so far (i.e. most stable ranking).
  3. If the submission does have a ranking, choose an opponent of similar rank

Original comment by folkerth...@gmail.com on 26 Oct 2010 at 1:04