I was reading some papers and ended up getting some references to this paper:
ftp://ftp.research.microsoft.com/pub/tr/TR-2006-80.ps

It is about a bayesian method to estimate skill of player. For those who do not want to go through the real paper, this website provides some information
http://research.microsoft.com/en-us/projects/trueskill/default.aspx

I do not have the time to implement one of these, but I am sure that at least Skrum would be interested on looking at the paper.

cheers,

me again