I'm trying to use Trueskill in Python, specifically this library: http://trueskill.org/, to generate rating from match results.
However, I didn't see obvious way to do this. From the document on the site, given two ratings, one can simulate the rating updated via rate_1vs1
. But I didn't the other way around, that is, given match results and generate ratings for all the players.
Any hints on this? Or should I use another implementation? Thanks.
Well, in fact rate_1vs1
is the way to go. By looking at the source code I think the assumption is that the first argument is the winner. i.e.,
# if alice win
alice, bob = rate_1vs1(alice, bob)
# if bob win
bob, alice = rate_1vs1(bob, alice)
# if drawn
alice, bob = rate_1vs1(alice, bob, drawn=True) # order doesn't matter in this case