The problem with Elo applied to a game like Toribash is that the system itself was made for Chess, a game with no random factors whatsoever. TB is just too chaotic with all the random fracs and inconsistent damage values across the board (10k rub damage ftw). It's also a lot easier to lose a game of Toribash than it is to lose a game of Chess when players of hugely disproportionate skill levels match up with each other.

I gave some thought to this, but I couldn't really come up with anything better to gauge players under vanilla circumstances. When I want to see how good people are, I just check what mods they play, their belt and their win rate, which tells me pretty much everything I need to know.