[ QUOTE ]
this thread is linked to in the discussion I beleive you are referring to. If that's what you're talking about, my point is that they all use % chance to win as a constant, which I am arguing it isn't. I beleive this makes variance significantly lower.
[/ QUOTE ]
You're right, win rate isn't constant, it varies from opponent to opponent and (hopefully) improves on average over time as your skill develops. I'm not convinced it significantly reduces variance though, unless you are excercising selectively stringent game selection (i.e. when you hit a run of poor results you tighten up your game selection, electing to play only poorer opponents until you turn things around). If you practice game selection as a constant your average win rate should reflect this.
I might run some crude simulations to settle this, it's an interesting question. If 60% is your average win rate, what do people think a sensible range of expected win rates against different individual opponents is? 40-75%?
RH