What is the mapping between Rybkas evaluation scores and winning percentages?
In normal positions, each centipawn of Rybka’s evaluation is worth roughly 3 Elo points. For scores reasonably close to 0, this translates to approximately 0.5% in winning chances. So, for example, an evaluation score of +0.20 is worth around 60 Elo and translates to an expected score of 60%. More discussion can be found here.