S_Alexander: I think what’s going on is the fast analysis works like this: we take a quick hunch of all the moves, pick what we think are the top 3 (actual top three is somewhat subjective, or at least non determinstic). Then we do a full analysis on those three moves, and however the stats work out from that is what we use for the change in outcome. So, while not a perfect system, I think it tends to give pretty good results. I think looking at the full analysis, it’s quite possible that it got it kinda right, but the outcome prediction was kinda weird none the less - i need to dig into how that value is computed again.