You can't really but if you were to think about it that's how I would go about it. You can test different values and probabilities that you think are (likely|conservative|optimistic) and go from there. That gives you a range and a ballpark for different scenarios.
Personally though, if I can't predict something with > 90% accuracy/reliability then I'm not interested in investing in it.
Personally though, if I can't predict something with > 90% accuracy/reliability then I'm not interested in investing in it.