Adding some kind of confidence bound or estimated probability distribution might make more sense of the huge ranges. You're already computing the 'height' of p

Adding some kind of confidence bound or estimated probability distribution might make more sense of the huge ranges. You're already computing the 'height' of partners, so you could potentially incorporate that into the model. Something like "$100 - $1000, best-guess $500 ± 200"