Re: Known risks with quantifiable outcomes
There is uncertainty, to be sure, but it seems extremely unlikely that Jill Stein will draw enough votes to matter unless she can find a vice presidential nominee of Bill Clinton's caliber. Johnson/Weld is a different matter, however, and are likely, especially if they manage to get into the debates, to draw both Democratic and Republican votes, although substantially more from Trump than from Clinton.
Utah will be an interesting state to study, as the Democrats went quite heavily for Sanders and do not much like Clinton, while the Republicans went quite solidly for Cruz and have a rather strong distaste for Trump. I do not think either VP candidate will change that much. I suspect the bordering areas of Idaho, Wyoming, Arizona and Nevada are rather similar, although they are less likely to affect the electoral college breakdown.