Concur. A *few* Republicans will cross over, vote 3rd party or not vote at all. But even more Democrats will vote for Trump. I suspect the majority of Independents will as well.
I understand the desire to prognosticate by noting the number of registered voting ballots, but I prefer dealing in facts. Might be because in my state we don’t declare a party when we register to vote, so the prognosticating is a moot point with us.
Assuming that a voter will vote the same as his/her party registration is a sort “worst case” scenario, if you also also assume that the best case is where any crossovers will be disproportionately in your favor.
On the other hand, if you assume that any crossovers will go against you, then it’s a best case scenario.
It’s a reasonable “null hypothesis,” because it takes a neutral position on crossovers.