Furthermore, Trump's formula for winning PA was not predictable in advance for anyone who wanted to stay politically relevant - if you went around saying that Clinton was only going to win Scranton by 3% after Obama won it by 27%, you'd be laughed out of politics in 5 seconds. Also, there was the general line of thinking that a state that votes for a Democrat 6 times in a row (PA) isn't going to turn around and start voting the other way.
So you have to herd with the consensus and ignore obvious facts to remain relevant?
We had polling that had Trump up by 20% in Luzerne County. We had polling that showed Clinton's favorables in the 20's in that area.
We had a fully encompassing analysis of the state that mentioned the Plateauing of Democratic registration gains in the Southeast and Republican gains in registration over the previous three years in the rest of the state. I know I repeatedly mentioned the impact of a protectionist message in expanding the out-state appeal beyond the War on Coal rhetoric, which helped Romney only in the SW, to a message that appealed to working class swing voters across the whole of the state. Plus, I just happened to be from NE Pennsylvania, and have a general feel for it.
The signs were there.