for starters, the "stay" camp was only slightly ahead of the leave camp - well within the margin of error:
http://elections.huffingtonpost.com/pollster/uk-european-union-referendum
Trump, however, is outside the margin of error in many states that he would need to flip. For example, every poll in New Jersey (
http://projects.fivethirtyeight.com/2016-election-forecast/new-jersey/#plus) is past the margin of error. And it's getting worse by the week. Same with Wisconsin save for a poll before the first debate:
http://projects.fivethirtyeight.com/2016-election-forecast/wisconsin/#plus
On top of that, you add in fivethirtyeight's aggregation of multiple polls -- sort of a meta-analysis -- that reduces the margin of error. This is their current electoral map:
http://projects.fivethirtyeight.com/2016-election-forecast/#plus
Relying on the polls being wrong is maybe a nice mental saving exercise but not a viable belief system. If the polls starting moving in Trump's direction so that he's closing the margin as the election gets near, then maybe that argument would be more realistic.