It's interesting. The "fundamentals" as I've heard them called were calling this as a close race two full years ago. Decent but not strong economic growth, America doing poorly abroad but in no immediate crisis, two-term incumbent stepping down but in a Congress ruled by the opposition, etc.
The personality and campaign race has been anything but. When election day rolls around we'll see which were correct. The issue is that if its a close result, it can be either because of the fundamentals or because of a relative tie in the personality/campaign way. If it's lopsided, it's a clear and obvious win for those who argue that campaigns and candidates matter, to the point that they can eclipse economics and statistics. Interesting from a political science PoV, certainly.
The timing of the shift in voting did coincide a little too closely to the FBI's announcement, but I really can't believe we'd have that many undecided voters waiting until the last week to make up their minds.
It's worth reimagining "undecided" for a minute.
When you hear "undecided", it implies either the low-info voter, or the hardcore neutral. But it can be larger than that, though. Consider, for example, that Trump is doing historically poor with Republican voters, winning only 78% ish of the vote consistently. In an era when winning 95% is standard, that's very bad. But what happens to that missing 17%? They don't disappear you know. They get stuck; unwilling to support Trump or just too sick of him to commit. but opposed to Clinton for ideological or partisan reasons. These are the sorts of voters who are on the edge; they aren't necessarily deciding upon both, they might merely be waiting to see whether their disgust of Clinton makes them vote or disgust of Trump stops them, or deciding whether to vote Johnson or Trump.