Three Weeks Out - First Prediction of US Election

I have a horrible track record predicting presidential elections. For as long as I have been blogging, I have been wrong, both in 2008 (when I got the right outcome, but with numbers that were woefully off) and 2012 (when I thought Romney would win).
So let's try again...
I now have a simple rule: if fivethirtyeight.com gives a state a better than 70% chance of going for a particular candidate, I give it to that candidate. At 3 weeks out, the chances of tectonic shifts in a state's results are approaching nil.
So let's look at this morning's 70% electoral map:

So even if Trump wins in all the tossups (Arizona, Iowa, Ohio, and North Carolina - as well as the single votes in Maine and Nebraska), Clinton wins 307-231.

OK, let's make this even tougher, let's look at an 80% or better map:

Note that three states move into the tossup column: Alaska, Nevada, and Florida. Even if all these former and new tossup states went to Trump, Clinton is the winner, 272-266. A squeaker.

Now let's slide the yardstick to 90% or better:

Now we can add South Dakota, Minnesota, Wisconsin, Missouri, Pennsylvania, New Hampshire, Maine, and Georgia to the tossup column. This map represents the absolute floor for each candidate.

So here is my prediction: barring some cataclysmic late "October surprise" of tectonic proportions (the release of 33,000 emails, chock full of classified documents; a foreign disaster; a cataclysmic 9/11-level event; Clinton physically collapsing in the final debate; oh hell - maybe an alien "first contact") - Hillary Clinton has won this election.

Which means, which means...

Let us assume that Bannon, Conway et al, can read a map. Let us further assume that Trump can do the same. What this means is that tomorrow night's debate is his almost last chance. I say almost because I suspect there is one more ploy at his disposal - in the final 2-1/2 weeks Trump can purchase prime time chunks of network television as Ross Perot did in 1992. Maybe we'll have a foretaste of TrumpTV. Let's assume that all of these certain and potential upcoming performances are desperate, no-holds-barred, unconventional presentations. I don't think it will matter.

The concrete has set, the cake has been baked, and the train has left the station.

The only questions now are the size of the electoral vote victory, and the margin of the popular vote win. My guess: 305-233 (less than the 2012 margin of victory); 47-45% (about the same 2 point margin as in 2012).

Be certain that I got it wrong yet again.

Actually, I hope for a 1964-style blowout (486-52; +20 margin in the popular vote). Maybe a Clinton-Dole 1996 result (379-159; +9 margin) is more realistic? (h/t @soverytired) The only reason I won't go with a blowout scenario is the "Brexit" factor - I am convinced that a sizable number of poll respondents are not answering honestly.