I also said that Hillary was obviously elected.
There's a difference between making a prediction and making an election call which is what ON progressive is talking about.
He's referring to my model that we can predict an election solely on the margin of the votes and the number of votes that are in. I've witnessed lots of elections, and it generally holds, that the inverse ratio of the votes counted will be the margin of the votes.
So, if 10 percent of the votes are in, a margin of 10 percent would be sufficient to call the votes. 20 percent of the votes being in, a margin of 5 percent would be sufficient to call the vote.
50 percent of the votes being in a margin of 2 percent would be sufficient to call the vote.
As you can see, occasionally this model doesn't work as it didn't with the early leads in VA and AL, Trump was up over 5 in VA, which triggered the early call. Same with Alabama.
The correct way to fix that would be 1/2x, meaning that 20 percent in and a margin of 10 percent is sufficient for a call, margin of 5 percent is callable at 40, and at 2 percent would be callable at about 75 in.
My data shows that this model would have gone 51-0 rather than 49-2. I may yet tweak it, but the math behind it is rather simple and profound.