I based 1.5% on two faulty assumptions. The first was that I used Michael McDonald's estimates of remaining votes. His estimates in many states were very low. He's since adjusted his spreadsheet. The second was that I expected the late vote to only be slightly more Clinton than election night. That was really wrong. Clinton won post-election night Colorado by 22 points. She won Maryland by 39 points. She won Georgia and Florida, states she lost, by 26 and 22 points. Clinton's post-election night margin in California was 34, up from 28 on election night. The post-election night vote has gone to Clinton 57.9%-34.5% so far. That's somewhat remarkable considering Clinton and Trump were separated by 0.1% after election night.
I got this one wrong. I apologize for the mistake.