Hacker News Reader

Here's one leading up to the election: https://www.nytimes.com/interactive/2016/upshot/presidential...

And here's the page I was talking about: https://www.nytimes.com/elections/forecast/president

Now it says >95% Trump, but on that day it was the other way around--something like >95% Hillary.

The early predictions were based on the early polls, and the later predictions were based on data after election results started to come in. A lot of the States that ended up swinging were on the eastern half, so the rapidly changing outlook makes sense if you look at the bigger picture (rather than just the number at the end).

That being said, predicting the outcome of an election with some 250 million potential voters based on a few hundred thousand polled individuals is a tricky game to play.

Reply on news.ycombinator.com