My last projection

Here is the data I used for the AMSTAT election prediction contest (data comes from 10/27-10/30). I predicted Red states with polls pulled on 10/27 in the evening, blue states and swing states (except for Iowa and Ohio which I reran on 10/30) on 10/29 in the morning.   Now I do not believe that the predictions from this weekend will be accurate. I think things are changing. This does not reflect the news of the possible reopening of the Clinton private email investigation.  I don't know if on election day it will work.
Another story broke today about the release of FBI files relating to the investigation of a presidential pardon of a Clinton Library donor.  I also think that some republicans may be waiting to vote for Trump while awaiting more information on all the candidates before they decide whether or not to vote for Trump.
Here is an CNN article about early voting: http://www.cnn.com/2016/10/29/politics/early-voter-data/
We don't have real results but states release data on who has voted and sometimes that includes party identification. However not all Democrats will vote for Hillary Clinton and not all Republicans will vote for Donald Trump.

States I am Watching Closely

Nebraska

Nebraska electors are assigned by congressional district, so I am currently debating which way the 2nd district will go.  There is only one poll that released information on that district.  The house race there has a democratic incumbent.  I predict a Trump win overall but it's hard to choose what will happen in the 2nd district which is almost entirely the city of Omaha.  Obama won in 2008 but not in 2012.  On Sunday I thought the 2nd would go blue because of lack of support for Trump.  Today I don't know what I think.

Ohio

The polls are tightening and my model has gone either way depending on the day.  If I have to make a call in a case where my model is inconclusive, I am going to lean towards Trump.  The heroin crisis is especially rampant in Ohio, so much so it makes the national news regularly.  Trump is tough on crime, and that along with the new news about the Clintons may tip the race to Trump.  The rule in Ohio is a statewide race within 0.25% will be recounted (https://ballotpedia.org/Recount_laws_in_Ohio).  I don't know if it will get that close but I think Ohio will be one of the closest states in the race.  Today I ran it and found a Trump lead.  This will be the last race I call because it is so close.  Right now registered republicans are leading but I don't know how many of those Republicans voted for Trump or went for a write-in or independent candidate.

Iowa

Iowa is not as close as the other states I am worried about.  Trump has been leading for a couple of weeks. My main concern with Iowa that Iowans can (and normally do)vote early and since Iowans started voting on September 29th, before the all the October surprises.  According to CNN, more registered Democrats voted  than Republicans as of 10/30. I am having trouble finding the original data.  I do think there are some republicans waiting to vote because they are deciding whether or not to vote for their parties candidate.  I think most republican/ conservative voters will vote for Trump, but I think they want a bigger picture and may vote later.

Florida/North Carolina

I am pretty confident of a Clinton win in Florida and North Carolina. As long as minority voters turn out (Clinton has a strong lead among both Hispanics and African Americans) I think Clinton  will win these states but it will be close.

Here is a text file of my detailed results: https://drive.google.com/file/d/0B1diLu79C03gR3o0TUJJVUEtNms/view?usp=sharing