let's remember that while Nate wrote a very long and convoluted explanation for his Trump calls in the primaries, the basic point (and failure of his predictions) was that he rejected the standard modeling in favor of endorsement based modeling even though past polling wasn't a worse base to use in the first place and probably was even better for determining overall competitiveness; there's also the fact that he tried to model future events with only current info, primaries have candidates drop out, explode, etc. the general is set in stone in this manner, Nate couldn't model where support went as candidates fell out because it was unpredictable (as shown in Pew's later polling), he saw Herman Cain, Rick Perry and Newt's implosions in 2012 as something to "correct" for rather than recognizing that they were "accurate" as of the time, but Perry and Cain both disqualified themselves before Iowa and weren't something to worry about (they only were slightly overestimated in late game polls)
had he stuck solely to the polling he would have been right about Trump and Cruz
there is so much polling now that his model both fluctuates daily because of it but stays with the same overall "answer" as the outliers are both factored in by the "now-cast" and filtered out by the longer term forecast, there's no point in adjusting for a few outlier polls because you know they'll be corrected for
this is the same reason RCP, which does nothing but run a straight average of only the latest polls (538 keeps old polls in to smooth out the hiccups) gets the same effect in their average, Hillary bumped up two points due to the polls and now she's actually lower than she was before those outlier polls came out because they moved into and out of the model along with tighter late election polls, her position on the RCP electoral map barely changed because that's controlled solely by the state polls
i think that too many people don't realize what Nate's punditry really is, it's based off the same stuff he did for Baseball Prospectus but with politics, he's not coming from the same political punditry world as everyone else...in that regard he's no worse than any other experts he just lays out the entrails to be criticized and tries to explain when he's wrong rather than just pretending it never happened...he used to get in the same fights and make the same type of calls regarding sports (and still does actually)
Nate's model is more conservative than everyone elses, but they're all saying the exact same thing 95% of the time, especially his and Wang who is somehow his rival?...and both of them rarely disagree with RCP, which again just does a straight average of the newest polls with no weights:
Nate: 306-232, 50-50 D
Wang: 321-217, 50-50 D
RCP: 304-234; 51-49 R
RCP would be in agreement on the Senate except they count even a 0.1 lead as a win in their no toss-ups method. (Heck is at +0.3 in NV.)