He didn't say he ignored it.  He said it was balanced out by something else.  I mean, maybe.  There's so much polling everywhere else that maybe one poll doesn't move the needle that much.  I don't know how his model works, exactly.  But I would think that this ought to move the average at least somewhat. 
In fairness to Nate, it's really hard to calibrate a model when you only get data every 2 or 4 years, the data you do get is highly flawed (i.e. polls), and you can't really measure the accuracy of your model except from the national popular vote, which is meaningless.