It's been a while since I've updated my weighted average of polling on the national polls of the race for president. The swath of polls in October showing Hillary Clinton charging ahead has made it somewhat of a distasteful task. Nevertheless, it needed to be done. Except my results were - well, unexpected.
As a reminder, not all polls are created equally. To account for this I have not included all polls from the RealClearPolitics average of polls in my analysis. I have excluded polls with no reported Margin of Error. Not reporting that makes the numbers dubious at best and unscientific at worst. I have also excluded polls with a margin of error greater than 4.5%, which is a pretty large margin of error. As an added twist, I have excluded polls where multiple instances of the same poll source are occurring within days of itself. For Example the NBC News/Wall Street Journal poll shows up 3 times in the RealClearPolitics average of polls. Yes, the dates are different but it seems like a way of manipulating the average of polls by including your own smaller polls, more frequently.
Given all of that, the race it turns out is a lot closer than the topline optics would indicate, although Hillary Clinton has not surprisingly given the media blitz on Trump, retaken the lead over the last 4 weeks (1.4%). However the lead is razor thin at this point which makes the final debate, the advertising blitz and the get out the vote effort coming up in the last weeks, uber-critical. I expect to see significant poll changes starting next week. Which direction they'll shift is anyone's guess.
For comparison purposes, here's what my unweighted results look like, although I've excluded polls for the same first two criteria I've mentioned above. You can see the gap is larger (4 points).