[Jack Black]

Hmm, well both Stan Wong and Nate Silver rely on poll averages to drive their models, both of which have proved extremely accurate in predicting actual results of US elections over the last 4 or 5 cycles. (And yes, occasionally they get one wrong).

Nate rates and weights the polls. His ratings are public, and provided in handy "grade" format.

So, I must respectfully disagree that the RCP average is worthless. It correctly predicted Obama / Romney result, but not the margin. ABC News / WaPo were very close, for instance, though.

For people who really hate poll averaging I'd say then confine yourself to looking at the "A" rated polls

I don't get your point about the turnout. Turnout is always a factor in elections, which is why pollsters try hard to factor it into the polls. "Likely voters". Etc.

However if one hates poll averaging on principle I would then say at least use the better polls for thinking about the race, not the outliers or completely new ones, like the USC that all the pro-Trump people are glomming onto.

I have thoughtfully made a list of these on a post yesterday, which you can access here

[Lagmeister]

pro-Trump people are glomming onto.

Well, he is our nominee... but I better understand where you are coming from. I will tell you again... you don't average percents because the outcome is meaningless to determine an actual spread.

use the better polls ... not the outliers or completely new ones

RCP uses the outliers and averages them into the mix to increase the Hillary spread. Furthermore, why flippantly discount new methodology?