Honestly.
They specifically say in their analysis that Taylor isn't on the ballot, yet they don't have a Roberts vs. Orman average.
That's probably because RCP has always been a Republican-leaning publication. Almost half of all articles they're linking to are science-denying far right propaganda articles. Not to talk about Sean Trende, who happens to seriously believe that a Republican presidential candidate would easily win even if every single Republican in the US were obliterated from the face of the earth. That's how serious RCP is.
Trende isn't always super optimistic about republicans. He wrote this in april:
http://www.realclearpolitics.com/articles/2014/04/17/could_democrats_gain_senate_seats_this_fall_122301.html--------------
I use RCP to help in gathering polling data, but I don't usually take their averages verbatim when it comes to my ratings. I'll eliminate things like yougov, outliers, quasi-internals, etc. and take what's left, sometimes adding in other polls that I know about but aren't included in the present average for whatever reason.
