A look into the past
I'm on record as saying the polls currently in the RCP average are crap. I think they are sampled with turnout models that are highly unlikely. However, let's keep in mind the historical perspective.
On October 5, 2008 Obama had a 5.9% average advantage over McCain 49.3 - 43.4.
On October 5, 2012 Obama has a 3.2% average advantage over Romney 49.3 - 46.1.
October 5th was after the first debate in 2008, where Obama was widely regarded as the winner. It was also after McCain suspended his campaign, due to the economic meltdown.
Note the interesting fact that Obama's support in both averages is identical. When election day rolled around the Independents broke hard toward Obama and the Republicans stayed home. This time, Independents will break for Romney and Republicans will turn out in very high numbers.
Even if the current RCP average is accurate (it isn't), this trend compared to 2008 is not a good sign for Obama.