The problem with all of the polls is that no one knows who is going to turn out for this off-year, gubernatorial election. Pollsters are using likely voter models that include the presidential elections in 2008 and 2012. To me, a better likely voter model would be who voted in the 2009 gubernatorial election?
My suspicion is that a lot of the Northern Virginia, young, hip, Obama voters are not going to show up for this election. This is going to be all about turnout.
You are right. I am not sure I fully trust the 2009 models but it will be much closer than 08 or 12. Everyone here is sick and tired of both candidates, so low voter turnout is expected and whoever gets their base out wins.
We have a 3rd party Libertarian (who is as much a Libertarian as Obama) that is complicating matters, but his support is soft and only been measured at 10% at the highest.