If Romney is unable to rebound...

[ShadowRocket]

How likely is it that Obama could match or excede his '08 margin of victory in the popular and electoral votes?

[sobo]

Beating 2008 in EV would be very difficult. MO is winnable, but taking IN again based on the recent polls would be very tough, and Obama really needs both to match or exceed 2008 because of the reapportionment in 2010 unless he can somehow pick off an extra state. Even if he picks up MO and MT, but loses IN, Obama would only be at 360.

[J. J.]

Ah, both tracking polls have the candidates tied. 

[Skill and Chance]

Ah, both tracking polls have the candidates tied. 

They are clear outliers and one doesn't call cell phones.

[Vosem]

Romney would have to sink further, and that would be really hard. Even at the level he's going now, he'll beat McCain. Not by much, but he will.

[J. J.]

Ah, both tracking polls have the candidates tied. 

They are clear outliers and one doesn't call cell phones.

Two separate tracking polls?  

The odd that both are outside of the MOE is about 400 to 1.  The odds that both are showing the same error is about 2000 to 1.

You must be desperate to make that claim.

[Skill and Chance]

Ah, both tracking polls have the candidates tied. 

They are clear outliers and one doesn't call cell phones.

Two separate tracking polls?  

The odd that both are outside of the MOE is about 400 to 1.  The odds that both are showing the same error is about 2000 to 1.

You must be desperate to make that claim.

What is the probability that all of the state polls, national media polls, and the RAND and Reuters/Ipsos trackers are all off in the same direction and these two are right?  It's vanishingly lower.  As Harry Enten says, if it were named anything other than Gallup we would all be ignoring it right now. 

Furthermore, even if they were right, a challenger needs to do more than tie.  Kerry led most of the summer in 2004.  So did Dewey.  Reagan was into the 50's in Gallup during his convention bump.

[Fmr President & Senator Polnut]

Ah, both tracking polls have the candidates tied.  

They are clear outliers and one doesn't call cell phones.

Two separate tracking polls?  

The odd that both are outside of the MOE is about 400 to 1.  The odds that both are showing the same error is about 2000 to 1.

You must be desperate to make that claim.

My question is, forgetting the "rules", how often do the standard polls deviate from the trackers... I don't recall 2004/8 having such contrary finding.  

EDIT: I did some quick reading... at this time in 2008 all polls seemed to be lining up relatively closely... even the trackers.