It's a little bogus to back out the 1-day tracking results from Rasmussen's posted 3-day tracking poll, but what the hell, I do it anyway. Yesterday's polling had Kerry up 49/46.
I think the 3-day tracking is N=1000, so I infer than the 1-day figures should have MOE=5.5% or so, which is high, which is why Rasmussen doesn't post these numbers in the first place.
Update [2004-9-17 12:33:25 by globecanvas]: Correction, the 3-day tracking is N=3000, so the 1-day figures are perfectly legitimate, in fact they have a lower MOE than Gallup's N=760 poll released to much fanfare today. In fact you could think of this graph as a vivid illustration of exactly what MOE=3.1% means.