Movement in a tracking poll depends on only two things: the new poll data being added and the old poll data being dropped. It is possible, even likely, that Bush had a good day yesterday (the date being added), but he had an even better day four days ago (the date being dropped). If this happened, Kerry would get an uptick in the three day results even though the latest poll favored Bush.
In any event, unless Kerry does at least as well tomorrow as he did on Saturday, Bush will get an uptick tomorrow -- even if Kerry wins the one day sample today. Of course, if Bush wins today's sample, the uptick will be even greater.
All of this illustrates that tracking polls are good to spot trends over a few days or a week, but do not necessarily reflect the state of the race as of the date the rolloing averages are published.
MHO, its garbage.