3%
I've been doing data entry of October and November polls in my spare time in order to run the simulations I would have run on the early morning of November 2nd if I hadn't been working. I only have a little left to go (the remaining amount shouldn't change the projections much), but it seems fairly clear that I would have called the election correctly. The popular vote is almost exactly right, and I called each state right except for Wisconsin and one of the Maine electoral votes. Furthermore, I would have called nearly every state with a high degree of confidence. I would have give Kerry roughly a 3% chance of winning. I'll post later with more exact results, a refreshed list of the swingiest states, and some maps.
Read more (maybe)!