Since we've had a mini-debate about quants vs. simple averages in polls I'll tell you this at this point in the race in terms of how close it will be so you can judge the performance of more sophisticated analysis at a later date.
The Big 4 regression guys in this election (Silver, Linzer, Wang, and Jackman) have all converged on 332 as the most likely number of electoral votes for Obama. The second most likely is 303 electoral votes. The third most likely is 347. I suspect all of those numbers reflect a race that is not as close as you believe (personally I suspect 303 is the right number) and it will be interesting to get your take on this after we get the final result.
I think they are counting on numbers that will not hold up today.