For the poll obsessed here, understanding how different odds makers are coming up with their projections can be difficult. My own model is pretty complicated, and I run about 20,000 election simulations to create it.
But the basic idea is pretty simple: you calculate the odds of winning a state based on the polling, and then you run a bunch of simulations to see what the probably result is. Now there are some flaws in this more basic approach (eg dependent versus independent variables) but the basic principle is simple.
So I created a more simplified model that you can use. If you follow the link below, you can enter your odds for each state, and generate the probability of an Obama victory. Because of limitations with Google docs, this runs only 1,650 or so simulations, but the basic idea is pretty straightforward (and it took only an hour to create).
If you find an bugs send me a message here or at on twitter at dcg1114.
Before anyone asks - I am part of the legal protection team here and will be working over the next week.