2004 Presidential Election Forecast
From Henry Chappell at Univ. of South Carolina. Currently (as of 10/19) predicting Kerry with 282 EVs.
Kerry swing/surprise states: AR, HI, IA, MI, MN, MO, NM, OR, PA, WV, WI.
Bush swing/surprise states: AZ, CO, FL, NH, OH

I built a monte carlo electoral vote simulator using Chappell's state-by-state estimated percentages. The only underlying heroic assumption is that all states are independent (clearly false). Running 5000 simulated votes, Kerry is elected 53.75% of the time, Bush 46.25% (giving to Bush all of the 269-269 ties, which occur 1.1% of the time).

Median result is Kerry 273 EVs. Max landslides are Kerry 396 EVs, Bush 399 EVs.

Adding correlation pushes Kerry's median EVs up [to 282 at 100% correlation], but reduces his win rate [to 52.8% at 100%]. Not surprisingly, the outliers get more extreme.