05/21/08 I added conditional probabilities to the main table, computed assuming mccain wins pivotal states ohio or florida. Clinton has a good chance, but it looks like Mccain should concentrate his efforts in Florida

05/20/08 I have added the electoral votes distribution chart to the main homepage display. I will explain what it is when I find some time :-)

05/18/08 what really hurt obama are Florida and Ohio. In today's polls, Obama is losing 43-44 vs McCain in Florida, which implies a 35% probability of winning the state. Florida receives 27 electoral votes, therefore on average he's getting approximately 18 electoral votes less than Clinton that is predicted to win the state. In Ohio he's losing 42-43 in the polls, with a similar outcome. Ohio has 20 votes, therefore he can only add about 7 votes, while clinton is getting all 20.

If I was a superdelegate I would commission a very detailed survey of voters in those two states.

05/08/08 I checked some of the days where Clinton's probability of winning is zero in the chart. Those zeros seem to make sense.

I compared April 12 (probability of winning: zero) with April 13 (probability of winning: 17%). The mean number of electoral votes to Clinton are, respectively, 215 and 268 on those two dates (remember that 270 votes are needed to win). There were lots of new polls on April 13, with big changes in Missouri (11 electoral votes), Minnesota (10), New York (31), and most importantly Ohio, where Clinton goes from 42-47 vs Mcain on April 12 (effectively losing the state) to 52-43 (which is tantamount to winning the state with probability close to one). Ohio alone provides 20 electoral votes. Needless to say, pivotal states are important ... I will soon put up some conditional probabilities, conditioning on McCain winning some of the pivotal states.

05/07/08 the history charts are back online showing huge swings in probabilities. I'll check later if they make sense. If anyone wants to look at the data they are based upon, the urls are in this format

../var/includes/Storage/rawdata-[mm]-[dd].html

where mm is the month number and dd is the 2-digit with leading zero day number

05/07/08 i have found an error in the code, and had to erase all historic data. I will try to reconstruct them later. Stay tuned.

04/28/08 i managed to recover predictions from past dates and to reconstruct the trends. Now all pictures are in the front page underneath the main table

04/27/08 one day online and first big swing already. Not sure which poll changed since yesterday, but clinton's probability increased considerably. I have added the "basic computation" that adds up electoral votes by just counting a win in the poll as if it was a win in the election in that state (if there is a tie, I split the college)