First is the same number shown on the xkcd page, the fraction of elections won by each major-party candidate in a 2-man race according to the state-by-state predictions. Unlike xkcd, I show two decimal places and go into a special exponential mode for numbers very close to 0% or 100%. (I don't know whether this will really be needed, but what the heck)
Second is the number of EV that are within X% of going for a given candidate. The numbers on the "50%" line should total to 538, while in the "20%" line only states within 80..100% for one candidate (that is, a 20% chance of being lost) will be shown in that candidate's column. Those numbers will go down to .1% but only show rows that aren't zeros.
Because this program uses intrade's xml data interface, each run takes only about 4 seconds to do 3 requests (limit 50 "stocks" per request, but 102 stats are requested). xkcd's program used a much slower scheme that included running perl and lynx, then parsing the html with regular expressions. (I never thought I'd say this, but Shame on you, Randall Munroe)
PS I don't have a position on whether the intrade predictions are actually predictive of this election.
Files currently attached to this page:
Entry first conceived on 4 November 2008, 0:23 UTC, last modified on 15 January 2012, 3:46 UTC
Website Copyright © 2004-2017 Jeff Epler