6 comments on “Intrade versus FiveThirtyEight

  1. I was looking at Betfair’s site at http://www.betfairpredicts.com and noticed they did a decent job vs. 538 also — it looks like they called Indiana correctly when 538 did not prior to election returns. It appears they also got north carolina ahead of the pack. Missouri seems to have been a toss-up late.

  2. Isn’t a lot of this moot because elections are not independent events? Elections which occur at the same time are correlated for a number of reasons, which would show up as bias (either for or against prediction markets), even though none may exist.

  3. Pingback: 2008 US presidential elections | Midas Oracle .ORG

  4. Not moot, but the correlations imply that the data tell us less than what a casual reading might lead one to expect.

  5. Interesting blog entry here. I do not see an email address, but I am a PhD candidate who has been following this 538 versus Intrade debate closely and I was hoping it would be possible for you to share the full set of results that you saved for 538 and Intrade?!? It was be very useful in a project that I am doing on this topic. Thank You!!!

Comments are closed.