Sunday, November 26, 2006

Polls or Tradesports? Which is more accurate?

I came across this article on Kausfiles at Slate this morning and the author does a modest analyzation of the 4 major polling firms (Rasmussen, Mason-Dixon, Survey USA, Zogby) vs the results of the actual 2006 results. The order of finish according to the spread or difference of the final polls were as follows: Mason-Dixon, Rasmussen, Survey USA, Zogby. Read the article for more info. Now I know in 2004, Tradesports predicted every race except one according to a Dartmouth study if I am not mistaken; but does anyone know how TS did this year? I know for example on the GOP Senate control (trading in the 60-70s) lost big while if you looked at the state by state contracts for Senate the Democrats were actually winning. I also think off of the top of my head TS was wrong on the gubernatorial election in Minnesota where Pawlenty pulled it out over Hatch (the TS contracts were pretty heavily tilted in Hatch's favor, but lack of liquidity could be the culprit). I wish I had more time to fully analyze the results but let me know what you found to be the case. While I believe polls tend to give the general "temperature" per se of the race (whether the race is close or not), I still believe TS is the better indicator of not only the closeness of the race but who wins as long as there is great enough liquidity.

No comments:

Post a Comment