Comments on: Damped polls outperform prediction markets. http://www.midasoracle.org/2009/03/07/dumped-polls-outperform-the-prediction-markets/ Prediction Markets, etc. Thu, 02 Feb 2012 23:13:18 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Chris F. Masse http://www.midasoracle.org/2009/03/07/dumped-polls-outperform-the-prediction-markets/#comment-23691 Chris F. Masse Mon, 09 Mar 2009 08:26:54 +0000 http://www.midasoracle.org/?p=13174#comment-23691 David Pennock: Sorry for the typo. David Pennock: Sorry for the typo.

]]> By: Chris F. Masse http://www.midasoracle.org/2009/03/07/dumped-polls-outperform-the-prediction-markets/#comment-23690 Chris F. Masse Mon, 09 Mar 2009 08:25:54 +0000 http://www.midasoracle.org/?p=13174#comment-23690 Panos Ipeirotis: Yes, that is what Justin Wolfers wrote in a comment on a blog post from the Andrew Gelman blog, which I already linked to months ago. Panos Ipeirotis: Yes, that is what Justin Wolfers wrote in a comment on a blog post from the Andrew Gelman blog, which I already linked to months ago.

]]> By: Panos Ipeirotis http://www.midasoracle.org/2009/03/07/dumped-polls-outperform-the-prediction-markets/#comment-23685 Panos Ipeirotis Sun, 08 Mar 2009 23:48:59 +0000 http://www.midasoracle.org/?p=13174#comment-23685 Interestingly enough, if this paper is correct, then this suggests an obvious trading strategy for the traders in future election markets. So, in a sense, the markets would incorporate the information in this article and produce more accurate results next time :-) Now, being critical: Being accurate in out-of-sample data just 4/5 times is not really impressive. Is someone wanted to be really a PITA, the result does not even suggest statistical significance using a sign test, as there is 18% probability that the result is due to pure chance. Now, being more realistic: they did not have that many elections to work with. Interestingly enough, if this paper is correct, then this suggests an obvious trading strategy for the traders in future election markets. So, in a sense, the markets would incorporate the information in this article and produce more accurate results next time :-)

Now, being critical: Being accurate in out-of-sample data just 4/5 times is not really impressive. Is someone wanted to be really a PITA, the result does not even suggest statistical significance using a sign test, as there is 18% probability that the result is due to pure chance. Now, being more realistic: they did not have that many elections to work with.

]]> By: David Pennock http://www.midasoracle.org/2009/03/07/dumped-polls-outperform-the-prediction-markets/#comment-23684 David Pennock Sun, 08 Mar 2009 15:31:05 +0000 http://www.midasoracle.org/?p=13174#comment-23684 In the title, do you mean "Damped polls"? In the title, do you mean “Damped polls”?

]]>