The idea behind the design of information markets such as Intrade and a few others is that they aggregate information from a diverse crowd of people and thereby, attain accuracy by errors on either side of a question canceling out. David Leonhardt therefore writes about the most important policy decision of the last few weeks where Intrade in particular was wrong by a large margin. As the article states, the US Supreme Court made a majority decision that affirmed the constitutional fitness of the health care mandate.
Being one of the largest decisions this year and one on which the incumbent president had lots of skin in the game, it is essential to consider the fact that Intrade in particular was so far off. David Leonhardt mentions that information markets are not perfect and may have been affected in this instance by the thinness of the market and may have reflected a very narrow set of views to the exclusion of the alternative views. It appears that on the part of the Obamacare, the participants reflected too much of those who information suggested that the mandate would be declared unconstitutional.
And yet in spite of the confirmed error in this instance, I think that the chatter about this miss is still uninformed. To my mind, confidence in information markets is based on the fact that the collection of view that generates its information is bound to be more accurate that individuals who hold court as "the experts". Thus information markets are bound to have a better predictive record than these experts but that is not to state that they are infallible. So the total failure in calling out the outcome of the case is still embarrassing to me, but I think that few experts had it as clearly too. Intrade only appears to have made it look like it was not even close because with a 65%-70% threshold, Intrade in particular performs very well.
The lesson of this event is that the experts claiming that information markets are imperfect are not saying anything new. On the other hand, the cheerleaders of information markets must always state clearly that they tend to be far more accurate than experts and that usually is good enough but does not mean they are infallible.
Being one of the largest decisions this year and one on which the incumbent president had lots of skin in the game, it is essential to consider the fact that Intrade in particular was so far off. David Leonhardt mentions that information markets are not perfect and may have been affected in this instance by the thinness of the market and may have reflected a very narrow set of views to the exclusion of the alternative views. It appears that on the part of the Obamacare, the participants reflected too much of those who information suggested that the mandate would be declared unconstitutional.
And yet in spite of the confirmed error in this instance, I think that the chatter about this miss is still uninformed. To my mind, confidence in information markets is based on the fact that the collection of view that generates its information is bound to be more accurate that individuals who hold court as "the experts". Thus information markets are bound to have a better predictive record than these experts but that is not to state that they are infallible. So the total failure in calling out the outcome of the case is still embarrassing to me, but I think that few experts had it as clearly too. Intrade only appears to have made it look like it was not even close because with a 65%-70% threshold, Intrade in particular performs very well.
The lesson of this event is that the experts claiming that information markets are imperfect are not saying anything new. On the other hand, the cheerleaders of information markets must always state clearly that they tend to be far more accurate than experts and that usually is good enough but does not mean they are infallible.
1 comment:
http://rcavanaugh.com/2012/07/how-not-to-interpret-intrade-results/
a/ it's not intrade that is right or wrong
b/ this issue vote was a snowflake, but IF it occurred 4 more times?
Post a Comment