First of all, thank you for your continued work on the IEM. [MicroSoft Word file] It’s great to see a continued and consistent trader base and methodology.

But I do have to admit that I am very uncomfortable with the way the paper compares the Vote Share market and the polls to the final Election Day result. Both a prediction market and a poll are trying to put a number on uncertainty, and that uncertainty is only ever known/fixed on Election Day. While the dynamics of how each reacts certainly can be compared and assessed, measuring an error based on comparing 2007 results to Election Day results strikes me as really unreasonable.

While I do find it fascinating that the markets bounced around the final result, leading to a potential interpretation that there was a structural reason why the Democrats won in 2008, I disagree with using prediction market results to imply that the result was pre-ordained.

I believe that prediction markets should be compared to polls through the election cycle to determine how each moves relative to each other (for communication and information flow). I believe that prediction markets and polls can only be assessed based on their results shortly before Election Day (for accuracy), and that as you get further before that the numbers become meaningless.

All that said, I’m willing to change my mind if there’s enough evidence otherwise.

Jed Christiansen

-

- Iowa Electronic Markets propaganda

- HubDub propaganda

- InTrade propanganda – Bis.

- BetFair propaganda

-

Yes, the infamous Judith Miller.