Are there any studies of the track record of the prediction markets?

The theory of them contains some obvious caveats, or rather "known unknowns" about the participants in the market. An examination of the track record could answer a lot of those questions. And indeed, point more closely to how much credence this prediction is.