Foreign Policy’s blog posted this image and used it as a launch pad to discuss how, immediately before Alberto Gonzales’ resignation was announced, the prediction market upped the odds that his resigation was coming.
I can’t help but look at the larger trend, though, which appears to indicate that the market consistently lowered the odds that Gonzales would not make it past September over time, right up until immediately before the actual resignation was tendered. It sure looks to me like the prediction market did a pretty poor job in this case, since Gonzales did resign before the end of September.
September 7, 2007 at 2:35 am
Bear in mind that you’re dealing with probabilities and one-off predictions here. Whether one particular bet went right or wrong doesn’t tell you much useful at all. A more useful question would be: how do prediction markets perform as a whole (or perhaps, some rule-based selection of prediction markets based on information that’s knowable before the event) and how do they compare to other methods of prediction?