Intrade on the Oscars

How did Intrade do with predicting the Oscar's? Hard to tell, but here are some interesting things to consider before we start looking at prices.

1. Oscar's began at 8:30 p.m. on 2/22/2009.
2. At least as early as 2/18/2009, a blog appeared (and has since been deleted) with "Oscar Leaks." The leaks can be found here.
3. The leaks were very inaccurate, comparing to the realized winners, only 10 of the 24 awards were correctly predicted by the leak list.

Economists, I think, would have been rightly skeptical of a leaked list ex-ante. Why would a person with inside-information reveal this info, why not use it to buy all shares less than 1 for the actual winner and sell shares for the losers at any price above 0? The insider could have easily made a reasonable penny on such a move, and surely someone with inside information would have realized the dollar value of this information in betting. We can consider two possibilities:

1. The list was a ploy to get attention by someone with no inside information.
2. An insider used the list as misinformation, hoping other traders take the bait and allowing the insider to make even more profit.
   

So, before we look over the Intrade data, we must consider that noise, BUT it does not excuse the prediction markets performance if they "miss"*. These possibilities should be discerned by anyone, who traded on the market and discounted the information accordingly. So let's see how the market behaved by looking in-depth at one where the leaked list was right (Best Picture) and one where it was wrong (Actor in Leading Role)

Best Picture - Leaked winner: Slumdog Millionaire; Actual Winner: Slumdog Millionaire


Conclusion: My view of it is that the market did not react much to the leaked message as volume of trading and price level remained pretty constant. Furthermore, the most probable winner was the actual winner. Kudos to Intrade traders!

Best Actor - Leaked Winner: Mickey Rourke; Actual Winner: Sean Penn

Conclusion: Like the leaked list, Intrade suggested that Mickey Rourke was most likely to win best actor, instead of communist Sean Penn (just saying). The most 2nd most probable winner still had about a 30% shot, based on the prices. Again, there seemed to be no response in price to the leaked list.

Best Actress: Actual-Winslet; List - Winslet; Intrade: Winslet (75%)
Best Supporting Actor: Actual - Ledger; List-Ledger; Intrade: Ledger (97%)
Best Supporting Actress: Actual - Penelope Cruz; List-Amy Adams; Intrade: Cruz (61%)
Best Director: Actual - Boyle; List-Boyle; Intrade: Boyle (95%)
Best Animated Feature Film: Actual - Wall-E; List-WallE; Intrade: WallE (95%)
Best Cinematography: Actual - Slumdog; List - Slumdog; Intrade: Slumdog (80%)

In general, I would say that knowing the leaked list was garbage ex-post, we can say that the prediction market accurately discounted its usefulness, as trading did not seem to react to the list. In the 8 event markets it posted, it assigned the highest probability to the eventual winners 7 times, but in these categories even the mostly incorrect list got 6 correct.

* I find it unusual that critics of prediction markets anoint them failure when the highest probability outcome does not reveal to be the true outcome. Prediction markets are not a crystal ball that tells you the future, they just ascribe a probability based on collective beliefs. The markets may have been right in assigning a high probability to a particular event, while still observing that when the event plays out the low-p event occurs.