Prediction Markets and the Palin Pick

Palin is McCain’s VP pick, but the inTrade markets seem to have been caught flat footed. Here is the graph of her contract’s performance over the life of the speculation:


Notice that she never seems to have traded at above 20 and immediately before the announcement she was trading in the single digits. I’ve had an on-going discussion with some friends about the value of inTrade numbers, and the anti-inTrade voices have been crowing this afternoon that Palin’s pick shows the bankruptcy of prediction markets. “I argue against putting too much weight on prediction markets,” one friend said, “which I think are interesting but not of tremendous political value or somehow more predictive that good polling data.”

So does this show that the inTrade is bunk and we ought to rely on expertise?

It’s hard to say. I can think of a couple of different stories that one might tell here. First, you could — as I think my friend does — assume that inTrade numbers are driven by essentially uninformed speculators responding to momentary fluctuations in the flow of information from the 24-hour news cycle. Better to ignore the noise and go with thoughtful experts who base their analysis on reasoned elaboration. The failure of inTrade to pick Palin is just a case in point.

There, is however, another story that we could tell. “Prediction market” is actually a misnomer. inTrade is not a predictor but rather an aggregator of information. If we assume that self-interested investors are not throwing their money away, then the markets should respond to the available information and provide us with a rough sense of the probability of particular outcomes given that information. Note, if we suppose that these markets will accurately gauge probabilities it doesn’t follow that their predictions always come true. After all, sometimes improbable events happen. Also, the markets are going to at best reflect available information and sometimes key bits of information just aren’t available.

Accordingly, it seems to me that we have three possible ways of evaluating the inTrade failure to pick Palin. First, inTrade may simply be bunk. Second, Palin’s choice may simply have been a very improbable event that happened to come to pass. Third, it may be that in the period running up to the choice of Palin there just wasn’t all that much relevant information available about the relative probabilities of who was being chosen.

If you think that the first answer is correct, then you ought get an inTrade account. There is good money to be made trading against the ignorant, CNN-driven masses. If you think that the second answer is correct, you should take comfort from the fact that fairy tales do indeed come true sometimes. If you think that the third answer is correct, then you ought to realize that the torrent of expert puditry is an illustration of the iron law of gaseous knowledge, namely that any amount of knowledge or information — no matter how small — can be puffed and expanded to fill any space of ignorance — no matter how large.

You may also like...

4 Responses

  1. Dave says:

    Perhaps it says nothing at all. My sense is that prediction markets do well at establishing probabilities for future events subject to either market or competitive forces (such as elections, sporting events, or other markets), and do not so good on what is ultimately a single person’s decision, and, as in the McCain VP pick, a decision where the decisionmaker is trying to surprise everyone with his maverick pick so as to take advantage of the news cycle.

    I wouldn’t read to much into this one. You might as well ask the prediction markets to determine which number between 1 and 10 I am thinking of right now.

  2. A.J. Sutter says:

    “There is good money to be made trading against the ignorant, CNN-driven masses:” but not the WSJ-driven masses. Palin’s name appeared in a WSJ list of contenders several weeks ago. From that point, choosing her seemed like an obvious “move to block”, to use a Hollywood Squares metaphor that seems to fit our electoral process all too well.

  3. Ocean says:

    I must agree with Dave. Since prediction markets aggregate publicly available information, they’re useful for foreseeing winners of elections, or other events involving many different people. But to determine what one person (or at best a small group of people behind closed doors) is deciding would probably require insider information. And, apart from being illegal, that isn’t what prediction markets are based on.

  4. Miriam Cherry says:

    Dave is 100% right on this one. People said the same thing when prediction markets failed on Harriet Miers. When it’s up to an individual, the markets don’t really work.

    Prediction markets are good when based on knowledge or perhaps what a crowd will do, all based on publicly available information. They are not mind-reading devices, and only McCain (or McCain & staff) was privy to that info.