Thursday, June 28, 2012

InTrade and the Decision

Dave Weigel tweets:
Fun fact: InTrade put "mandate will be overturned" at 73% yesterday. Another fun fact: InTrade is useless.
Disagree! InTrade isn't useless; it's just that you have to use it for what it's good for, which is putting a number on conventional and insider thinking. That's quite useful, indeed; it just isn't the same thing as saying that InTrade will always be good at predicting the future. For that, you need to know whether conventional wisdom and insider thinking is likely to be correct or not.

So, for example: during the GOP nomination battle, there was a ton of hype at one point about Herman Cain, and you could find plenty of examples of people saying that Cain really had a significant chance of winning. But if you look at InTrade, you find that at least according to those in that market, it was all phony hype. As opposed to the Newt Gingrich bubble, which really did involve (foolish, but still) people thinking he really might win.

Now, how accurate is InTrade at judging conventional and insider wisdom? Obviously, that's impossible to know, but I think it holds up pretty well. In the Supreme Court example, I think in fact that the bulk of opinion from oral arguments on was that the mandate was toast. So InTrade did what it does; it's just that even the wise guys were wrong on this one (collectively, that is).

So how do you use it? Exactly as you would use any non-numeric indication of what insiders are thinking. If there's good reason to think that insiders know a lot, then what they predict is apt to be a good prediction. If not, not.

It's a tool. As with all tools, be careful to use it only for what it's design allows it to do.


  1. Why is it useful to put a number on insider and conventional thinking? Is "most insiders think" or "The conventional wisdom is....." too many keystrokes?

    1. Numbers just make everything better. There's a big difference between a 60% score on InTrade and a 90% score on InTrade, and being able to say that that difference is an extra 30% gives you something nice and concrete to work with. Numbers can be averaged and statistically analyzed, whereas "most people think" is much harder to work with rigorously.

      Of course, it's very easy to get hung up on false precision. A change from 59% to 62% isn't really worth all that much considering the inherent noise of InTrade. But again, that's something you can describe mathematically by talking about standard deviations and stuff, so numbers win again.

    2. Numbers can be averaged and statistically analyzed.....

      I'm not trying to be contrarian about this, but I'm seriously curious. What statistical operations use InTrade numbers? It's a betting pool. It can't be averaged with random-sample polls or anything else I can think of other than other betting pools, if there are any comparable to it. And even then, what would we know? The average bets people are placing? What analysis would gain from knowing that numerically?

    3. Jeff,
      there are people who go around predicting seat gains and election results using intrade.

      I happen to not be a fan, because there's very little theoretical content there, and I happen to dislike the theory it's based on (efficient market hypothesis); or, more specifically, the theory is fine, it just rarely is applicable.

    4. Matt, thanks -- I did not know anyone was doing that. It does still puzzle me, because I thought the InTrade number WAS that kind of prediction, not that it was "used" in arriving at some other one. But OK, whatever keeps the quants busy and happy, fine by me.

  2. I saw intrade as high as 82 yesterday evening.

  3. InTrade is a market; markets are often wrong, but why in the world does that invalidate them, Dave Weigel?

    Poor battered Research in Motion reported its latest disastrous earnings a few minutes ago after hours; though the stock closed today at ~$9/share, off 73% from its 52-week high, its currently down an agonizing additional 19% in afterhours trading.

    If you - and I'm sorry if this is true - were still clinging to RIMM through today's close, your hopes of a revival have been dashed yet again. What in the world would that have to do with the efficiency or effectiveness of the market? You were just wrong - so what?

    A similar thing happened today with intrade and the ACA SCOTUS decision. I actually feel exactly the opposite of Weigel: if the market can't be wrong - even badly so - about something like the ACA decision, what's the point of having a market? Without uncertainty, how would anyone ever trade?

  4. So Dave Weigel doesn't know the difference between 73% and 100%? Sounds pretty useless to me.

  5. You've all seen the argument that Scalia's dissent refers to the eventual majority as "the dissent"; apparently highly unusual in a actual dissenting opinion, and suggestive of the notion that some entity made Roberts an Offer He Couldn't Refuse.

    If the implication is correct, then the mandate stood on the basis of a highly unusual, low-probability turn of events. Intrade thus placed a 27% probability, the night before, on an occurrence that turned out to be, in fact, a low probability event.

    For this Weigel would throw out Intrade? Aside to Anastasios wrt 'smart' people: Weigel's a Northwestern grad. FWIW.

    1. To put it in perspective, suppose InTrade made a market for "Giants pitcher throws perfect game in June". Those options would have traded for $0.01 or so a share for the first two weeks of the month, and then the morning after Cain's perfecto, Weigel might have taken to the Twittersphere to declare how useless Intrade is.

    2. The dissent refers to Ginsburg's partial concurrence/partial dissent as a dissent, not Robert's majority opinion.


Note: Only a member of this blog may post a comment.

Who links to my website?