Risk Markets And Politics

Wednesday, October 15, 2008

The gamble of downplaying manipulation

Whether it is GOP bias, manipulation, or simply confident well-funded traders, there is some agreement that the Intrade presidential markets have been affected by "non-informational" trading. To be clear, this is not a condemnation of Intrade. The exchange's liquidity and trader diversity are hamstrung by archaic laws in the U.S., the continuation of which will frustrate a fair assessment of market accuracy. The point is that arguing for legal and regulatory change while downplaying the viability of manipulation and other market pathologies is counterproductive.

That prediction markets may be manipulated with some persistence should be no surprise to anyone who has followed the subject in the past couple of years. Here is a sampling of some of the warnings:

The HRC attack, part 2
The Giuliani manipulator buyer is back.
Manipulation can affect prices.
Is there manipulation in the Hillary Clinton Intrade market?
Is there manipulation in the Hillary Clinton Intrade market? Redux
Measured Enthusiasm For Prediction Markets

We now even find some academic papers that admit that manipulative trading may be profitable given certain assumptions. It is up to readers to decide which papers contain the most "stylized" assumptions.

No-one argues whether, in the long run, in general, manipulation is a losing proposition that subsidizes other traders — but is it really prudent to deploy that message, in comments to CFTC for example?

First, if Obama wins the election, based on the other available markets and poll projections, it would seem that an error had been introduced into the largest and most widely-cited of prediction markets. When comparing market and poll accuracy over time we are usually talking about only a few percentage points difference, so this error isn't trivial. Furthermore this is a market that takes place only once every four years, so long-run arguments ring a little false. There's no reason why something similar couldn't happen in 2012. At least, one is optimistic that the regulatory situation will improve and Intrade's traders will be more numerous and less capital-constrained at that time, which should make manipulation more difficult on average. Those who downplay the dangers of manipulation risk such goals by sacrificing their general credibility. It's a negative skew proposition.

Second, some markets can irreversibly affect the outcome they predict. This happens infrequently and requires some fundamental basis, but specific cases can spectacularly undermine a general argument. This is the old bit about trying to cross a river that's three feet deep on average. An example we've seen recently: when a business is predicated on maintaining a deposit base or borrowing short-term at certain rates, manipulation might be irreversible if it targets confidence or attacks the business's funding costs. In essence, the manipulator forces the (possibly quite liquid) market to "settle" as the firm approaches insolvency, and prices do not snap back. Breaking a currency peg has a similar dynamic. Now, there is currently no real analog to these situations in prediction markets as such, but either these markets will continue to be relatively small and not widely-followed, or ....

Kenneth Arrow and Intrade CEO John Delaney are making the right arguments here: transparency in the form of more public markets, along with less concentrated risk, would have helped avoid this crisis. But don't try to sweep uncomfortable subjects under the rug. That won't end well.


  • Nice post.

    It seems to me that a major problem is the lack of usability on the other exchanges for the US market. Betfair does 10x the volume as Intrade but no one ever quotes it as (a) it is hard to find the market and (b) it is impossible for 99.9% of the US audience to understand.

    Bringing pricing transparency between these separate markets is something that could happen without regulatory changes.

    By Blogger nigeleccles, at 9:00 AM  

  • Thanks Nigel, Yes, people are working on making markets more usable. I know this is one of Jason Trost's main points, but having a public market where people can make bids and offers, even if it is difficult to understand, is the first step.

    Americans don't feel welcome at Betfair for a number of reasons including that fact that political markets are hidden between pelota and snooker. Even among non-Americans, the exchanges appeal to different audiences and that's probably the biggest reason for lack of arbitrage between the two.

    Also I think the 10x number is way off for political markets. When you convert the Betfair matched amounts to Intrade accounting logic (sum all of the risk for each transaction, not just the "backer's stake"), I get $13.6m Betfair vs. $9.9m Intrade in the Obama market and $9.1m Betfair vs. $10m Intrade for McCain.

    By Blogger Jason Ruspini, at 5:58 PM  

  • I agree with your points around Betfair. The amazing thing about Betfair's interface is that it hasn't changed since 2002. I know because I helped design the interface that inspired it, the flutter.com racing exchange interface which we launched in the summer of 2001. (You'll also notice that the BETDAQ interface is almost identical to the Betfair interface. That isn't a coincidence as I also designed that and launched it in early 2003!)

    The problem is that since 2002 Betfair's design seems to have frozen. It was a good design back then but now it looks aged and doesn't take into account the advances in web design that have occurred in the past 5 years.

    On the numbers I have just under £5m traded on Obama on Betfair (I think that is the sum of the backer's stakes) and just under 1 million contracts traded on Intrade. Do you get the same?

    By Blogger nigeleccles, at 4:54 AM  

  • http://uk.site.sports.betfair.com/betting/LoadRunnerInfoAction.do?marketId=20739353&selectionId=1171627&rfr=3925

    I may be wrong, but I think the numbers in this link are only the backers' stakes. If that's true, to get a fair compare with Intrade, for each traded price the total exposure is:

    "Traded" + ("Price" - 1) * "Traded"

    So if it says that $11607 traded at 3.0, the backers' exposure is $11607 and the layers' exposure is 2 * $11607. I got $13.6m by summing those two risks for every price that traded. Then you have a fair comp to the number of Intrade contracts traded * $10, which was $10m.

    By Blogger Jason Ruspini, at 5:43 PM  

  • Thanks for the clarification. I believe they are backer's stakes.

    I was counting each Intrade number as $1 instead of $10. My mistake.

    By Blogger nigeleccles, at 12:07 PM  

Post a Comment

<< Home