Idea Futures

Manipulating futures markets

Posted by Peter on October 29, 2004
Posted in: Idea Futures, Investing, U.S. Politics. Tagged: .

Alex Tabarrok writes about the apparent attempts to manipulate the Bush re-elected contract at Tradesports.com (which just dropped to exactly 50!), and CNBC has mentioned the same report today (with a denial from George Soros that he is responsible).

I want to warn people not to treat the failure of this manipulation as strong evidence that manipulation won’t have much effect on the reliability of the prices. If an experienced trader such as Soros tried to engage in this kind of manipulation, he would use a much more patient and cost-effective strategy than quickly driving the price down from 55 to 10.

To estimate the harm done by manipulation, we need to look at careful studies of how accurate markets have been, plus experiments such as the one Robin Hanson arranged. Note also that Robin’s attempt at a theoretical argument on this subject is unconvincing because it unrealistically assumes that traders aren’t risk-averse.

Efficiency of presidential election futures markets

Posted by Peter on October 17, 2004
Posted in: Idea Futures, Investing.

Tyler Cowen and Mike Linksvayer discuss the somewhat confusing reaction of Bush wins futures to the first debate.

I have some different theories about what might have happened. The size of a typical trade seems to be a few hundred dollars or less, so the typical reward for being quick to exploit inefficient prices is probably in the $10 to $30 range. It’s fairly easy for me to imagine that the most sophisticated traders value their time enough that such rewards don’t cause them to react quickly. I’ve been making a small but steady return from fairly regular trades on Tradesports the past few months, but I only log in about once every two days.

Another possibility is that the best informed traders get their information by talking to undecided voters over several days after the debate.

Either way, as an experienced investor it doesn’t surprise me that markets are slow to react to information that isn’t very clear. Markets often show more frequent patterns of prices following a trend than I would expect from random behavior. I interpret this as evidence that some information gets reflected slowly in those prices. That doesn’t mean it’s possible to get rich by any simple trend following rule – enough of those trends are false signals created by traders trying to exploit the naive trend follower that it’s hard to get useful information out of the trends unless you combine it with good information about what the efficient price should be.

Idea Futures markets obsolete?

Posted by Peter on October 14, 2004
Posted in: Idea Futures. Tagged: .

I attended a fairly interesting talk at PARC last week by Bernardo Huberman. He claimed to have created something that sounds like a hybrid between a survey and an idea futures market.
It involves selecting a small group of people (the minimum for it to work well is about 9 people). It includes some market-like trading between participants, but the results are apparently better than what markets would produce under similar conditions, because it does additional calculations such as adjusting for measured risk aversion of each participant, and has something that treats private information (i.e knowledge available to only one participant) differently from public information. At least some of the time Huberman gives participants monetary incentives to get the right answer.
This approach has an advantage over idea futures markets that bad guys are less likely to be able to manipulate it by spending money to produce inefficient prices, because the organization running the system exercises more control over who participates and how much influence each one has. Not that I think idea futures markets suffer any significant problems due to forms of manipulation like this that require the manipulator to transfer unbounded amounts of money to those who are trying to make the market prices accurate, but the difficulty of analyzing the problem has made such markets susceptible to criticism by demagogues. Also, markets tend to foster some secrecy (as traders want to maintain their advantage), while a more survey-like approach makes people less likely to have a stake in having better information than others, since they’re less likely to be involved in future surveys.
Huberman’s approach does have some disadvantages. It presumably requires the person who wants the accurate information to pay for it (with some associated public goods problems in many cases), whereas idea futures markets can often be financed by small commissions on trades. Also, Huberman’s approach probably fails to select the most informed participants in a number of circumstances, thereby failing to extract information that an open market would be able to extract.
A paper covering the most interesting parts of Huberman’s work is available here.
You might also want to see a related criticism of idea futures markets by Manski at http://www.faculty.econ.northwestern.edu/faculty/manski/prediction_markets.pdf, although that paper doesn’t do a very good job of analyzing whether any alternative is better.