hile prediction markets have been in the spotlight this year, they are still unfamiliar to many folks. As one small step towards improving their visibility, along with my colleague James Lemieux I ran a prediction market at the University of Kansas School of Business. The markets ran for three and a half months and almost all traders were undergraduate business majors (you can see the very end stages of the market at: http://kufin400.inklingmarkets.com, username: myfoxkc and password: myfoxkc).
These markets were quite popular. The 475 traders made over 27,000 transactions in the 139 available markets. As a matter of reference, that is about 200 transactions per market while in Google’-s market this ratio is 260.
There was a mix of both socially redeeming topics (issues of interest to the Business School such as how many internships the undergrads would get this school year) and others designed to attract interest (politics, sports, entertainment, finance). I was surprised to see that passions–- and trade volume–- ran quite high even in the more serious markets. For example, one contract’-s expiry was based on whether the XM-Sirius merger would be consummated by March. When the DOJ announced its approval at the end of that month, there was only a small price increase. As the comments below suggest, this was not because the traders were asleep at the wheel but rather because they had a good understanding of the regulatory environment.
Inkling Markets provided the platform for our markets (if you are unfamiliar with Inkling, they have active public markets which you can sample). Inkling’-s software and support is really ideal for classroom markets. There are nice features for both the people running the markets (James and I) as well as for traders (the students).
For the market admin:
– it is a snap to set-up and administer new contracts
– Adam Siegel and Nate Kontny are very responsive to questions, often responding within the hour
– an intuitive trade interface, which is accessible even for those without experience with financial markets (though this can be a drawback if you would also like students to become familiar with order books)
– lots of goodies (customizable profile pages, market-specific discussion boards, graphs) leads students to visit the market a lot
– the daily/weekly top traders list encourages participation
I would strongly recommend others give prediction markets in the classroom a try. I found them to be both a great pedagogical tool and also one which the students really, really like. Students learned first hand about the role of information discovery as well as the biases often seen in prediction markets (though I will add it was difficult to illustrate the home town bias given the success of the athletic teams at my school this year). Feel free to get in touch with me if you have questions about how to set-up your own classroom markets.
Inkling Markets is a great company, and Adam Siegel is great.
At one point, I was hoping that play-money InTrade let people create their own event derivatives, but it never materialized. Maybe later…
Pretty cool success! Congratulations Koleman.
Just a heads up on making comparisons between stats from MSR markets (Inkling) and CDA markets (internal Google): Don’t compare the number of transactions. It isn’t an apples-to-apples comparison.
In MSR, every buy or sell order is automatically matched by the market maker. Each time someone submits an order, that’s a new transaction. In the CDA market, some orders to buy or sell are never matched by other players — typically because of disagreements over the proper price. These unmatched orders do not count as “transactions” even though users have been actively participating.
Yes, it’s “cool”.
Good point, Bo. I never thought of that point. Sounds true. Mike, Jason, any comment?
Thanks for the props–I have a much better appreciation of the difficulties of running a PM now (though I am sure your job was far tougher since all my traders were in one building!)
Good point on CDA/MSR, but I still think the comparison is appropriate. When someone makes a trade in a MSR, they are willing to trade at the prevailing price. There are likely to be many others who do not trade, but would be willing to do so if the price were quite different. These latter traders are much like those who place limit orders in a CDA, and just like with the CDA they do not get included in the transaction tally.
Regardless of the microstructure, the number of transactions is a reasonable measure of the market activity.
@Chris F. Masse: Chris, take a look at this previous post:
I agree with Koleman that the number of transactions is a reasonable measure of market activity and with Bo that a direct comparison of number of transactions in CDA and MSR isn’t an apples-to-apples comparison. But it is only a very generic indicator, and for more understanding we’d probably want to look at number of traders in each market, size of transactions, and so on.
I’m not sure the issue is important enough to warrant deeper study, but such a study could readily be done experimentally. Alternatively, a comparison of trading in relatively similar markets at, say, Inkling and NewsFutures might show how number of trades per market is associated with market efficiency in MSR and CDA markets. But I can imagine a lot of problems in getting a clean comparison in the latter approach; an econ lab experiment could do the trick.
@Michael Giberson: Excellent synthesis and excellent suggestion. Well… still waiting for David Pennock’s input…
I don’t see why someone would invent a concept like ”market activity”, which tells you nothing, if there is a definition available at all, then try to measure it using different methods.
We have a valid concept, it’s called market liquidity.
Tbh, I don’t see what this page is about. I was hoping it would be about reliability of predictions markets, but that turned out to be wishful thinking, I think.
Chris asked me to comment, so here goes: I agree with Michael and Bo and Koleman.
One way to think of the market maker is as if it is just another trader, albeit an extremely active one. Thus the comparison of number of transactions does have some meaning, if not a perfect correspondence.
@David Pennock: Good synthesis. Let’s see whether the people above agree.
[…] Surprisingly, Koleman Strumpf’s story ranks #5. Not that I’m […]