I would like to comment on the post from the Hatena Diary blog. (By the way, please note that my URL has changed, because I corrected one word in the post title. Sorry for the inconvenience.)
#1. Speculation-oriented prediction markets/exchanges: TradeSports, BetFair.
#2. Hedging-oriented prediction markets/exchanges: HedgeStreet and all the Chicago exchanges that will do binary, European call options.
#3. Forecast-oriented prediction markets/exchanges: Iowa Electronic Markets, AS CLAIMED BY THESE SCHOLARS WHOSE TASK WAS TO CONVINCE THE CFTC TO GRANT THEM A NO-ACTION LETTER. (They would have not gotten it, had they emphasized “-speculation”-. And, of course, “-hedging”- was out of question.) It’-s a “-claim”- that might be discussed, since we’-ve seen that TradeSports-InTrade is a much more powerful predictive tool for the US elections. Ditto for BetFair for U.K. elections.
#4. Decision-oriented prediction markets/exchanges: I would put here the kind of stuff that Robin Hanson is so excited about.
#5. Entertainment-oriented prediction markets/exchanges: Hollywood Stock Exchange, Washington Stock Exchange, Inkling, NewsFutures.
#6. Education-oriented prediction markets/exchanges: The Iowa Electronic Markets fits here, partially, regarding the use that professors around the country make of their markets in classrooms.
—-
– I disagree with Google in #4. Maybe the Google internal prediction markets would fit in #3.
– I disagree with NewsFutures in #3 —-I acknowledge (at least partially) the predictive power of play-money prediction exchanges, of course.
—-
Should we judge markets/exchanges on INTENTIONS or on RESULTS? I don’-t give a damn that TradeSports-InTrade and BetFair were created for speculation– if they have better predictive power than IEM, I’-m fine with them. Ditto for the HSX. I don’-t give the first fig that it was created as an entertainment tool. It’-s the best forecasting tool for the movie business, period.
—-
For the links to the prediction exchanges, see CFM.
—-
Previous Blog Posts:
– Prediction Markets DEFINITIONS – not a “taxonomy”
– Professor Robin Hanson’s draft definitions is validated by professor Eric Zitzewitz.
– Prediction Markets Definitions – REDUX
– Prediction Markets Definitions – by Robin Hanson – 2006-11-21
—-
Addendum: Robin Hanson has posted a comment…-
“Oriented” is not clear enough for my tastes. Is this about trader motives? Trader results? Price results? Exchange motives?
—-
My Answer: I meant “-exchange motives”-. […- See my comments. …-] But now that I think of it, another classification taking account of the “-price results”- makes more sense.
Previous blog posts by Chris F. Masse:
- Collective Error = Average Individual Error – Prediction Diversity
- When gambling meets Wall Street — Proposal for a brand-new kind of finance-based lottery
- The definitive proof that it’s presently impossible to practice prediction market journalism with BetFair.
- The Absence of Teams In Production of Blog Journalism
- Publish a comment on the BetFair forum, get arrested.
- If I had to guess, I would say about 50 percent of the “name pros” you see on television on a regular basis have a negative net worth. Frightening, I know.
- You can’t measure the usefulness of a system by how many resources it consumes.
“Oriented” is not clear enough for my tastes. Is this about trader motives? Trader results? Price results? Exchange motives?
ORIENTED
I meant that to state the INTENTION of the creator of each prediction market.
IEM scholars’s intention is clearly forecasting.
etc.
I meant “exchange motives”.
Of course, one could look at each prediction market/exchange according to:
– Trader motives
– Trader results
– Price results
– Exchange motives
I meant “exchange motives”, because the exchange managers would object to the other criteria. (IEM scholars will never admit that their simplifying the futures made speculation easier for the average, unqualified, individual traders, and that has been the key of the IEM success.)
But now that I think of it, another classification taking account of the “price results” makes more sense.
It is not clear to me that the motives for creating the IEM were forecasting. Quoting from their ‘92 AER paper:
“There are several reasons why this market is interesting. First, its finite duration coupled with payoffs determined by an observable outcome allows us to ask how well the market worked. … Also there are alternative means of generating forecasts of election outcomes (namely, public opinion polls), and these serve as a comparative standard to judge market performance. Second by placing this market in the untraditional area of political science we can study the effect of judgement biases on market performance. … Finally the data collected in this market allow for the detailed analysis of the behavior of traders. In full-scale financial markets even the most finely detailed available data consist merely of anonymous prices and quantities of separate trades.”
This doesn’t make it sound like they were interested in the forecasts themselves at all.
“First, its finite duration coupled with payoffs determined by an observable outcome allows us to ask how well the market worked. … Also there are alternative means of generating forecasts of election outcomes (namely, public opinion polls), and these serve as a comparative standard to judge market performance.”
Here, it seems to me that IEM is interested in studying the absolute and relative accuracy of the market-generated predictions. (Am I correct?)
The IEM co-founders, according to my reading of these two lines, are not interested “in the forecasts themselves”, but in the efficiency of the “prediction generator”. (Am I correct?)
I would say that a genuine scientist interested “in the forecasts themselves” logically becomes interested in applying the scientific approach to investigating the “prediction generator”. His/her objective would be to know better the mechanism, so as to propose improvement to the machine, and so as to generate more accurate forecasts later on.
Next comment, I will bring a damaging document (or so I hope).
Decision 2000 – Predicting Elections Using Financial Markets: The Iowa Electronic Markets – (PDF – edited transcript of a speech) – by Prof. Robert Forsythe (IEM) – @ Milken Institute Forum – 2000-02-09
http://www.milkeninstitute.org…..hetran.pdf
QUOTE
The political markets are open to the public at large. We run some other markets that are open, at this point, only to members of the academic community. We’ve also run earnings markets on corporate earnings. That was an experiment for us. If you think about political markets, what do we compare ourselves with typically? We compare ourselves to polls. For those of you that follow current corporate earnings announcements, there are people that are like pollsters. There are analysts that are marking earnings forecasts all the time. And so we’ve tried to run some markets on corporate earnings.
UNQUOTE
QUOTE
What started us thinking about this was the Michigan primary in 1988. If you remember, Jesse Jackson won that primary and it was an event that the pollsters completely missed. The pollsters thought that Jackson would do well. Nobody had him running in that primary. Four of us went out to lunch that day and three or four hours later we were convinced this was a good idea. I should say at the outset that this was not an idea that’s unique to us. There is a whole history of markets like this being run at other universities but it’s always been sort of informal – the notebook in the faculty lounge where people could trade. So we decided to take this much more seriously and to try to run a larger scale, more formal market mechanism to see whether markets could be used to predict a non-market event such as an election.
UNQUOTE
QUOTE
The market does well in this country in predicting the outcome, not only in an absolute sense, but also relative to the polls.
UNQUOTE
MY READING: The IEM was created with the purpose of investigating the absolute and relative accuracy of market-generated US political predictions.
MY CONCLUSION: No researcher would be interested “in the forecasts themselves”. A researcher would focus his/her interest on the “prediction generator”, its mechanism, its possible improvement, so as to generate more accurate forecasts later on.
And so I’m entitled to classify IEM in the “forecasting-oriented prediction markets/exchanges” category, if I stay true to the IEM co-founders’s intentions (as I read them, but I could be wrong). And, of course, as Robin Hanson seemed to suggest, there could be other ways to classify, for instance with the “price results” criterion.
Does what I say above help in the “Prediction Markets Definitions” debate? I welcome any comment(s).
To me, the purpose of investigating the accuracy of markets relative to polls is very different from the purpose of wanting to know who will win the election. This is part of why I want us to have clearer terminology, so we do not lump together things that are importantly different.
I think it’s pretty stupid and fruitless to try to fit every prediction exchange out there into a *single* category. Every one of these websites have multiple purposes, and their traders and/or visitors have multiple motivations. Why should Tradesports not be also in the entertainment category? I, and many others, certainly use it just for entertainment, not to “speculate” nor to earn a living. And why should NewsFutures and HSX, for insance, not be also in the “forecasting” category? It’s certainly also one of their primary purposes, and the primary reason why the media occasionally cares to write about these sites. This classification exsercise, at least its methodology, feels arbitrary and confusing.
Emile, yes markets serve many purposes and trader have many motivations. So unless we are to forgo making any distinctions between markets, it seems to me that we must focus on the *primary* purposes and motivations. These would be the ones most causally responsible for the existence of the market or the volume of trade.
To Emile:
– I proposed a classification based on what I thought was the primary motives of the founders of the exchanges.
– I didn’t say that it was a *good* and/or *definitive* classification. Please, don’t put words in my mouth. At the contrary, I said that another classification based on price results would be maybe more interesting.
– I agree with what Emile Servan-Schreiber said.
– I understand what Robin Hanson is trying to do.
I stay tuned.
Thanks for your attention,
“We must focus on the *primary* purposes and motivations. These would be the ones most causally responsible for the existence of the market or the volume of trade”
I am unclear how a market established to predict when a protein-folding problem will be solved, but in which the vast majority of trade is uninformed gambling would be classified. If “both”, then I am not sure what we are supposed to do with the definitions.
Pursuing an establishment-purpose set of definitions again seems mainly oriented towards regulatory and legal concerns (and to a lesser extent, public perception – including defining that “public”).
With that in mind, the more timely clarification might be what constitutes a “commodity”, an “option” or a “security”, and whether a “negotiable conditional note” is a viable animal.
Hmmm… “Commodity” seems clear to me. There’s a physical good whose ownership is being traded, even if physical possession never changes. Orange Juice, crude oil, but not payoffs based on outcomes.
I like “negotiable conditional note” as a generic description of a prediction market claim, though it doesn’t obviously rule out convertible debentures (see http://www.investorwords.com/1…..nture.html or Wikipedia), some of which are conditional on corporate events, IIRC.
An option is clearly different in that it gives someone the right (but not the obligation) to buy or sell an underlying instrument (which might itself be an option or a future.)
“Security” is the most general concept, encompassing all of these instruments and yet others.
Chris, Except for the negotiable conditional note, I was only alluding to the CME Credit Event Derivatives controversy, where the other three terms are in play to some extent:
http://www.aleablog.com/2006/1…..ghts-back/
Because of jurisdictional friction between the SEC and CFTC, the meanings of these terms are more contested and less intuitive than your descriptions.
Insofar as getting clear on definitions is aimed at securing favorable legal/regulatory treatment, this seems like a more urgent discussion than say differentiating hedging from information markets.
QUOTE
“Commodity” seems clear to me. There’s a physical good whose ownership is being traded, even if physical possession never changes. Orange Juice, crude oil, but not payoffs based on outcomes.
UNQUOTE
With all due respect to Chris Hibbert, I side with Jason Ruspini.
http://www.futuresindustry.org…..asp?a=1106
QUOTE
A broad interpretation of “excluded commodity” might include betting transactions on sporting and other events. Wagers on sporting events might satisfy the definition because, absent chicanery, the occurrence or contingency is not within the control of the parties to the relevant contract and the outcome may be “associated with an economic consequence,” i.e., a payout to the winner. A narrower interpretation would require the associated financial commercial or economic consequence to be other than a payment to the winning counterparty on the contract itself. However, even under the more restrictive reading, it could be argued that there are general economic consequences that could be associated with the outcome of high-visibility sporting events. These might include increases or decreases in restaurant receipts, subsequent advertising rates in local markets, or sales of particular types of merchandise. A broad reading of “excluded commodity” would therefore make such event-market wagering eligible for listing on an EBOT (albeit only between eligible participants.) Moreover, were such events to be considered to be “commodities,” they might even be offered on designated contract markets without restriction as to market participant. The CFTC has not yet provided guidance, however, as to how it interprets this language.
UNQUOTE