-

Robin Hanson:

One proposed distinguishing criteria includes the size of an individual trader’s stake, and the number of traders. The Iowa Electronic Markets are limited on both of these parameters. Such limits do succeed in preventing large hedging markets from masquerading as info-motivated event markets. But they do little to prevent generic gambling markets from masquerading as info-motivated event markets.

-

Total absurdity.

-

UPDATE: Robin Hanson comments…

Again, important for what purpose? The CFTC was clear that they are concerned about how to keep generic gambling from slipping in via whatever they might approve. I was addressing that concern. I don’t see how you can read anything I said as arguing against liquidity.

-

It’s a major contribution to the general discussion about prediction markets.

-

Please, allow me to disagree on one thing he said.

-

Robin Hanson:

One proposed distinguishing criteria includes the size of an individual trader’s stake, and the number of traders. The Iowa Electronic Markets are limited on both of these parameters. Such limits do succeed in preventing large hedging markets from masquerading as info-motivated event markets. But they do little to prevent generic gambling markets from masquerading as info-motivated event markets.

-

I have a fundamentally different view. What is important is not what Robin Hanson thinks of TradeSports-InTrade, BetFair-TradeFair, and HedgeStreet —and what their motivation(s) are.

What is important is whether those prediction exchanges do generate trustworthy and accurate probabilistic predictions. Period.

They do or they do not. Period.

-

And guess which prediction exchange has been more than willing to host Robin Hanson’s conditional prediction markets: Hint: it’s not the Iowa Electronic Markets.

Did Robin Hanson tell that to the CFTC?

-

UPDATE: Robin Hanson’s comment, posted below this post.

Chris, I did not take the CFTC call for comments as asking for what I would choose if I were king. I instead took it as them asking for help negotiating treacherous political waters. Part of the reality of their difficult situation is that the public, and hence Congress is quite wary that new rules might in effect legalize general gambling. Convincing the public that betting markets provide accurate predictions is not enough to convince them to legalize such markets in general. If there is any doubt, if I were king I would in fact legalize general gambling.

-

Date: Mon, 07 Jul 2008 10:12:46 -0400
To: secretary@cftc.gov
From: Robin Hanson <rhanson@gmu.edu>
Subject: Comment on “Concept Release on the Appropriate Regulatory Treatment of Event Contracts”
—————————————————————-
I am an event market innovator, having published the first detailed discussions envisioning their widespread application, having designed a widely used trading mechanism (the market scoring rule), and having co-developed the first internal corporate markets (at Xanadu), the first public web markets (the Foresight Exchange), and the aborted-but-influential Policy Analysis Market.

As I am less well trained in law than social science, I will not comment on what the C.F.T.C. is legally authorized to do, but only on how various policies correspond to public interest and public opinion. I speak here only for myself and not for any organization with which I may be affiliated.

The degree and type of regulation appropriate for a financial market depends on traders’ motives. Long ago most everything beyond direct physical exchange was widely discouraged or prohibited as “gambling” or “speculation.” The motives imputed to traders seemed to be some combination of mistakes, overconfidence, thrill of action, love of risk, and showing off one’s confidence and risk tolerance.

While public opinion on gambling has changed little, eventually legal exceptions were carved out for markets where, though speculation was still possible, enough participants had more sympathetic motives to garner public support. Securities markets allowed business managers to hedge ownership, insurance markets allowed hedging of various idiosyncratic risks, and commodities futures markets allowed hedging of various common risks.

It has long been noted approvingly that such speculative markets often had the desirable side effect of inducing people to collect info and aggregate it into prices. But until recently such info was not considered or accepted as a primary explanation or justification for a market’s existence. Given the myriad ways our society now suffers, often dramatically, from failures to aggregate info, I am very optimistic about the long term potential for such markets to offer substantial social value. However, the question remains of how such info-motivated markets should be regulated today.

Ideally an entire new regulatory regime would be carved out, on par with regimes for securities, insurance, and commodities futures regulation. But who would bother with such an effort before such markets had proven themselves able to realize substantial social value? And how could such markets prove themselves without at least tentative legal spaces in which to experiment? I know of no good reason why the C.F.T.C. should not provide one of the first such spaces.

Two key issues face a new regulatory regime for info-motivated event markets, especially one carved out of a common-risk-hedging commodities-future regulatory regime:

• How does optimal regulation of info-motivated event markets differ from that of common-risk-hedging markets?
• How can regulators ensure that this new regime is not used as a back door to escape prohibitions on other commodity futures trading, or to escape general prohibitions against gambling?

How Does Optimal Regulation Differ Here?

On the first question, the largest difference I see, by far, is the appropriate scale. When hedging risks it makes sense to focus first on risks, and hence trades, which are a large fraction of the wealth of the individuals or organizations involved. If risks are common there should be many who trade if any trade, and so market volume should be many times individual wealth levels. It also makes sense to devote a small fraction of this volume to efforts to avoid foul play. I have heard that it costs on the order of a million dollars to jump the regulatory hoops to gain approval for such markets, and I cannot say that this is not roughly the right cost magnitude.

For markets whose main function is to collect info, however, the appropriate scale seems far smaller. To collect info on a topic, those who know or could find out need only be offered a sufficient incentive to bother. In the lab, experimental economists see substantial effort and price info aggregation when only a few tens of dollars are at stake, and field data seems consistent with this estimate. If most of the social value from info-motivated event markets were concentrated in a few very important topics, it would not matter much if regulatory barriers prevented markets on topics with small info values. But if, as seems more plausible, much of the value is found in a long thick tail of smaller topics, then to realize this social value it is essential that regulatory barriers to creating such markets be reduced to the lowest feasible level.

For example, consider a topic where a social value of one thousand dollars could be realized, if only people were allowed to trade in a market on that topic. It is hard to see how this value could actually be realized if the regulatory cost to create this market were more than a few hundred dollars. If there were a million such topics, the total social value such markets could create would be one billion dollars.

A related difference is when it makes sense to limit participation. If most of a certain kind of risk is held by wealthy individuals or large organizations, then it can make sense to limit participation to such traders. But for info collection it is crucial to allow participation by the sorts of people who could plausibly obtain that info. For a great many topics these people will be spread out in the population, and not easily distinguished from most other people. A broad permission to participate will thus be desired in such cases.

How Can We Distinguish When This Regime Should Apply?

On the second question, we seek a reliable way to distinguish markets where the info collected is a strong rationale for its existence, a rationale strong enough to justify overturning the usual public presumption against generic speculation, and strong enough relative to hedging rationales to justify using this new regulatory regime, rather than other hedging regulatory regimes.

One proposed distinguishing criteria includes the size of an individual trader’s stake, and the number of traders. The Iowa Electronic Markets are limited on both of these parameters. Such limits do succeed in preventing large hedging markets from masquerading as info-motivated event markets. But they do little to prevent generic gambling markets from masquerading as info-motivated event markets.

Another proposed distinguishing criteria is the form of the organization that hosts the market. Some have proposed that tax-exempt, research, and government organizations be given wider latitude than for-profit businesses. I understand that this matches a common public perception, but honestly it seems mostly wishful thinking to believe that such organizations are substantially more likely to create markets with a strong info rationale, or to avoid whatever problems one fears with for-profit businesses.

Some have suggested that topics could be used to distinguish the strength of info rationale. Markets on sporting events might be presumed to have low info rationale, while markets on public policy might be presumed to have high info rationale. This approach seems to open a proverbial “can of worms,” however, requiring a great and continuing effort to categorize topics.

To ease this effort, one could inherit some other topic categorization. For example, regulation of speech distinguishes topics where free speech is presumed to perform very valuable social functions, and so has strong legal protection, from topics where such functions are less clear, allowing speech to be more easily regulated. Event markets might be permitted on topics where free speech has a strong legal protection.

In contrast to such weak indicators let me propose a stronger indicator of when a speculative market has a strong info rationale. I am not proposing that only markets which sport this indicator be allowed, but rather that at least such markets be allowed. My proposal is to permit markets where a sponsor pays to ensure that traders on average do not lose financially from participation, as this payment creates a strong presumption that this sponsor expected to gain substantial value from that info.

It is hard to see many of the benefits that traders may gain from trading, but we can more easily see the average financial costs that traders suffer. Traders may have to pay for permission to trade, to deposit into a system, to check prices and trading history, for each trade, and to withdraw their winnings. In addition, trader deposits may not earn competitive risk-adjusted rates of return. Payment is sometimes in the form of seeing ads. Such fees are essential to the profitability of “gambling” businesses today that rely primarily on traders’ speculative motives.

If for a particular topic, a sponsor were willing to ensure that traders paid none of these common trading fees, that sponsor would have credibly suggested that his or her market would not exist if that sponsor did not expect related info to have substantial value. If this sponsor furthermore subsidized this market, allowing traders to gain on average by trading against ignorant automated market makers, this would show even more clearly that this sponsor valued the resulting info. Such measures would ensure that traders suffered no average financial loss from their participation, though traders could still lose on average, such as by wasting too much time dealing with these markets.

Of course we do not expect sponsors to arise to support all topics where info collected by trading would have substantial social value. We expect businesses to sponsor markets on topics where they can profit from info, and charities to collect donations to support markets on topics they consider more broadly valuable. But we also expect many coordination failures, where each party prefers that others pay for commonly valuable info. So the case for prohibiting markets that fail my proposed criteria is much weaker than the case for permitting markets that meet this criteria.

It also remains possible that even when a sponsor finds info to be valuable enough to pay for, the social value of that info could be much less than the private value to this sponsor. If we could identify classes of such cases, these classes might form the basis of exceptions to this general permission I propose.

I have many other opinions about how such markets might be defined and regulated, but I’ve already gone one for quite a bit here – if you like what you see here and want more, you know where to find me.

In Summary

In addition to existing regulatory regimes for ownership-hedging securities, idiosyncratic-risk-hedging insurance, and common-risk-hedging futures, it could make sense to have a distinct regulatory regime for markets whose main reason to exist is the info that they collect. Compared with existing commodities futures regulation, such a regime should set a much lower barrier to creating such markets, as much of the social value may be distributed in millions of small markets. And while it is hard to determine in general which markets would create high social info value, relative to cost, we should presume such high value when a sponsor is willing to pay to ensure that traders suffer no average financial cost from their participation.
—————————————————————————

Robin Hanson rhanson@gmu.edu http://hanson.gmu.edu
Research Associate, Future of Humanity Institute at Oxford University
Associate Professor of Economics, George Mason University
MSN 1D3, Carow Hall, Fairfax VA 22030-4444
703-993-2326 FAX: 703-993-2323

Robin Hanson

-

1) Market manipulation will not be tolerated. PPX defines market manipulation as engaging in any activity which may artificially affect the market price of a security, or any other PPX market information. Market manipulation may include, but is not limited to: coordinated trading by different PPX accounts, unauthorized computer-assisted trading using a single or multiple PPX accounts, inappropriate disclosure of trading activity, and misrepresentation of information to the PPX community. PPX will impose heavy fines and/or trading restrictions on traders who make use of market manipulation for any purpose. Any determination as to whether market manipulation has occurred will be in the sole discretion of the PPX Administrator; all determinations are final and noncontestable.
1st offense: a fine of 25% of Portfolio Value.
2nd offense: a fine of 50% of Portfolio Value.
3rd offense: the confiscation of the Account Portfolio by the PPX Administrator.

2) Each individual trader may operate only one PPX account. An “individual” is defined as a person whose given name and physical address is represented on a valid Federal photo ID. Monthly contest winners must present a photocopy of said Federal photo ID as well as a signed affidavit in order to collect contest prizes.

OUTSTANDING TRADING SYSTEM
Amazingly accurate daily and intraday market forecasts of the Dow, NASDAQ, S&P 100, S&P 500, and the major overseas markets.
TradingPRO is and easy-to-trade trading system, specially designed to day trade the market indexes or correlated products (future indexes, e-minies contracts, options, and cash market) giving you easy to follow trading signals. [...]

More than a decade developing technology for the trading community.
Trading Pro has been for the last years the most technologically advanced leading indicator available to the individual trader and financial institutions.
As in the military and aerospace industrial sectors, certain technologies are not available for the consumer market until some years later. Products and technologies as complex as the GPS or as simple as the Velcro-strips where originally developed by the military or aerospace industries for high professional use. Even the Internet was born as a military project.
The financial industry is no different. Large financial institutions, investment banks and hedge funds, are the first to find and exploit complex technologies and trading strategies that are only available to the individual trader years after.
Trading Pro is giving access to individual traders to a set of cutting edge technologies, only available before to large financial institutions. For the last 3 years our systems have been used by more than 3,000 traders world wide.
Our technology has been in developement by physicists, mathematicians and programmers since 1993. System dynamics, chaos theory and artificial intelligence are only a piece of the technology used to obtain our amazing results.
Can you afford not to trade with this technology at your hands? We invite you to test and trade our trading models.
Trading has just become easier.

Paper: a measure of statistical significance of the forecasting models employed at Trading Pro.

- No link from the frontpage to the listing of their play-money prediction markets. (These two Chicago Boys have a screw loose, in my view.)

- I don’t like the blablating under the “Inkling Markets Explained” headline. Too long. Not accurate. (Mitigating risk, you say??? You can’t be serious.)

- 18 clients listed. Great. (Take that, Robin Hanson! ) Psstt… “A gaming software company is using Inkling to predict industry quality scores of their products.“, on the FAQ page. Intriguing.

- The “Get an Inkling” page is good.

- Free trial. Great. (Previous: Inkling Markets: Try Before You Buy. — REDUX)

- The “Feature rundown” page is very detailed. I noticed two great things. Number one, they finally pay tribute to professor Hanson. “Use of (Robin Hanson’s) battle tested Market Scoring Rule“. Number two, they seem to have built a step-by-step explainer for th newbies: “Step-by-step wizard to create a market (5-10 minute process)“. Great on both points.

- There’s a long FAQ page. Here are my remarks:

a) As I wrote here many times, the stock market metaphor (used in their first paragraph) is wrong. Let’s talk about event derivatives or event futures, rather.

b) The wisdom of crowds is incorrectly explained (and not linked to). What count are participant diversity, judgment independence, decentralization and a collective verdict mechanism.

c) The prediction market scene is well described overall, with some minor omissions —IEM, TradeSports, HedgeStreet, HSX, WSX and the Sim Exchange are cited, while NewsFutures is not, for instance. (Payback for the NewsFutures blog listing only IEM, InTrade and HSX as prediction exchanges, maybe.)

d) The section “What value do businesses get from using prediction markets?” oversells prediction markets, in my view.

e) Another plug for Robin Hanson:

How is the price set? Does Inkling work just like the real stock market?
The real stock market works using an auction (called a Continuous Double Auction in market vernacular.) When you buy shares of a stock, you are actually buying them from someone else. You never know who that is because there are so many trades happening, a computer handles it all. The price represents whatever the last trader was willing to pay for shares. In Inkling, the price is set automatically according to the volume and sentiment in the market. For example, if someone buys 10 shares of a stock, we increase the price according to an algorithm. In essence, Inkling is always the seller if you want to buy shares and the buyer if you want to sell shares. The algorithm we use is called the Market Scoring Rule, invented by Professor Robin Hanson at George Mason University.

Are there advantages/disadvantages to using the Market Scoring Rule?
Using a market scoring rule means there can be far fewer traders in a market because there does not have to be as much “liquidity” for the market to reach a conclusion. Using an algorithm also enabled Inkling to remove a lot of concepts related to stock markets that are difficult for people to understand, i.e. bid-ask spreads, etc. Because an algorithm is setting the price of the stock vs. the traders themselves, price jumps and drops are much more methodical. Where as in an auction, the price is whatever someone is willing to pay for a stock so the swings can be much more dramatic.

f) The issue of manipulations is addressed:

Can people manipulate the market results?
An individual trader with a single account cannot manipulate the markets. If one person has multiple accounts, it is possible to “game” the markets by constantly buying and selling shares between accounts to pump up one account. In a corporate setting where people only have one email address (our method of authentication,) this is a non-issue. There is of course always the possibility of collusion among market participants, which would require checks outside Inkling’s purview. Regardless, we deploy a rules engine that examines each user that joins the marketplace to detect signs of manipulation.

How much money do people get to start?
New accounts receive $5,000 inkles to start.

Why only $5,000?
We determined after much experimentation during the development of our product that $5,000 enables a new trader to make one larger trade or many smaller trades, but does not allow someone to immediately have great influence on any individual market. This is important because for someone to exert great influence over a market, they must have exhibited the ability to predict accurately in previous markets.

Overall, Inkling Markets has a very good marketing approach with their “try it before you buy” motto. We have a winner, here, folks.

Addendum: I forgot this from the FAQ…

How many people must participate in a market for it to be accurate?
We’ve found if markets have at least a dozen participants who make one or more trades over the lifetime of the market, the outcome has a higher probability of being accurate than a market with fewer traders.

Another global influenza pandemic, similar to the 1918 occurrence, is considered inevitable by many experts. Recent cases of avian influenza (H5N1) in birds and humans have led many to fear that this strain of the “bird flu” may produce the next worldwide influenza pandemic. Recent observations indicate that H5N1’s pandemic potential is increasing. However, the likelihood that a strain of H5N1 will cause the next pandemic and where and when such an infection will begin are still not known.

Despite considerable uncertainty, information does exist that may help predict when and if the next pandemic will be caused by an H5N1 strain. For example, an increase in the number of human clusters might suggest an increase in the likelihood of an H5N1 pandemic. Collection and analysis of such information is difficult because it is geographically dispersed and not always shared among health professionals (e.g., epidemiologists, clinicians, basic scientists). Currently available methods do not efficiently collect and interpret these data. We believe that a web-based prediction market designed to collect disparate information from widely dispersed health care professionals can serve as an avian influenza “barometer”.

An avian influenza prediction market will not replace existing influenza surveillance systems nor will it eliminate the need for improvements to the existing systems. Instead, we propose this prediction market as a supplement that can quickly aggregate expert opinions based on existing surveillance information. Prediction markets have been shown to be especially effective at quantifying subjective data. The probabilities generated by this market could help policymakers and public health officials coordinate resources, facilitate vaccine production, increase stockpiles of antiviral medications, and plan for allocation of personnel and resources.

The consensus within the health-care community is that the world is unprepared for the next influenza pandemic. This lack of preparedness exists at all levels: global, national and local. Several months or even weeks of warning would help with public-health preparedness and could prove crucial to saving lives. The World Health Organization (WHO) has produced a blueprint for pandemic preparedness with recommendations for certain actions to occur at specific levels of alert. Each level of alert is linked to specific, confirmable events (e.g., identification of a new influenza strain with unusual pathogenicity). The goal of this research project is to provide more accurate estimates for the likelihood of certain events — by date and location — than are currently available.

There is nothing extraordinary about prediction markets. As an alternative method for gathering and processing information about disease occurrence, pathogenicity, and progression, however, they are novel and hold much promise. A successful avian prediction market will include epidemiologists, clinical microbiologists, clinicians, virologists, and veterinarians. The Program for Monitoring Emerging Diseases (ProMed) is the largest assembly of individuals concerned with monitoring and tracking emerging infections diseases. We will recruit exclusively from within ProMed’s membership. To join ProMed, go to www.promedmail.org.

Iowa Electronic Markets: Influenza Prediction Markets

At the beginning of the influenza season, we recruit a diverse group of health care workers with information about influenza activity. Each of the traders is given an educational grant of $100 with which to trade. This grant grows or shrinks over the course of the influenza season depending upon the accuracy of the individual trader’s predictions. Potential traders include physicians, nurses, pharmacists, clinical microbiologists and epidemiologists. Each of these traders has access to unique information regarding current and future influenza activity. For example, if a clinical microbiologist suddenly sees an increase in the number of respiratory cultures positive for influenza, influenza will likely increase in her community during the next few weeks. Such a trader could log-on to our website and purchase shares for “widespread influenza activity” over the next few weeks. She would also likely try to sell her shares of “little-or-no influenza activity”. On the other hand, if a nurse had seen several patients with influenza in the emergency room for the last few weeks and then saw no more patients, he would be tempted to sell all his “widespread influenza activity” shares and try to buy as many “little-or-no activity” shares as he could.

Each influenza “share” or “contract” is associated with a specified level of influenza activity during a specific week and is worth $1.00 when trading ends if that turns out to be the observed level of activity. Thus, the price at which trades take place represents the consensus belief about the likelihood of that level of activity. For example, if the price of widespread activity for a given week was $0.90, the market expects influenza to be widespread with 90% probability.

Each influenza contract is based on the Centers for Disease Control‘s weekly color-coded system: RED=Widespread, BLUE=Regional, PURPLE=Local, GREEN=Sporadic, YELLOW=No Activity. Individual trader’s predictions are judged accurate or not based upon the levels of activity reported by the CDC. [...]

Traders in our market actually do get a check in the form of a financial grant to use for education purposes (e.g., buy books, or to help pay for a conference). The grant is awarded at the end of the season and the size depends on how successful participants were in predicting influenza activity.

The financial amount involved (traders start with $100) is not enough to really encourage traders to trade up to several times a week for an entire influenza season. Most participants reported that they truly enjoyed the process of trading itself and seeing how accurate their forecasts are compared to those of other traders.

About 100 Press infiltrations about the IEM’s Avian Influenza Prediction Markets – AP story.

A predictive market for seasonal flu, operated in Iowa and North Carolina, correctly predicted the current level of flu activity in an area 71 percent of the time, even though the predictions were made before any solid surveillance reports. The market had a 50 percent success rate at predicting flu activity a week in advance, according to a recent report in the journal Clinical Infectious Diseases.

And Tyler Cowen of Marginal Revolution took the bait, too. His comment to a someone who said that TradeSports-InTrade has avian flu contracts, too:

Note that the Tradesports contract is on the uninteresting question of whether avian flu arrives in the United States, and not whether it starts killing lots of people.

Exactly. (See my previous remark on the XM-Sirius merger prediction markets.)

Psstt… Tyler Cowen made a mistake, naming the NewsFutures avian flu forecast… a “betting market”. It’s not a series of prediction markets; it’s “competitive forecasting”. See: Caught Red-Handed Emile Servan-Schreiber

I am no academic expert, but after first bursting out laughing at M Servan-Schreiber’s suggestion, I now concede that when the costs of aggregating information are very low, reward-only markets can–at the moment–be expected to have greater predictive power than “real-money” (TS) markets, because there is no cost to admitting a mistake, so reward-only markets can be expected to inflect at earlier times than real-money markets–when the reward for successful aggregation is enough to compensate for the time lost. For things like the NFL futures markets (the subject of the Wolfers / Pennock / Galebach / Schreiber paper), I’d infer that virtually all serious NFL traders are looking at very similarly aggregated information, so the costs of aggregating information are extremely low, if they exist at all. Political information, on the other hand, is extremely diffuse and time-intensive to aggregate. Although a comparison of NF (reward only) and TS futures does show that NF did arrive at the correct conclusion earlier than TS markets did, the NF futures also exhibited extreme volatility–another indicator that reward-only traders are much quicker to arrive at individual, randomly dispersed inflection points (much less information pileup required for an individual trader to recognize the trend and shift his position, due to much lower cost to realizing error and executing a shift).

However, judging from the Tradesports political forums, the TS trader population was significantly right of center in its political leanings. It’s impossible to judge how predictive of a factor that would be, but most people would probably agree that it’d be significant enough that it’s definitely too early to crown NF the victor.However, even assuming for the sake of argument that reward-only markets will inflect earlier and more accurately no matter what the aggregation cost is, the strengths that currently allow reward-only markets to inflect earlier also make them infinitely more prone to manipulation. Once some real-money traders perceive that the rest of the real-money trading community places predictive value on reward-only markets, they will be able to distort the reward-only markets’ predictive value all they want, because there’s no cost to being wrong.

Even if there is a legitimate coupling between reward-only and real-money markets now, it wouldn’t be long before real-money incentives manipulated that coupling to extinction.

Addendum: A reader e-mails, “Suppose you had insider information and were inclined to use it, [and] once used it is likely to be aggregated as public information–would you use that information on a real money market or a play money market?

“Similar question–which market would be more likely to extract private information?”

Exactly. When 1) traders are using basically the same core data, and 2) public/private aggregation does not pay off (whether it’s trading on insider info, drilling down into the data with advanced statistical methods, using time-consuming inference to discern pieces of private information, etc), traders can’t get a data edge, so there is little reward for “superior traders” doing superior analysis. But when the picture of all the data is more difficult to discern, the real money PMs should pretty quickly run away with the crown.

PS–what are NF’s transaction fees? Do they have play money transaction fees?