• Markets are not prophets, prophets do not exist.
• A mechanism’s forecastability should not be judged against a virtual fool-proof prophet; we’d better compare it with other existing or widely-used mechanisms and -to my partial and context-bound knowledge- markets outperform all those.
• Markets are the only tool that intrinsically suggests their probability of failure. If Obama’s stock is traded at 70 cents, this suggests that there is a 30% probability of Obama losing; I’d say markets are by character modest and no fanfare has any place in describing their suggestions.
• Markets are primarily an aggregation/meta mechanism; as such, garbage-in-garbage-out effects are expected to happen, so we’d need to keep focus on minimizing garbage rather than blaming the market/compiler.
• Maturity of the mechanism and its use, as long as trading volume (in real-money intrade for example), have not yet reached a fully efficient level (more on this to come soon), but these result in significant profit opportunities, so I expect things to just keep getting better.

cross-posted from my blog

So, we start by giving you, the most demanding audience for such a service, early access to first stress-test askmarkets and to email us your impressions, proposals and bugs you encountered (we bet there are still some hiding somewhere), before really going ‘wild’, ..em public. Just navigate to our homepage, enter your email and password ‘midasoracle’ and the first 50 of you will get in (we’ll also invite the latecomers later on).

Textual ads are auction powered. Speaking of Google, let me repeat the well-known process (which is valid with some variations also in Yahoo’s Search Engine Marketing and Microsoft’s ad center -forgive my partial knowledge). Each time you view or click on an ad link, the advertiser rewards the site owner for the use of her screen real estate and the redirection of your attention span. More accessible real estate costs a lot, as the more the advertisers pay the higher their ad is ranked and exposing their message to more visitors. Ad ranking occurs mainly from a dynamic Vickrey auction model, namely cost-per-click bids is the most influential factor defining your ads rank. But, to my eyes, the system’s current status seems sub-optimal. From the ad platform engine’s perspective, earnings aren’t maximized, as the advertiser pays only for actual clicks and not for impressions. From the advertiser’s view, the ad’s impact is also sub-optimal, as high rank doesn’t guarantee more visitors, in the case of low ad relevancy to the user’s query and actual interests. Finally, and most importantly, the auction-based fundamental model doesn’t accumulate the collective intelligence of the previous visitors’ behavior and probably results in a poor user experience. (Well, in practice, clickthrough rates are also evaluated and the adrank algorithm is much more complicated, but this doesn’t reduce the validity of comments on the fundamental choice of an auction model).

While the auction-based approach apparently works, let me propose a more simple, direct and transparent market-based variation. In such a case, the ad-space of the universe of all potential keywords combination consists a gigantic marketplace, while each keywords’ combination will form a market in this marketplace. In each market, an advertiser’s submission triggers the creation of a stock, which is initially traded at the defined cost-per-click price. But this price is nothing but constant; it goes up each time a user clicks on the ad, or down each time a user clicks on a different ad (this second action could even be omitted). As a result, stock prices, therefore ad ranking, evolves dynamically to enhance previous visitors’ choices and leverage the wisdom of crowds in forming an elegant user experience. Moreover, it maximizes engine’s gainings and advertisers’ impact, while enables a fully trackable procedure for advantage of both the platform and its users.

I would like to stress my lack of extended or insiders’ knowledge on the topic, but I’m more than eager to discover relevant attempts, or -even better- participate in some attempt to put this idea into reality.

cross-posted from my blog

The question is, are the currently available prediction markets web services compliant with the era of attention economics? Are we able to attract a critical mass of users, thereby surpassing the tipping point needed to turn the mechanism of markets to a typical decision support and forecasting tool?

If a picture is worth a thousand words and assuming that a website’s source code is an unbiased descriptor of its complexity, I attempted to take a look at the homepages of some popular prediction markets web services, using this ‘websites as graphs‘ tool. In the results that follow, each cycle represents an html tag.

• intrade

• hsx

• newsfutures

• thewsx.com (by consensus point)

• buzz game of yahoo, a source of inspiration to me

• my beloved inkling markets

• our approach at askmarkets.com (yet in alpha version)

P.S.: I didn’t include betfair because the graph occurred wasn’t descriptive of the true complexity of their homepage.

Cross-posted by gtziralis.com.

The quality of presenters is excellent and the workshop has succeeded in attracting high level contributions. The program and complete list of papers to be presented are available here, while directly related to prediction markets papers are:

• Are Political Markets Really Superior to Polls as Election Predictors? by Bob Erikson, Columbia University and Chris Wlezien, Temple University
• Optimal Price Setting in Fixed-Odds Betting Markets Under Information by Raphael Markellos and Vasiliki A. Makropoulou, Athens University of Economics and Business
• Hitting Home Runs and the Art of Corporate Valuation: Do Managers or the Prediction Market Make Better Predictions? by Orly Sade, New York University and Hebrew University
• Prediction Markets as a Tool for Management of Political Risk by Ole Jakob Bergfjord, Norwegian School of Economics
• Public Signal Bias and Prediction Market Accuracy by Tom Gruca and Joyce E. Berg, University of Iowa
• Ignorance Prior Bias in Prediction Markets by Lionel Page, University of Westminster
• An Analysis of Tradesports’ 2005-06 National Football League Prediction Market by Feng Zhou and Philip O’Connor, University of Waikato
• Testing the Efficiency of Markets in the 2002 World Cup by Steve Levitt, University of Chicago and Ricard Gil, University of California at Santa Cruz
• Price Formation in Sports Prediction Markets: A Cross-Cultural Study by Stefan Luckner, University of Karlsruhe
• Socially Embedded Prediction Markets by Yiling Chen and David M. Pennock, Yahoo! Research
• Manipulators Increase Information Market Accuracy by Robin Hanson, George Mason University and Ryan Oprea, University of California, Santa Cruz
• Is There a Favorite-Longshot Bias in Election Prediction Markets? by Justin Wolfers, University of Pennsylvania, Eric Zitzewitz, Stanford University and Andrew Leigh, Australian National University
• Using Markets to Inform Policy: The Case of the Iraq War by Eric Zitzewitz, Stanford University and Justin Wolfers, University of Pennsylvania

The exponential pattern of growth recorded in my literature review of the field remains consistent, at least.

Cross-posted from gtziralis.com.

The integration with business intelligence software, corporate data warehouses and document indexes will definitely facilitate the task of information collection by traders. I claim that such an integration will also enable the introduction of artificial intelligence agents, namely virtual actors that would trade on the relevant information and knowledge extracted by the CRM/ERP software.

This introduction of a hybrid prediction market, a prediction market populated by both human and artificial agents (term just coined), could solve some problems, such as low trading volume and would by default incorporate extracted explicit knowledge, while also creating some new, like for example the need to fine tune its algorithms. But, after all, I believe that it could trigger a new fascinating research field (machine learning powered, data and textual actors featuring in a prediction market) and empower new hot business applications.

A lot of questions arise. To name a few:

• Could or will such hybrid markets perform better than traditional pure ones?
• Do virtual agents negate the typical virtues of a prediction market (e.g. tabula rasa opening, harnessing ‘just’ the collective intelligence, etc.)?
• Does a market-based meta-algorithm perform better than traditional data mining meta-algorithms (e.g. pure artificial markets vs adaboost)?
• Will such markets be included in the declaration of establishment of The Institute of Prediction Markets or in a forthcoming literature review?

The journey will probably be long but fertile; the ideas exchange is now open.

Originally published at my log.

I’m happy to report you the findings of my latest work, which has been accepted for publication in the inaugural issue of The Journal of Prediction Markets and is entitled as Prediction Markets: An Extended Literature Review. The final draft is available for download here.

The article, performed for the needs of my doctoral dissertation in NTUA, Greece, presents an attempt to collect and study the totality of the related to prediction markets academic work. 152 papers (journal articles, conference proceedings, book chapters, master’s theses and doctoral dissertations published between 1991 and 2006) were finally reviewed and classified. For further convenience of the researchers, the bibliography references are available for download in bibtex, endnote or txt format through my webpage.

I present here the basic findings of the literature review:

• There is a significant increase, which could be roughly described as exponential, in the volume of prediction market literature during the last years

• The terminology used to refer to the concept of prediction markets is very dispersed. The need for a universally adopted descriptor is therefore highlighted

• A classification scheme is also provided, which shows off the diversity of the research field and maybe suggests directions for further research

I hope that this work would serve as an expedient source for anyone interested in prediction markets.

Thank you for your time and comments.

after using the Google Custom Search Engine, which was launched yesterday, I made an attempt to create one on prediction markets. The result Prediction Markets Search Engine focuses on a directory of nearly 100 websites, which are highly relevant to the concept and produces promising results. For example, you can see the results occurred for the query artificial markets in both the google default search engine (left) and the prediction markets search engine (right).

The directory is and should probably always remain under construction, moreover is open to everyone interested in viewing or adding relevant sites (you just have to send me an email to invite you). What’s more, a search box of the prediction markets search engine is also available in the hopeful case that someone is interested to include it in her/his site.

Thank you in advance for your time and comments.