Main Page

From P-MART wiki

Jump to: navigation, search
m (Changed protection level for "Main Page" ([edit=sysop] (indefinite) [move=sysop] (indefinite)))
 
Line 1: Line 1:
'''Welcome to the Prediction Markets wiki of the Graduate School of Management, University of Haifa'''
'''Welcome to the Prediction Markets wiki of the Graduate School of Management, University of Haifa'''
-
This wiki was initiated by Daphne Raban and Dorit Geifman as a collaborative knowledge repository on the topic of Prediction Markets. We intend to update this wiki with resources and findings as we pursue the topic. True to our belief in the value of collaboration we encourage you to contribute.
+
This wiki grows out of ongoing research work in the field of Prediction Markets and Collective Intelligence which is performed at the [http://mba2.haifa.ac.il/index.php/english- Graduate School of Management] of the [http://www.haifa.ac.il/index_eng.html University of Haifa], under the supervision of Prof. [http://rafaeli.net/ Sheizaf Rafaeli] and [http://gsb.haifa.ac.il/~draban/home/ Dr. Daphne Raban]. At its core lies a mapping of domain vocabulary to market implementations which is presented in a paper ''"P-MART: Towards a Classification of Online Prediction Markets"'', currently under review.
-
To open an editing account please contact [mailto:dorit.geifman@gmail.com Dorit]
+
The wiki is structured as a cross reference of '''[[:Category:Markets|Prediction Market implementations]]''' and '''[[:Category:Market properties|market attributes]]'''. The description of each market is contained in a separate wiki article and tagged with its attributes (in the form of wiki categories). In this way it is possible to view the property of each market, and access any market from any of its market property. It is also possible to access all markets and market properties from the sidebar. In the main page you will find an overview of the Prediction Markets classification, where each of the market attribute is listed in '''bold''' and links to its category page.
-
Consult the [http://meta.wikimedia.org/wiki/Help:Contents User's Guide] for information on using the wiki software.
+
The objective of this wiki is to act as a knowledge repository for researchers and practitioners in the field of Prediction Markets. True to our belief in the value of collaboration we encourage you to contribute to this wiki. It is expected that through on-going contribution the domain vocabulary will be fine-tuned as the domain knowledge expands, and the implementation repository will grow as new examples emerge. Contributors need to register first. In order to maintain the structure of this wiki please follow [[Editing guidelines|editing guidelines]].
 +
 
 +
For more details contact [[User:Dorit.Geifman|Dorit Geifman]]
== Introduction ==
== Introduction ==
 +
We can trace the origins of the Prediction Markets concept to the works of [http://en.wikipedia.org/wiki/Friedrich_Hayek Friedrich Hayek], the winner of the Nobel Prize in Economics, and a social and political philosopher. Hayek claimed that the price mechanism serves to share and synchronize local and personal knowledge, allowing society's members to achieve diverse, complicated ends through a principle of spontaneous self-organization.
 +
 +
We are now starting to witness the proliferation of Information Markets into the corporate decision-making environment. One of the pioneering organizations to use Prediction Market as an internal decision-support tool is Hewlett Packard, which is using it as a sales forecasting tool. Intel too is using Information Markets for managing products forecast and production plans. There is evidence that many other organizations like Google, Eli Lilly and Microsoft are jumping into the water, but most of this activity is still in an experimental stage. Organizations are still trying to figure out the kind of incentives to be used, how to make sure that established hierarchies are not threatened and how to maintain a cost effective process etc.
== What are Prediction Markets used for? ==
== What are Prediction Markets used for? ==
Line 16: Line 21:
Prediction Markets cover a wide range of topics. These may be in the areas of '''[[:Category:Current or recreation events|Current or recreation events]]''', '''[[:Category:Business topics|Business topics]]''', '''[[:Category:Policy issues|Policy issues]]''', '''[[:Category:Synthetic events|Synthetic events]]''' etc.
Prediction Markets cover a wide range of topics. These may be in the areas of '''[[:Category:Current or recreation events|Current or recreation events]]''', '''[[:Category:Business topics|Business topics]]''', '''[[:Category:Policy issues|Policy issues]]''', '''[[:Category:Synthetic events|Synthetic events]]''' etc.
-
== Prediction Markets in Context ==
+
== In what setting are markets used? ==
-
We can trace the origins of the Prediction Markets concept to the works of [http://en.wikipedia.org/wiki/Friedrich_Hayek Friedrich Hayek], the winner of the Nobel Prize in Economics, and a social and political philosopher. Hayek claimed that the price mechanism serves to share and synchronize local and personal knowledge, allowing society's members to achieve diverse, complicated ends through a principle of spontaneous self-organization.
+
Prediction markets are implemented in a variety of environments
 +
* '''[[:Category:Public domain markets|Public domain]]''' – open to the general public. May take the form of stand-alone web-sites e.g. [[Intrade]] or add-ons to larger web-sites e.g. [http://www.ftpredict.com/ FTPredict] of the Financial Times.  
 +
* '''[[:Category:Corporate|Corporate]]''' - may be internal, limited to the employees, or external, open to other stakeholders such as customers or suppliers
 +
* '''[[:Category:Laboratory|Laboratory]]''' – a closed and controlled environment, set up specifically for academic experiments
 +
* '''[[:Category:Field experiment|Field experiments]]''' – controlled experiments in real-world settings
-
== Online Markets and Platforms ==
+
== How do Prediction Markets work? ==
-
The [http://en.wikipedia.org/wiki/Prediction_market Prediction Markets entry] in Wikipedia provides an extensive list of operating markets, categorized according real-money and play-money markets. It also lists a wide range of proprietary as well as open-source prediction markets software. The intention in this section is to mention only markets and platforms that can contribute in some way to deeper understanding of prediction markets and assist in the teaching and research activity.
+
Prediction Markets differ in their internal design, operational model and structure of assets. In this section we will provide an overview and references for more detailed information.
-
=== Iowa Electronic Market (IEM)===
+
=== Market models ===
-
[http://www.biz.uiowa.edu/iem/ Iowa Electronic Market (IEM)] is a real-money, non-profit market, operated by the Henry B. Tippie College of Business at the University of Iowa as part of its research and teaching mission. IEM was first opened on 1988 and is since known and widely cited mainly for its political prediction markets. Some of its markets are open to the public while others are available for classroom traders only. For the latter, the IEM web-site also includes [http://www.biz.uiowa.edu/iem/classroom/ course modules, sample assignments] and reference to many [http://www.biz.uiowa.edu/iem/archive/references.html research papers]. True to its research mission, IEM has opened its [http://www.biz.uiowa.edu/iem/archive/historicaldata.cfm historical trading data] for research activity and consequently it is widely cited in academic papers as well as by the media.
+
The mechanism behind the market exchange and pricing may be:
-
=== Zocalo - an open-source prediction markets platform ===
+
* '''[[:Category:Continuous Double Auction (CDA)|Continuous Double-Auction (CDA)]]''' - buyers bid prices up and sellers bid prices down until a consensus is reached. CDA can be implemented with or without an automatic market maker.
-
[http://zocalo.sourceforge.net/ Zocalo] is an open-source toolkit for building prediction markets. The project is lead by [http://mydruthers.com/ Chris Hibbert] and supported by [http://www.commerce.net/ CommerceNet]. The software was first released on July 2006 on SourceForge.net and there have been ongoing updates since. Hibbert's short-term plan is to support experimental markets in economic labs, at a later stage support internal pilot markets to businesses and for the long-term deploy public markets with partners. Currently it is in used by George Manson University (GMU) to experiment whether manipulators affect the market price. Hibbert is looking for collaboration and the software can be [http://sourceforge.net/projects/zocalo/ downloaded] from the Zocalo project page at SourceForge.net.
+
* '''[[:Category:Market Scoring Rules (MSR)|Market Scoring Rules (MSR)]]''' – all traders can continuously update their score rating i.e. the estimates of the probabilities of the event, and the underlying pricing function calculates the new probabilities (price). The model implements an automatic market maker.
 +
* '''[[:Category:Dynamic Pari-Mutuel (DPM)|Dynamic Pari-Mutuel (DPM)]]''' – in a pari-mutuel model, all bets are placed in a pool which is later divided between the winners. This model takes the form of a continuous pari-mutuel system.
-
== How does it work? ==
+
=== Security types ===
-
Prediction Markets differ in their internal design, operational model and structure of assets. In this section we will provide an overview and references for more detailed information.
+
The security type defines the payoff model. It can be one of the following:
 +
* '''[[:Category:Winner-takes-all securities|Winner-takes-all]]''' – the security is cashed out at a predefined price if its underlying event occurred, and is nullified otherwise. This model has been the most commonly used for binary (Yes / No) or multiple-option contracts.
 +
* '''[[:Category:Index securities|Index]]''' – also known as linear security. In this case the price of the security reflects the estimated value of the outcome and its payoff is tied to the actual outcome of the event.
 +
* '''[[:Category:Volume Weighted Average Price (VWAP)|Volume-Weighted Average Price]]''' – the security is cashed out according to its average price weighted by transaction volume in X days prior to its expiration date. This method is used mostly in idea / opinion markets, when there is no absolute truth that can be used as a reference for cashing out the market
 +
* '''[[:Category:Conditional securities|Conditional]]''' – the outcome of these securities is evaluated according to a combination of events i.e. the security pays $1 if some event happens conditionally to the occurrence of some other event.
-
=== Prediction Market Architectures ===
+
=== Trading currency ===
-
The different nature of Prediction Markets, mainly its special attributes stemming from it being used as an information aggregation mechanism, drove researchers to reconsider the traditional market architectures and propose some new market models. The main issues that have been addressed are market liquidity and number of traders, its quality as an information aggregation platform and the risk taken by market institutions.
+
Trading may be executed through the use of '''[[:Category:Real-money|Real-money]]''', i.e. trader’s money or play-money with a well defined monetary exchange rate, or '''[[:Category:Play-money|Play-money]]''', which is not directly tied to the final reward.
-
This section will first outline the basic market models and then describe the architectures that have been proposed and are used for Prediction Markets. For each model, data or platforms suitable for research purposes will be mentioned.
+
-
==== Market Models ====
+
=== Trading funds ===
-
Prediction Markets implement various exchange models and pricing mechanism:
+
The source of the funds which are available to the traders may be:
 +
* '''[[:Category:Limited own funds|Limited own]]''' – traders invest their own money, but the amount is limited by the market institution
 +
* '''[[:Category:Unlimited own funds|Unlimited own]]''' – traders invest their own money and the amount is unlimited
 +
* '''[[:Category:Initial capital endowment|Initial capital endowment]]''' – traders are endowed by market institution with initial capital, usually play-money.
 +
* '''[[:Category:Initial securities endowment|Initial securities endowment]]''' – traders' accounts are initialized with a number of securities.
-
* '''[[:Category:Continuous Double Auction (CDA)|Continuous Double Auction (CDA)]]''' where the market mechanism continuously matches orders to buy an asset (bid) with orders to sell an asset (ask).
+
=== Market size and duration ===
-
* '''[[:Category:Market Scoring Rules(MSR)|Market Scoring Rules (MSR)]]''' – traders can continuously update their score rating i.e. the estimates of the probabilities of the event, and the underlying pricing function calculates the new probabilities (price). The model implements an automatic market maker.
+
The market size is defined by the number of ''active'' traders during market activity period. This may range from '''[[:Category:Small size markets|small]]''', 15 or less, through '''[[:Category:Medium size markets|medium]]''' – 15 – 50 traders, to '''[[:Category:Large markets|large]]''' – larger than 50, common in public markets but can also be found in corporate settings
-
* '''[[:Category:Dynamic Pari-Mutuel (DPM)|Dynamic Pari-Mutuel (DPM)]]''' - a hybrid between the Pari-Mutual model, where all bets are placed in a pool which is later divided between the winners, and the Continuous Double Auction models
+
-
* '''Wagering markets'''
+
-
This model is typically found in sports and horse betting institutions. In this case the bookmaker sets the odds according to some expert opinion. The bookmaker may change the odds according to the level of betting, however the odds or the price are fixed at the time of the bet. The bookmaker profits by offering different odds for the two(or more if applicable)sides of the bet.
+
-
* '''Scoring rules'''
+
-
Scoring rules are used in the case(the simple scenario)where the outcome of an event has 'n' states and the sum of their probability to occur is 1. The trader is asked to report her opinion on the probability distribution between the states. Once the true state is known, the trader gets a reward which is proportionate to the probably s/he indicated for that event. This requires the market to have a patron that agrees to pay the reward.
+
-
=== Operation Models ===
+
Markets should be open as long as new information streams in, whether information that is exogenous to the market or information that is inferred from market price, and as long as the actual outcome is yet unknown. Here their duration is arbitrarily defined as:
 +
* '''[[:Category:Short duration|Short]]''' – does not exceed one day, in a laboratory setting usually lasts a few minutes
 +
* '''[[:Category:Medium duration|Medium]]''' – a few days, does not exceed one week
 +
* '''[[:Category:Long duration|Long]]''' – longer than a week
 +
* '''[[:Category:Variable duration|Variable]]''' – the market closing date is defined ad-hoc, and is dictated by the topic of the market and the external events which affect the subject of trade.
-
==== The Iowa Electronic Markets model ====
+
=== Interaction means ===
-
The Iowa Electronic Markets (IEM) is operated by the University of Iowa, the Henry B. Tippie College of Business, as part of its research an teaching mission. Consequently the exchange is run as a non-profit operation.
+
Does the market platform or trading procedure enable complementary information exchange between traders?
-
The contracts are issued for trading by means of '''unit portfolios''', which consist of one of each of the contracts as defined in the market prospectus. The price of the unit portfolio is the guaranteed aggregate liquidation value of the its contracts. At any point in time the trader can buy or sell the IEM exchange these unit portfolios through a fixed price bundle order. Once a trader buys the unit portfolio the contacts are put into circulation and executed upon through the different market actions.
+
* '''[[:Category:No interaction|No interaction]]''' – platform does not facilitate interaction among traders. Traders are instructed to refrain from off platform interaction.  
-
The market actions that are available to the trader are:
+
* '''[[:Category:On-platform interaction|On-platform]]''' – the platform provides for online interaction means such as chat or forums.
-
* '''Fixed price bundle orders''' – a request to buy or sell the IEM exchange a set of contacts which form a unit portfolio at a fixed price that is the guaranteed aggregate liquidation value of the assets within the unit portfolio.  
+
* '''[[:Category:Off-platform interaction|Off-platform]]''' – traders may exchange views outside the market about the events being traded.
-
* '''Market price bundles orders''' – a request to simultaneously buy or sell a set of contracts which form a unit portfolio at their respective market prices
+
-
* '''Limit orders''' - a request to buy or sell a contract at a specified price for a specified period of time
+
-
* '''Withdrawing of outstanding orders'''
+
-
* '''Market orders''' – requests to buy or sell contracts at the current ask and bid prices
+
-
=== Contract payout structures ===
+
== The traders ==
 +
The traders may '''[[:Category:Self-selected traders|self-select]]''' to join the trading activity, usually when they perceive themselves knowledgeable or interested in the topic of the market. Alternatively, they may be '''[[:Category:Invited traders|invited]]''' by the market owner according to the estimated potential of their contribution and expertise.
-
==== Linear ====
+
The traders may be '''[[:Category:Stakeholders|stakeholders]]''', having vested interest in the outcome of the event being traded, or they may have '''[[:Category:No-stake in outcome|no-stake in the outcome]]''', being fully neutral with respect to the outcome.
-
==== Winner-takes-all ====
+
The information which is available to the group of traders may be '''[[:Category:Perfect information|perfect]]''', i.e. the all information which is needed exists but is private to a subset of the traders, '''[[:Category:Complete information|complete]]''', i.e. each trader only holds uncertain or incomplete information but the combination of the information which is accessible to all traders forms a complete picture of the subject of trade, or '''[[:Category:Incomplete information|incomplete]]''', the available information, private or public, contains a variable level of uncertainty.
-
=== Incentive Models ===
+
== Why do traders participate? ==
 +
Market institutions use a variety of incentive schemes in order to encourage participation. These may take the following forms:
 +
* '''[[:Category:Performance incentive|Performance]]''' - payoffs are directly tied to the performance of the trader, i.e. gains profits from trading.
 +
* '''[[:Category:Performance-rank incentive|Performance rank]]''' – reward is associated with the rank of the trader in the descending list of performance.
 +
* '''[[:Category:Participation incentive|Participation]]''' – reward is tied to participation. Can be associated with participation level or drawn by lottery.
 +
* '''[[:Category:Contribution incentive|Contribution]]''' – participants are rewarded for their contribution to the marketplace, e.g. creation of new markets.
 +
The traders may be awarded a '''[[:Category:Monetary reward|cash]]''' payment, a material, non-monetary '''[[:Category:Gift reward|prize]]''', '''[[:Category:Bonus shares reward|bonus shares]]''' or an intangible reward such as '''[[:Category:Reputational reward|increased reputation]]''' or '''[[:Category:Intrinsic reward|intrinsic satisfaction]]'''.
-
== Is it Reliable and Accurate? ==
+
== Is the market outcome accurate? ==
-
== Classification of Prediction Markets ==
+
== Does traders' behavior affect market accuracy? ==
-
A link to the main classification page [[Information Markets - Classification]].
+
-
== Corporate Prediction Markets ==
+
=== Manipulation ===
-
We are now starting to witness the proliferation of Information Markets into the corporate decision-making environment. One of the pioneering organizations to use Prediction Market as an internal decision-support tool is Hewlett Packard, which is using it as a sales forecasting tool. Intel too is using Information Markets for managing products forecast and production plans. There is evidence that many other organizations like Google, Eli Lilly and Microsoft are jumping into the water, but most of this activity is still in an experimental stage. Organizations are still trying to figure out the kind of incentives to be used, how to make sure that established hierarchies are not threatened and how to maintain a cost effective process etc.
+
-
=== Examples ===
+
-
 
+
-
==== Google ====
+
-
As posted in the [http://googleblog.blogspot.com/2005/09/putting-crowd-wisdom-to-work.html Google official blog] on September 2005, a prediction market was set up inside the company with the intention of forecasting product launch dates and other things of strategic importance to Google. More than a thousand Google employees bid on 146 events in 43 subjects. This was developed as one of the projects under Google's policy of employees spending 20% of their time developing whatever new ideas they find interesting. Recently the results 2.5 years running the experiment were published [http://bocowgill.com/GooglePredictionMarketPaper.pdf Cowgill, Wolfers, & Zitzewitz, 2008]. It is interesting to note that the experiment did not only provide the outcome of the future events that were traded, but also an insight in the organization's internal information flows.
+
-
 
+
-
== Implementations of Prediction Markets ==
+
-
* General Electric's [[Imagination Markets]]
+
-
* [[Hubdub]]
+
 +
=== Judgment biases ===
 +
== Resources ==
== Resources ==
-
=== Academic Articles ===
+
=== References & Bookmarks ===
-
* [http://siepr.stanford.edu/papers/pdf/03-25.pdf Wolfers,J. and E. Zitzewitz (2004). "Prediction Markets." SIEPR Discussion Paper No. 03-25] is a comprehensive paper on Prediction Markets that addresses various perspectives of the topic
+
* [http://del.icio.us/doritgsb/predictionmarkets del.icio.us Prediction Markets tag] - various bookmarks on the topic.
 +
* [http://www.midasoracle.org/predictions/exchanges/ Prediction Exchanges] - a list from Midas Oracle .Org
 +
* [http://www.forecastingprinciples.com/PM/ Special Interest Group on Prediction Markets] - International Institute of Forecasters, IIF
 +
* [http://www.crowdworx.de/de/news/crowdworx-library CrowdWorkx]
=== Books ===
=== Books ===
-
* '''Surowiecki, J. (2005)''' - ''The wisdom of crowds: why the many are smarter than the few''. This book is starter into the world of the wisdom of the crowd and specifically Prediction Markets.
+
* '''Abramowicz, M.''' (2008). ''Predictocracy: Market Mechanisms for Public and Private Decision Making.'' Yale University Press.
-
 
+
* '''Surowiecki, J.''' (2004). ''The Wisdom of Crowds: Why the Many Are Smarter Than the Few and How Collective Wisdom Shapes Business, Economies, Societies and Nations.'' Doubleday.
-
=== Wiki, Blogs & Bookmarks ===
+
-
* [http://del.icio.us/doritgsb/prediction-markets Dorit's del.icio.us Prediction Markets bookmarks] - maintains a list of related bookmarks, tagged with sub-categories.
+
=== Wikis & Blogs ===
-
* [http://en.wikipedia.org/wiki/Prediction_market Wikipedia on Prediction Markets] - has good articles and references. It is being updated on a frequent basis.
+
-
* [http://www.midasoracle.org/ Midas Oracle] - the most extensive information source on Prediction Markets.
+
-
* [http://www.forecastingprinciples.com/PM/ Special Interest Group on Prediction Markets] - a good source for various types of resources in this domain.
+
-
=== Prediction Markets in the media ===
+
* [http://en.wikipedia.org/wiki/Prediction_market Wikipedia on Prediction Markets]
-
* [http://www.time.com/time/magazine/article/0,9171,660965,00.html Time Magazine – The end of Management?] - dated mid 2004, it is one of the first articles that brings the concept of Corporate Prediction Markets to the public awareness and lists some of the pioneers in the area.
+
* [http://crowdcast.com/wordpress/ Crowdcast Corporate blog]
 +
* [http://blog.mercury-rac.com/ Mercury's Blog] - Jed Christiansen's blog
 +
* [http://blog.inklingmarkets.com/ Inkling Corporate blog]
 +
* [http://www.midasoracle.org/ Midas Oracle] - Chris F. Masse's blog
 +
* [http://blog.oddhead.com/ Oddhead Blog] David Pennock's blog]
 +
* [http://www.overcomingbias.com/ overcomingbias] - Robin Hanson's blog
 +
* [http://torontopm.wordpress.com/ Toronto Prediction Market Blog] - Paul Hewitt's blog
=== Research Data ===
=== Research Data ===
-
Some public exchanges open their data for academic exchange
+
Public exchanges that open their data for academic research
* In [http://www.biz.uiowa.edu/iem/archive/index.html Iowa Electronic Markets (IEM)] you can find historical data and research papers.
* In [http://www.biz.uiowa.edu/iem/archive/index.html Iowa Electronic Markets (IEM)] you can find historical data and research papers.
* In [http://www.tradesports.com/data/ TradeSports] you can gain access to historical data (2004-2006), mostly sports, some political and entertainment.
* In [http://www.tradesports.com/data/ TradeSports] you can gain access to historical data (2004-2006), mostly sports, some political and entertainment.

Current revision as of 20:11, 30 December 2011

Personal tools