Aggregation and manipulation in prediction markets: effects of trading mechanism and information distribution.
ABSTRACT We conduct laboratory experiments on variants of market scoring rule prediction markets, under different information distribution patterns, in order to evaluate the efficiency and speed of information aggregation, as well as test recent theoretical results on manipulative behavior by traders. We find that markets structured to have a fixed sequence of trades exhibit greater accuracy of information aggregation than the typical form that has unstructured trade. In comparing two commonly used mechanisms, we find no significant difference between the performance of the direct probability-report form and the indirect security-trading form of the market scoring rule. In the case of the markets with a structured order, we find evidence supporting the theoretical prediction that information aggregation is slower when information is complementary. In structured markets, the theoretical prediction that there will be more delayed trading in complementary markets is supported, but we find no support for the prediction that there will be more bluffing in complementary markets. However, the theoretical predictions are not borne out in the unstructured markets.
- SourceAvailable from: Jan-Paul Lüdtke
Conference Proceeding: THE IMPACT OF OVERCONFIDENCE ON THE EVALUATION OF INNOVATIONS[show abstract] [hide abstract]
ABSTRACT: The evaluation of new products and innovative technologies is a core task of innovation management. Without valid innovation evaluation, companies fail to identify promising endeavors for future business success. However, the evaluation of innovations is characterized by environments where uncertainty is high and strategic context is poorly understood. A relatively new strand of research promisingly taps heterogeneous individual expectations or the wisdom of crowds to generate valid predictions in such environments. But there is strong evidence, that aggregating subjective expectations can yield biased results as individuals often draw highly biased conclusions from available information. Among important individual evaluators of innovations such as entrepreneurs, inventors and business decision makers, overconfidence is the most prominent and important bias. Overconfidence leads individuals to systematically overestimate their evaluation capabilities. We report an experiment that explores the impact of overconfidence on individual behavior in innovation evaluation tasks. We focus a promising method to tap crowd wisdom known as information-or prediction markets, which aggregate individual evaluations of new product success or innovative idea potential via market mechanisms. We induce overconfidence experimentally and study how overconfident individuals interact on these platforms. Our results show, that overconfident individuals turn evaluations into actions earlier, they pursue their predictions with more vigor, are more likely to act according to initial evaluations by disregarding contradictory information and less willing to change a priori predictions after innovation evaluation tasks finish.19th IPDMC; 06/2012
Article: Designing Markets for Prediction.AI Magazine. 01/2010; 31:42-52.
Conference Proceeding: A Multi-agent Prediction Market Based on Boolean Network Evolution[show abstract] [hide abstract]
ABSTRACT: Prediction markets have been shown to be a useful tool in forecasting the outcome of future events by aggregating public opinion about the events' outcome. Previous research on prediction markets has mostly analyzed the prediction markets by building complex analytical models. In this paper, we posit that simpler yet powerful Boolean rules can be used to adequately describe the operations of a prediction market. We have used a multi-agent based prediction market where Boolean network based rules are used to capture the evolution of the beliefs of the market's participants, as well as to aggregate the prices in the market. We show that despite the simplification of the traders' beliefs in the prediction market into Boolean states, the aggregated market price calculated using our BN model is strongly correlated with the price calculated by a commonly used aggregation strategy in existing prediction markets called the Logarithmic Market Scoring Rule (LMSR). We also empirically show that our Boolean network-based prediction market can stabilize market prices under the presence of untruthful belief revelation by the traders.Web Intelligence and Intelligent Agent Technology (WI-IAT), 2011 IEEE/WIC/ACM International Conference on; 09/2011
Aggregation and Manipulation in Prediction Markets:
Effects of Trading Mechanism and Information Distribution
School of Information
University of Michigan
Ann Arbor, MI 48109, USA
School of Information
University of Michigan
Ann Arbor, MI 48109, USA
We conduct laboratory experiments on variants of market
scoring rule prediction markets, under different informa-
tion distribution patterns, in order to evaluate the efficiency
and speed of information aggregation, as well as test re-
cent theoretical results on manipulative behavior by traders.
We find that markets structured to have a fixed sequence
of trades exhibit greater accuracy of information aggrega-
tion than the typical form that has unstructured trades.
Prior theoretical predictions of differing strategic behavior
under complementary information distributions and substi-
tute information distributions are confirmed when the trad-
ing order is structured, but not in markets with an un-
structured trading order. In the case of the markets with
a structured order, we find that the information aggregation
is consequently slower when information is complementary,
as traders more frequently engage in bluffing and delaying
strategies. In comparing two commonly used mechanisms,
we find no significant difference between the performance of
the direct probability-report form and the indirect security-
trading form of the market scoring rule.
Categories and Subject Descriptors
J.4 [Computer Applications]: Social and Behavioral Sci-
prediction markets, experiments, market scoring rule
Prediction markets are markets designed to aggregate traders’
information and to forecast future events. Recently, a new
market form for prediction markets, the market scoring rule
∗This work was supported by the National Science Founda-
tion grant CCF-0728768. We thank our research assistants
M. Bombyk and R. Jimenez. A full version of this paper is
posted at http://www.umich.edu/~rsami/papers/JS.pdf.
Copyright is held by the author/owner(s).
EC’10, June 7–11, 2010, Cambridge, Massachusetts, USA.
(MSR) , has become popular.
human-subject laboratory experiments to study the speed
and efficiency of information aggregation in MSR markets,
while varying the mechanism form, constraints on trade tim-
ing, and information distribution pattern.
The first dimension of variation is in comparing two com-
monly used forms of the MSR mechanism: a direct mech-
anism in which traders report their beliefs as probabilities,
and an indirect mechanism in which traders reveal their be-
liefs through buying and selling securities. There is an active
debate about which interface is more effective; our labora-
tory experiments provide insight into this question.
MSR markets have a myopic honesty property: A trader
trading only once maximizes her expected profit by report-
ing her true beliefs .However, for traders using non-
myopic strategies over multiple trades, the theoretical re-
sults show a sharp distinction based on the pattern of infor-
mation distribution. Roughly, when signals are substitutes
honest reporting of beliefs is optimal even in a non-myopic
sense; when signals are complements honest reporting is not
a sequential equilibrium . This motivates the second di-
mension along which we vary our experimental design: We
study market performance under a complementary signal
structure, and under a substitute signal structure.
The third variation we study is in providing the structure
of a fixed sequence of trade opportunities, or an unstructured
market in which traders choose when to trade.
We conducted two-trader market trading experiments for
each of the 8 treatments generated by a factorial explo-
ration, with the following results: First, we find that struc-
tured markets aggregate information more efficiently than
unstructured markets.Second, in the first experimental
comparison between the direct and indirect MSR forms, we
find no significant difference in performance. Third, in test-
ing the theoretical results on the effect of information dis-
tribution on manipulative strategies, we find that they are
borne out for the structured market but not for the unstruc-
tured markets. This suggests that the timing of trades is an
important feature to include in future theoretical research.
In this paper, we use
 Y. Chen, S. Dimitrov, R. Sami, D. Reeves, D. Pennock,
R. Hanson, L. Fortnow, and R. Gonen. Gaming
prediction markets: Equilibrium strategies with a
market maker. Algorithmica, 2009.
 R. Hanson. Combinatorial information market design.
Information Systems Frontiers, 5(1):107–119, 2003.