[Swarm-Modelling] agent based prediction market simulation

[Sjors van Fulpen]

Hi there,

I’m a master student at the Rotterdam School of Management.
I am doing my master’s in Business Information Management,
and am currently writing my final thesis.

My thesis is focused on prediction markets. I am interested in the
effects of trader composition on the accuracy of the outcome of the
prediction market. In my research I wish to investigate what
combination of informed and uninformed traders (or experts and
non-experts) results in the most accurate prediction results in
relatively small markets (6-30 traders). I focus on small markets,
since these will be most applicable to actual businesses.

To gather data, I wish to run simulations using Zero Intelligence
agents (or preferably Zero Intelligence Plus agents) in which the
traders (agents) are assigned certain probabilities of the outcomes to
reflect their level of knowledge (similar to Plott, 2000). I want to
distinguish only two levels of ‘informedness’: informed and
uninformed, wherein the uninformed traders share equal probabilities
for each for the contracts traded in the market.

Not being able to program myself, I have been searching for software
that allows me to run my proposed simulation, thus far unsuccessful.
My question is whether someone has an idea as to where I can find such
simulation
software useful for my research, as you probably have more experience
in the field of agent based simulation..
I appreciate all types of input and thank you in advance for your time and help.

Sincerely,

Sjors van Fulpen

E-mail: address@hidden

Reference:
C.R. Plott, Markets as information gathering tools. Southern Economic
Journal 67 1 (2000), pp. 2–15