On Markets Structures and some thoughts about Information

1
On Markets Structures and some thoughts about
Information Working Index.

• Alessandro Cappellini
• February 3rd 2006
• cappellini_at_econ.unito.it

2
Index

• Introduction
• Humans Experiments and Policy Rules
• MAD
• Investor confidence Survey
• Information
• Extracting useful information from a simulated
  limit-order book
• MIB Memetic Information Behaviour

3
Introduction

• The main aim of this work, is to line the
  experiences collected from the master thesis, a
  stock market experiment, during the Ph.D.
  program. The knowledge obtained is large,
  starting from practice in simulation and coding,
  enlarging literature familiarity, cooperation
  with other discipline with a general enrichment
  and awareness.
• The thesis is basically a collection of
  heterogeneous and heterodox works on financial
  markets.

4
Humans Experiments and Policy Rules

• The chapter is mainly based on the paper
  presented in ESA conference held in Alessandria,
  about market manipulation.
• There will be a description of regulations
  against manipulation (EU directive) and a
  description of some behaviours in SumWEB
  experiment. Taking idea from these facts, I,
  Enrico Scalas (University of Eastern Piedmont),
  Michael Kirchler and Juergen Huber (University of
  Innsbruck) started in regulation
  analysis/comparisons on different stock markets.
  Early results could be ready in spring 2006.
• This idea became

5
MAD - Market Microstructures and Architectures
Database

• Stock markets are complex systems. Many
  autonomous agents interact through a mechanism
  know as "continuous double auction". They are
  sensitive to exogenous as well as endogenous
  influences, that can cause domino effects or
  avalanches, known as bubbles and crashes.
• Not only two-agent interactions are relevant, but
  also many-agent interactions cannot be neglected.
  
• Agents playing in these markets are intelligent
  and try to manipulate the outcome. Recent results
  on asymmetric information, behavioural and
  experimental economics clearly show that the
  neoclassical paradigm of walsarian markets is not
  sufficient to describe every feature of
  real-world stock exchanges.
• With this Database, our aim is to provide
  scholars an updated set of information on market
  architectures and rules. This should be helpful
  both for further studies on agent-based computer
  simulations (Boer-Sorban, de Bruin and Kaymak,
  2005) and on analytical models (Hommes 2005).
• http//www.complexity-research.org/mad/

6
Investigate traders behaviours surveys and
experiments

• During the months of September-December 2005 two
  sets of questionnaires will be filled by a group
  of skilled people and by students.
• During January-March a survey will be published
  on the internet.
• Results will be collected and analysed with
  Enrico Rubaltelli (University of Modena and
  Reggio Emilia) and A.I.Fin.C. (Italian
  Association of Behavioural Finance).
• In late spring an experiment

7
Investor confidence Survey

• Two paper and pencil surveys (150 student 80
  professionals)
• One Internet survey (three month survey,
  actually 101 complete answers on 814 contacts,
  quasi halving function)
• Main Features
• Bias
• Affect
• Prospect
• Semantic Values
• Survey on methods/knowledge of Traders
• Confidence (comparison with Shillers)
• Risk Propensity

8
Information

• External
• publicly broadcast
• News (media)
• Data providers (semi-public?)
• Analysis
• privately spread
• Gossip
• Word-of-mouth
• Insider trading
• Internal
• Technical (collected from the market)
• Fundamental (balance sheets)

9
Two Papers

• Both will be presented at
• Complexity 2006 Workshop
• May 17th-21th 2006, Aix-en-Provence, France
• Special session on Information in Financial
  Markets

10
Extracting useful information from a simulated
limit-order book

• Co-Author Gilles Daniel
• book imbalance can be exploited to forecast
  future price changes and outperform the simple
  buy-and-hold strategy
• Statistical analysis (MatLab)
• Smart Agent introduction in NatLab (Lev
  Muchnicks platform) (and in SUM?)

11
MIB Memetic Information Behaviour

• We would like to describe a social information
  process based on both spreading and learning
  algorithm in a stock market.
• We propose an alternative decision making process
  mainly based on communication, and opinion
  building without any consideration on financial
  stylized facts. Those opinions will be treated
  according to the memetic paradigm.
• The information ows among agents is establish on
  a fixed social network.
• The memetic agents, by informal chatting into
  the group, increase their knowledge, and after
  pondering, raise believes used both acting in the
  market and spreading to others. To think,
  agents exploit services supplied by an original
  focused algorithm.
• Our simulation model is the well know SUM-SumWEB
  project, an artificial stock market with a
  continuous double auction mechanism inspired by
  Milan stock exchange regulation.

12
Schema
Book
13
Memes examples
BUY
Verdi 10.5
BUY Rossi
Trust 2
Merry Lunch Analist Tim Hooker BUY Gialli 15,5
SELL Blu Reference 4
14
Implementing

• Co-Author Gianluigi Ferraris
• Adapt a GA
• Develop a MA
• GA against MA literature review
• TOTA Trader Oriented by Talked Actions