MSc Math Defence: Social Network Effects on Investment Behaviour in the Stock Market

CANDIDATE:  MICHAEL GLAZER

ABSTRACT:

We develop an agent-based model to examine the behaviour of brokers trading in a financial marketplace and determine whether social communication between brokers influence their overall trading decisions. The agents' decision-making process is determined by their total utility, which is broken down into objective and subjective utility; an agent's perceived value of the stock and the trading decision suggested from one's peers. To accurately represent an agent's trading decisions, sensitivity analysis is conducted to determine the required number of average trial runs within each simulation. Random and preferential attachment networks are implemented into simulations with 50, 100, 150 and 200 agents and the trading decisions over the course of 30 days are analyzed to examine if different social networks can influence a trader's decision. Affected total utility (ATU) represents the agent's decisions which are influenced by their peers. The frequency of affected total utility decreased significantly with the implementation of weak links in the social network. The occurrence of affected total utility remains stable for 50, 100 and 150 agents when the social network is a preferential attachment network.