The research in stock market dynamics used to be a key task for financial experts and mathematicians, but nowadays engineers also play an important role in this area. In the thesis, I will establish and test a hypothesis of agent-based trading through the newly formed connection between the model of stock market and Agent-Based Models.
Traders in the market continuously try to achieve a better performance than the risk-adjusted profit, or with a more common phrase, beat the market, which leads to the incorporation of information into the actual price at a large scale. They form their market environment themselves, which they also have to adapt to, in order to own a positive balance at the end of the day.
Trading participants can be viewed as abstract agents, who trade adaptively in a stock market, which is the environment of this agent-system. This scenario will be realized with the simulation of the agent-system, stock trading and evolutionary algorithms for optimalization.
Considering the complexity of the problem domain, the solution had to be designed towards scalability, with the rational expectation of time-saving properties of the model-testing phase. The software was implemented in Java, backed up by a MySQL database. To validate the hypothesis, I ran two test-cases with the final conclusion of the model’s high information dependency.