Cognitive-Agent-Based Modeling of a Financial Market

This paper describes our experience in building an evolutionary system for agent-based modeling of a financial market. The system uses a type of BDI agents, which are deliberative agents with a mental state defined by a belief base and by a set of desire-generation rules. Beliefs are fuzzy and trust in information sources is taken into account. At any moment, an agent generates a set of desires and selects a consistent subset thereof, whose elements are adopted as goals, to be achieved by executing actions. The market structure is realistic, albeit simplified: a single emph{asset} is assumed to exist, that is traded on the market in indistinguishable, standardized contracts against payment of emph{money}. The price of trades is continuously determined by means of a single-price auction, which crosses buy and sell orders to maximize market liquidity. Every agent has, at any moment, an available amount of money and an inventory of asset contracts, which, valued at market price and added to the available money, yields the agent´s emph{net asset value} (NAV). To determine their economical behavior, the agents have access to a set of technical indicators made available by the market, which constitute the sources of their beliefs, and to their current balance, which constitutes their knowledge. The agents participating in the market evolve by means of an evolutionary algorithm: they undergo selection based on their NAV, replication with random mutations, and recombination. When an agent replicates, its properties are divided among the offspring. This evolutionary process favors the reproduction of profitable traders, allowing the emergence of ``sensible´´ behaviors. The market thus simulated has been analyzed both in quantitative and in qualitative terms. The analysis has demonstrated that the model exhibits several traits that are typical of real-world financial markets.