Abstract
Since its inception, the Efficient Market Hypothesis (EMH) has faced persistent challenges, as numerous anomalies - such as volatility clustering, excessive trading volumes, and herding behaviour - exposed gaps between theoretical predictions and actual market dynamics. In response, economists developed alternative frameworks that relaxed EMH’s strict assumptions, distinguishing between different types of investors (e.g., "chartists" and "fundamentalists") and incorporating bounded rationality, learning, and adaptation. This line of research gave rise to agent-based models, which conceptualize financial markets as adaptive ecosystems and rely on simulations to capture investor interactions and the evolution of trading strategies. This paper reviews central modelling choices - such as the definition of investor heterogeneity, the specification of preferences, the mechanisms of price formation, and the processes of strategy selection - and discusses their implications for balancing realism with the complexity of calibration.