Agent-based artificial financial markets are bottom-up models of financial markets which explore the mapping from the micro level of individual investor behavior into the macro level of aggregate market phenomena. It has been recently recognized in the literature that such (agentbased) models are potentially a very suitable tool to generate or test various behavioral hypotheses. One of the psychological biases that received a lot of attention in financial studies, both mainstream and behavioral, is the phenomena of investor overconfidence. This paper studies overconfident investors in the agent-based artificial financial market based on the Levy, Levy, Solomon (2000) model. Overconfidence is modeled as miscalibration, i.e. as underestimated risk of expected returns. We find that overconfident investors create less frequent but more extreme bubbles and crashes when compared to the unbiased efficient market believers of the original model. When investors are modeled to exhibit a biased self-attribution, they quickly move to the state of high overconfidence and remain there. With an unbiased self-attribution, on the other hand, investor overconfidence varies greatly, but around a moderate level of overconfidence.

Agent based, Aggregate market, Artificial financial markets, Artificial intelligence, Commerce, Expected return, Finance, Financial data processing, Financial market, Micro level, Miscalibration, Original model,
Erasmus School of Economics

Lovric, M, Kaymak, U, & Spronk, J. (2009). Overconfident investors in the LLS agent-based artificial financial market. doi:10.1109/CIFER.2009.4937503