Self-referential behaviour, overreaction and conventions in financial markets

We study a generic model for self-referential behaviour in financial markets where agents build strategies using correlations estimated using the past history itself, between certain quantitative information and the price. The impact of these strategies on the price modify the observed correlations and create a feedback loop that can destabilize the market from efficient behaviour. For large enough feedbacks non-trivial correlations spontaneously set in and the market switches between two long lived states, that we call conventions. This mechanism leads to overreaction and excess volatility. We provide empirical evidence for the existence of such long lasting anomalous correlations in real markets.