The paper examines the properties of financial market dynamics, under different trading protocols. We start with an empirical analysis of the statistical properties of daily data from the world’s major Stock Exchanges, comparing the behavior of different market phases characterized by different trading protocols. The evidence lends support to the importance of investigating the outcome of alternative market mechanisms. Motivated by this finding, we present an agent-based model allowing the consistent treatment of agents’ behavior under three different trading set-ups, namely a Walrasian auction, a batch auction and an ‘order-book’ mechanism. The results highlight the importance of the institutional setting in shaping the dynamics of the market but also suggest that the latter can become the outcome of a complicated interaction between the trading protocol and the ecology of traders behaviors. In particular, we show that market architectures bear a central influence upon the time series properties of market dynamics. Conversely, the revealed allocative efficiency of different market settings is strongly influenced by the trading behavior of agents.
