Using a few nonlinear econometric tools, this paper examines some time-series properties of GP-based artificial markets. We find that GP-based artificial markets are able to replicate several stylized features well documented in time series generated by the GP-based artificial markets are consistent with the efficient market hypothesis in the linear sense. Furthermore, the emergence of these stylized features may be caused by some institutional factors, such as position limits and transaction factors. By introducing the complexity of evolved GP-trees, a bottom-up analysis of the impact of transaction taxes on GP-based artificial markets is also provided.
展开▼