Trading

Trading, the market mechanism, is the second most important design question which faces the researcher when building an agent-based market. Caution and efficiency in designing the actual details of trading in the market models is essential.

The majority of agent-based markets follow three methods of solving this design problem. The simplest way is to assume that price movements respond to the excesses of demand and supply. Thus, if there is an excess of demand, the price will rise. Conversely, if there is an excess of supply, the price falls. This method is very fast to compute but comes with some cost. An earlier version of the Santa Fe Artificial stock market used this trading method. In addition, there are some agent-based market models that use the same method for the trading mechanism including Farmer (1998), Cont and Bouchaud (2000), Chen and Yeh (2001), Markose et al. (2003), and Martinez-Jaramillo and Tsang (2009). Farmer and Joshi (2002) the author used a similar trading mechanism but the excess demand or supply for assets is filled by a market marker.

Alternatively, there is a model of the market where a local equilibrium price can easily be found. This method requires us to define a market with more economic structure so that a certain tem­poral equilibrium price can be found (Martinez- Jaramillo & Tsang, 2009). Examples of market models using this type of trading mechanism are Arthur et al. (1997), Levy et al. (1994).

The last trading method is to model the ac­tual trading mechanisms found in markets. This method implies the implantation of a realistic trading mechanism as used in a real market. It is applied frequently for modelling high-frequency data. Examples of models with such a type of mechanism are Gode and Sunder (1991), Yang (2002), and Chan (1999).

3.3.