Re: [Swarm-Modelling] [Fwd: Re: [ABMs in finance]

[Marcus G. Daniels]

Pietro Terna wrote:
> > Some patterns seem to emerge just from the trading rules.
> *********
Just to be clear, in this case the `trading rules' were not agent 
behaviors.  The agents behaviors were assumed to be random, and the 
market mechanism, an order book, itself was modeled in a realistic way.  
This type of analytical model explained around 70% of the variance in 
the spread of book prices as a function of a set time-aggregated 
measurements from historical London Stock Exchange data -- different 
subsets of millions of observations per different stocks. Delving into 
the ecology of agents wasn't necessary.

This is in contrast to the approach of building up a simulation model 
like ASM and expecting to reproduce global phenomena from different 
ensembles and parameterizations of agent behaviors.  I don't doubt there 
are situations where it is useful to do employ the synthetic ASM 
approach, but if an explanations seems simple from theory or data 
mining, perhaps it just is?