Competitive Advantage for Multiple-Memory Strategies in an Artificial Market

ABSTRACT We consider a simple binar y market model containing N competitive agents. The novel feature of our model isthat it incorporates the tendency shown by traders to look for patterns in past price movements over multipletime scales, i.e. multiple memory-lengths . In the regime where these memory-lengths are all small, the averagewinnings per agent exceed those obtained for either (1) a pure population where all agents have equal memory-length, or (2) a mixed population comprising sub-populations of equal-memory agents with each sub-populationhaving a dierent memory-length. Agents who consistently play strategies of a given memory-length, are foundto win more on average switching between strategies with dierent memory lengths incurs an eective penalty,while switching between strategies of equal memory does not. Agents employing short-memory strategies canoutperform agents using long-memory s trategies, even in the regime where an equal-memory system would havefavored the use of long-memory strategies. Using the many-body Crowd-Anticrowd theory, we obtain analyticexpressions which are in good agreement w ith the observed numerical results. In the context of “nancial markets,our results suggest that multiple-memory agents have a better chance of identifying price patterns of unknownlength and hence will typically have higher winnings.Keywords: econophysics, multi-agent games, limited resources, prediction