Crowd-Anticrowd Theory of Multi-Agent Minority Games
arXiv:cond-mat/0212088
Abstract
We present a formal treatment of the Crowd-Anticrowd theory of Minority Games played by a population of competing agents. This theory is built around a description of the crowding which arises within the game's strategy space. Earlier works have shown that this theory provides a simple, yet quantitatively accurate explanation of the time-averaged behavior of these multi-agent games. We also discuss the extent to which the Crowd-Anticrowd approach provides a useful tool for analyzing a wider class of Complex Systems.
30 pages + 15 figs (postscript). Minor typos corrected