Glen E. P. Ropella wrote: > But you're focusing on extrapolation, right? It strikes me that you're > not talking about heuristic (a.k.a. explanatory) models but about > aggregative extrapolation. More like looking for exploitable, repeatable cause/effect inefficiencies in an ocean of activity. My objection to ABM in this kind of context is that the individual `strategies' may range from super smart to virtually random. The data stream can tell you when someone made money, and there are lots of these examples each having particular state of the book (for that equity and maybe a bunch of related ones) and real world when it happened. Correlating and deconstructing the two into a taxonomy seems much more direct than tweaking agent rules which are just educated guesses to begin with. One could think about skipping the deconstruction/taxonomy stuff for an automated learning procedure (e.g. hidden markov model) that reproduced the individual firm trading patterns given certain signals, but since money is on the line perhaps that's not a great plan..
============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org
