In late 2003 and early 2004, I put some effort into getting the software used in the Nasdaq research above up and running again, and to attempt to reproduce some of the results reported in these papers. The original results were developed over more than 2 years of ongoing paid research, and were validated by Nasdaq's management and research teams. While this certainly wasn't as rigorous a process as one might imagine for a presitigous peer-reviewed academic journal, for example, the research was of high quality, and quite unique in its attempt to model by simulation very complex markets and answer difficult questions about them (and therefore unpublishable in any such journal!). The end result was that Nasdaq viewed the research highly enough to continue and expand funding of it for a number of years.
Over the few days I have been able to devote to this in my spare time, I have so far managed to get the software from 2001 up and running in the newer versions of Java, and begin to set up a reasonably diverse, realistic population of agents. I am currently trying to calibrate and validate this population for a large number of different simulation runs at varying tick sizes, and with varying amounts of parasitic strategies in order to gather sufficient data to statistically reproduce our previous results.
Others have shown some interest in this recent work, and we are willing, under appropriate conditions, to share the simulation software with other researchers. I would certainly like to encourage that!
Please do get in touch:vince.darley@eurobios.com , or return to the main page.