ASM question/observations

[Ken Gosier]

Questions/observations on ASM (esp. for Paul Johnson, but
also for other interested persons. ;-)

I'm studying the behavior of the 20000602 snapshot of ASM
from sourceforge. Specifically, I changed the specialist
type from 2 to 1 in ASMModelSwarm.m. This changes the
specialist from 'eta' to 'slope.' The behavior with eta is
just price impact: the specialist measures the order
imbalance, and applies a price impact coefficient eta to
the imbalance, and uses that to change the price. The slope
specialist instead makes a number of trial prices, and sets
the price where order imbalance is minimized.

I find quite different behavior for the slope specialist.
Whereas price is relatively stable for the eta specialist,
price is nowhere near stable for slope. With many different
random seeds, the price always shoots down to zero.
(Sometimes it shoots off the top of the chart, stays there
for a while, then shoots to zero.) After a while at zero,
it picks itself up and starts moving again, although at a
low level. Usually stays below 25, though it can get above
25 sometimes.

There is a minprice parameter set at a default of 0.001. I
find the same behavior of hitting zero, staying there, then
moving at a low level, whether or not I leave minprice at
0.001 or change it to 0.

So, these are just observations. I'm curious if ASM needs
some other parameters to be changed in conjunction with
sptype, in order for the slope specialist to produce
reasonable behavior.

=====
Ken Gosier
address@hidden

__________________________________________________
Get personalized email addresses from Yahoo! Mail - only $35 
a year!  http://personal.mail.yahoo.com/


                  ==================================
   Swarm-Modelling is for discussion of Simulation and Modelling techniques
   esp. using Swarm.  For list administration needs (esp. [un]subscribing),
   please send a message to <address@hidden> with "help" in the
   body of the message.
                  ==================================