|
PureBytes Links
Trading Reference Links
|
This is what people try to do by training neural nets with genetic
algorithms or by creating rules-based systems using genetic algorithms.
The problem with neural nets are their tendency to fit to noise in the
inputs. The problems with rules-based systems is that you get too many
rules in the system and overfit that way.
It's not impossible but they require a lot of work. Like any other method
you may use to approach the markets, you must understand the limits and
pitfalls of these tools.
Kent
----- Original Message -----
From: "LPrice" <lprice1023@xxxxxxxxxxxxx>
To: <omega-list@xxxxxxxxxx>
Sent: Thursday, August 08, 2002 11:15 PM
Subject: Off-Topic
I am curious if anyone on the list has experimented with trading bots?
I am particularly interested in bots that could learn and could also try
many different strategies at the same time. Could a trading bot be
trained to emulate a successful day trader for example? It would seem
that a trainable bot could ultimately be a successful trading vehicle.
Sincerely,
Lawrence Price
|