Mechanical vs Neural trading systems
- Thread starter pandabear
- Start date
I'm sorry if I've not made myself plain. More than once on these fora, opinions (frequently stated as established fact) have been offered that neural networks (or whatever empirical system) cannot be used to trade productively. While it may be true that neural networks have not been used profitably to that end, my only point is that this alone does not assure that they will not.
Others have asserted, in turn, that there are examples of such systems. I cannot vouch for that one way or the other: I am only saying that the mere absence (if such is the case) of such examples is not sufficient evidence to support the extreme claims made against neural networks.
So, how is it that learning systems best people in other contexts (credit scoring, actuary, etc.)?
? ? ?
Fraud detection systems and actuary models do this all the time.
I wonder if any of these posters have any real expertise in computers, not even mentioning artificial intelligence ... At the current stage of science and technology, a computer can't be smarter than the algorithm that was programmed with by a human. Hence it can't be smarter than a human. It may be more consistent, faster, obviously unemotional.
On the other hand where is this discussion headed? It seems only a waste of time with no potential value, between people who don't know what they're talking about and have nothing better to do. No disrespect intended
EDIT: But this is ET: most posters don't know what they're talking about, and have nothing better to do. Fortunately there are a few remarkable exceptions most posters challenge, as expected.
On the other hand where is this discussion headed? It seems only a waste of time with no potential value, between people who don't know what they're talking about and have nothing better to do. No disrespect intended
EDIT: But this is ET: most posters don't know what they're talking about, and have nothing better to do. Fortunately there are a few remarkable exceptions most posters challenge, as expected.
Quote from cnms2:
I wonder if any of these posters have some real expertise in computers, not even mentioning artificial intelligence ... At the current stage of science and technology, a computer can't be smarter than the algorithm that was programmed with by a human. Hence it can't be smarter than a human. It may be more consistent, faster, obviously unemotional.
On the other hand where is this discussion headed? It seems only a waste of time with no potential value, between people who don't know what they're talking about and have nothing better to do. No disrespect intended
I agree-it would be nice to get a substantive discussion going.
Here is a link to Futures Mag where Murray Ruggiero analyzes the use of neural nets in trading. I am not sure if the artiles are accessible to non-subscribers (I suspect not).
http://futuresmag.com/cms/Futures/M.../aug06ruggieroTechT?searchfor=neural networks
The article is from 8/2006 issue and it is the third of three articles on Neural Nets. In this article an actual system is shown with analysis of how to use a neural net to improve the system results without making the neural net the main part of the system.
I will run the tests described and post the results to show how the system has performed since the article was published.
I am also attaching the reference to the three articles from Futures as a reference (on the gif attached).
My point: instead of arguing on something you don't know about, better instruct yourself. You can use Google ... A quick query gives you, for example: Neural Networks: Where are they Now?
Let's see how well this logic works in another arena: A machine can't be stronger than its mechanical design that was constructed by a human. Hence it can't be stronger than a human. No, that doesn't make much sense, does it?
The simple fact is that computer systems have been built which are measurably much more effective than their human creators at solving many problems.
...or you could actually address my question?