Neural networks really have nothing directly to do with technical or
fundamental analysis. What neural networks do in mathematical terms is
approximate a function based on a series of training inputs and
outputs. Then, you can run the neural network with the same types of
inputs you were training with, and it will attempt to approximate the
real result.
Applying neural network analysis to stock market is
a pretty broad topic
in itself: you generally are attempting to pick something like % change
5 days out, and you can feed the neural network anything: technical
analysis functions, stars alignment, fundamental analysis, Google
Zeitgeist, or whatever you can think of. If there's a relationship, the
neural network will find it. Of course, it's much more difficult to get
the neural network to perform to anywhere near the same level out of
sample as it does in sample (which typically has near perfect results).
You can view neural networks as a very sophisticated method of parameter
optimisation, and it comes with all of the caveats that optimisation has.
Regards,
Matt
k j wrote:
> Well Matt what are you waiting for... priorities?
> Ahhh just kidding
>
> Matt Hi, I am not a programmer but I am interested... When one says
> " NEURAL ANALYSIS " what are we talking about. Relative to just
pure
> analysis or TA of a stock how does this possibly differ. When
> googleing the term I get all the new Software and no real definition
> as to what the term means.
> Soooo in your words....
>