Data normalization for stock values.

I have some doubts about data normalization for stock prediction using a NARX network.

- Which is the best way to normalize stocks data?. Is there built-in functions to perform this tasks for this kind of data?

- Why data in the Matlab samples is not normalized?

- Let's suppose I have a time serie with 200 values. Is there any difference in entering the data in the NN in the form of 1x200 versus 200x1? Where do I set that?

Thanks for your answers. Rgds.

Answer 1

[Marc]

Really Francisco?

Normalization for stock prediction? If any of us had that answer we would all be rich.

If Matlab had a true solution to stock prediction, do you think we would see a Matlab 2014a?

Not sure about your last question with the "NN". If there is a difference between column and row, then a simple transpose by using " ' " should fix that or check that.

Other than helping you get rich beyond your wildest dreams, is there something that we can really help you with?

I expect an ACCEPTED ANSWER on this.

[Fran Mat]

Let's put thing in order: The question is not about "Solution for Stock Pred."....it "Data Normalization for Stock Pred."....that's it....is it clear now?, or is it required more info?

Better would be an answer about this one: "- Why data in the Matlab samples is not normalized?". I see some sample data is not between (0 and 1) or (-1 and 1).(?)

Last but not least: Nothing to do with money...it is just a simple proof of concepts for a school demo.

To accept an answer I would need some contribution to the questions.