Neural Network Models & Inputs

Hi, I want model stock prices using neural networks with functions (such as market indicators) I can optimize. Can you tell me where I can research how to do this?

E.g. of Market Indicators I will use - SMA, Williams' %R, St. Deviation, MACD You could make an e.g. to show me if you like. I wouldn't mind that, lol!

[Greg Heath]

What do you have in the way of inputs and targets (i.e, desired outputs).

[3nore]

I don't have any particular desired inputs or desired outputs. I just want to know how it is done.

Answer 1

[Ken]

You can use a SOM to figure out which parameters are correlated with your output. Then take those parameters to create a NN.

I.E. Create a SOM with SMA, Williams' %R, St. Dev, MACD, and Price. Check the weight planes to see which match with price. Then use those as inputs to a NN (perhaps pattern recognition) with Price as the target.

You will need a good amount of data to do this accuratly.

[Greg Heath]

It is usually better to output change in price instead of price itself.

[3nore]

Hi Ken, Can you give me a code to show me exactly how you do this? I am new to programming and matlab