To be really honest with you this is really a difficult job. Finding answers to questions such as What constitutes the stock market ? when does share price rise or fall? what are the factors that affect stock market ? ( for eg a natural disaster can cause havoc in stock market there by influencing its swing in price. Again this is one factor. there can be lot others too). How does the decision made by world countries affect a market ? and so on can provide some basic information on how stock markets work. But again that wouldn't be all. Many unknown factors can easily affect stock pricing.

All I want to say is that its difficult identifying the factors that affect stock prices. But if you are able to do that, then the rest of the procedure would be a cake walk. I may be wrong here too. If so, I request my fellow peers to correct me.

Nirmal is right. It also depends on how detailed you want to simulate the stock market i.e small changes over a day or large changes over months. I guess large changes over months would be easier.

Then you have to find out information about what affects the stock market. Things like changes in weather, unrest (war), oil prices, news etc. You can get these data from past events/a history of past data set. These will serve as part of the input into your neural network later on.

The next question is what neural network is best to use to develop a model of the stock market. You can use Backpropagation network to develop an initial model to see if your idea works.

Bear in mind that randomness in human behaviour could affect your results. But your model might give some good results if you are predicting over large periods of time as mentioned previously. e.g what the stock price would be next week.

Another approach might be to cast everything into a probabilistic framework. That might give you better results than using a neural network.

but Neural Networks are being in use to predict Stock market... i have studied alot of stuff on it. can u suggest my any model which can predict stock market ? and data analysis methods can be used to find out factors that effect market. am i right?

I would suggest breaking the problem into sub problems to identify the stage were you are at, and from there maybe we can suggest next steps:

1) Have you chosen the neural network type that you are going to use?simple BP time delay, recurrent, etc.

2) Have you selected the data that you are going to use? is it binary coded or just continuous values

3) Is the output going to be buy/sell/hold signals or you are trying to predict up/down movement or just the next sequence number?