This preview shows page 1. Sign up to view the full content.
Unformatted text preview: work
The following conclusions can be drawn from this project work:
• A feedforward neural network has been successfully applied to the problem of a oneweek ahead prediction of the weekly closing prices of the S&P 500 index.
• The dependence of the stock index on interest rates is established using crosscorrelation values and the inclusion of this dependence is seen by the performance of
the network for the second test case.
• The trained neural network performs very well even for worst cases where there are
sudden rises or falls in the stock market index.
The original aim of this project was the application of the available data for medium term
prediction of the stock market index using some form of a recurrent network.
• For medium term prediction, the method used in  and described in Section 2, can
be applied. There, the network predicts the stock value for the next few weeks and the
training for that is performed by feeding the error for outputs back as inputs during
• A r...
View Full Document
- Spring '12