Unformatted text preview: irst order
difference. The value of the FAZ index is predicted for the next week based on this data.
The network is trained using data for the past M weeks and is then tested based on data
for the next L weeks, where M is called the training window and L is called the testing
window. For successive prediction, the windows are moved ahead and the network is
retrained. Three different networks are compared each having a different set of inputs, one
of which has only the last 10 FAZ index values as the input. It is seen that the network fed
with technical indicator data performs better than the one trained only on past index
values. Normalization of training data is done so as to keep the data within 0.1 and 0.9,
however, the normalization method is not given. This approach is particularly suitable to
the aim of this project and is hence, used as the basic method in this project with certain
Radial basis function approach
Komo, Chang and Ko  use a radial basis function network to predict the future stock
index values based on past data of the index and other technical indicators such as
transaction costs, bond market valu...
View Full Document
This note was uploaded on 07/20/2012 for the course ECON 203 taught by Professor Girishdev during the Spring '12 term at Indian Institute of Technology, Kharagpur.
- Spring '12