This preview shows page 1. Sign up to view the full content.
Unformatted text preview: of rt
for some weeks. The feature extraction is not explained in  except for the fact that
some irregularity is removed and logarithm function is used before normalization. The
authors claim that the use of the weighted sum of the outputs of many neural networks
reduces the error, especially since the returns are predicted for a few weeks. A buy/sell
system is setup based on the predicted returns and this system is shown to perform much
better than a buy-hold strategy. However, the teaching data uses future returns, so that
this method cannot be used for actual stock trading. (The authors do mention this as part
of their proposed future work).
One week ahead prediction using feedforward networks
This work  uses a simple feedforward neural network trained using past and present
data to predict the value of the FAZ-Index which is the German equivalent of the DJIA.
Input data includes the moving average of past 5 and 10 weeks of the FAZ-Index, a first
order difference of the FAZ-Index and its moving averages, the present bond market index
and its first order difference and the Dollar-Mark exchange rate along with its f...
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