Unformatted text preview: ereafter referred as the trend. These two inputs along with the
present S&P500 index value make up the first 3 inputs of the network.
• To reduce the effect of the noise present in the weekly stock market variation, the
average of the last 5 weekly closing values of the index and the average of the last 10
weekly closing values of the index, along with their absolute first differences and
trends are used. These constitute the next 6 inputs of the network.
• To account for the long term interest rate, the long term interest rate delayed by 25
weeks and its absolute difference and trend are used as inputs 10 through 12. The
short term interest rate is accounted for by using the rate delayed by 48 weeks as well
as the rate delayed by 20 weeks along with their respective absolute differences and
trends, thus giving inputs 13 through 18.
The desired output of the network is the stock index value for the next week. Thus, the
neural network to be used has 18 inputs and 1 output. A simple feedforward network is
used with s...
View Full Document
- Spring '12
- Neural Networks, Artificial neural network, Stock market index, neural network