Trading ebook Finance - Neural Prediction of Weekly Stock Market Index

4 and the test results are detailed in the next

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: size of 0.4 and the test results are detailed in the next section. Fig. 12: Training and testing data (S&P500 index) Training data 1 0.5 0 0 50 100 Data number 20 30 40 50 Data number Testing data 1 150 200 0.8 0.6 0.4 10 60 70 14 Fig 13: Training and test data for interest rates Long rate 1 0.5 0 Short rate 1 100 150 200 250 48 week delay 0.5 0 1 Short rate 50 50 100 150 200 250 20 week delay 0.5 0 50 100 150 200 250 Test results Fig. 14 shows the test results for this case. The plot shows that the network is able to predict the severe crash one-week before it occurred. Although, the network output is lower than the actual values in the initial part of the test range, there is no trend in the stock index, which would force the network to predict a crash. Thus, a network using only the past index values would have been unable to predict this crash, which can be attributed to the rising long term interest rates. For this case, the network predicts the correct trend of the stock index 65 times out of the test sample of 75 points. The maximum percentag...
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.

Ask a homework question - tutors are online