{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

Trading ebook Finance - Neural Prediction of Weekly Stock Market Index

Trading ebook Finance - Neural Prediction of Weekly Stock Market Index

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon
Application of Neural Networks to Stock Market Prediction Amol S. Kulkarni ª 1996 Amol S. Kulkarni All rights reserved. Material in this report may not be reproduced in any form. This material is made available for learning and may not be used in any manner for any profit activities without the permission of the author.
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
1 Introduction The aim of this project is to predict the future values of the Standard & Poor’s 500 (S&P 500) stock index based on its past values and the past values of some financial indicators. There have been many attempts at predicting stock market movements, most of them based on statistical time series models [1]. Most of these attempts have been unsuccessful due to the complex dynamics of the stock market. The efficient market hypothesis says that stock prices rapidly adjust to new information by the time the information becomes public knowledge, so that prediction of stock market movements is impossible [2]. This hypothesis seems to be correct for static and linear relationships explored traditionally using multiple regression analysis. However, it is possible that dynamic and non-linear relationships exist which cannot be modeled by traditional time series analysis methods [3]. This, is the motivation for application of neural networks to financial time series analysis. A huge amount of research is being done on the application of neural networks to stock markets. Some of the applications include prediction of IBM daily stock prices [4], a trading system based on prediction of the daily S&P 500 index [5], short term trend prediction using dual-module networks [6], weekly index prediction [7], monthly index prediction using radial basis functions [8] etc. Some of these papers use the past values of the stock index only, as the input to the neural network so as to obtain the future values, while some use additional fundamental and financial factors as inputs. This project explores the effect that short and long term interest rates have on the stock market, in particular on the S&P 500 index. It is well known that an increase in interest rates tends to lower the stock market and vice versa [9]. The hypothesis is that the current stock prices indicate the cumulative sum of the present and future worth of any company. If the interest rates increase, the equivalent future value of a stock in terms of today’s dollars reduces, causing the stock price to reduce. Financial experts report that the long term value of a broad based index is affected by interest rates after some unknown delay. However, the exact effect is unknown and hence, a neural network can be suitably applied to find this non-linear mapping. The next section is a brief review of some of the similar work done while section three describes the selection of features from the raw data for training of the neural network. Test results for different strategies are given in section four and the last section lists the conclusions from this project.
Background image of page 2
Image of page 3
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}