Stock Price Prediction using Artificial Neural Networks: Case Study – Karachi Stock Exchange

Authors

  • Shams Naveed Zia State Bank of Pakistan
  • Muhammad Zia Usman Institute of Technology, Karachi

DOI:

https://doi.org/10.31384/jisrmsse/2005.03.2.1

Keywords:

Stock Price, Prediction, Artificial Neural Networks, Karachi Stock Exchange, Technical Analysis, Fundamental Analysis

Abstract

Stock Price Prediction has always fascinated researchers because it not only involves the satisfaction of ‘beating the market’ but also the financial gains. Research and Market Data have proved that Efficient Market Hypothesis can be wrong. Various techniques have been used to predict stock prices. The techniques could be divided into two basic categories: which are Technical Analysis and Fundamental Analysis. Technical Analysis [1][2] involves historical data, trends, and their analysis. Where as Fundamental Analysis [1][3] involves analysis of Dividend Yield, Sector’s Performance, Company’s Performance, and Economic Outlook etc. Analysts use Statistical Tools, Expert Systems, and Linear Analysis Tools to predict stock prices. However, it has been proved that Neural Networks are more capable of predicting stock prices [4]. The reason is that stock markets are not linear in their behavior and often behave in chaotic manner. Neural Networks are non-linear and can perform with insufficient data. In this backdrop authors chose to apply Neural Network Techniques in Karachi Stock Exchange to predict stock prices. We have selected Backpropagation as it is the most widely used Neural Network Technique used today [5]. Backpropagation network has three types of layers: Input Layer, Hidden Layer, and Output Layer. Inputs for the Input Layer are indicators like Relative Strength Index (RSI), Moving Average Convergence Divergence (MACD) and Time Series Analysis using Hurst Exponent [6]. The steps involved in our analysis are: Pre-Processing of Data, Feed Processed Data as Input in Input Layer, FeedForward the results to Hidden Layer, Generate output, Backpropagation to adjust weights, and Using GradientDescendent to reach to optimal value. The stages in our analysis are Training and Testing the Network.

Downloads

Download data is not yet available.

Downloads

Published

2005-12-31

How to Cite

Zia, S. N., & Zia, M. (2005). Stock Price Prediction using Artificial Neural Networks: Case Study – Karachi Stock Exchange . JISR Management and Social Sciences & Economics, 3(2), 1–4. https://doi.org/10.31384/jisrmsse/2005.03.2.1

Issue

Section

Original Articles