FORECASTING STOCK PRICE USING BACKPROPAGATION ALGORITHM AND NONLINEAR AUTOREGRESSIVE EXOGENOUS MODEL (NARX)

  • วสันต์ ศิลปะ
  • สมพร ปั่นโภชา
  • บำรุง พ่วงเกิด
Keywords: Backpropagation, Fundamental analysis, Artificial Neural Network

Abstract

Nowadays, fundamental analysis is one of the most popular methods. This is depending on the analyst’s experience which can make the biased analysis. Artificial neural networks are the most widely used in various careers. Therefore, it is interesting to use Artificial neural networks for stock price prediction.

The aim of this study is to use artificial neural networks for prediction of The Siam Cement PCL. (SCC) stock price by using fundamental analysis. This is because this company stays in the stock market for long periods. Moreover, the company has enough data to test in artificial neural networks using financial ratio and financial statement. In this study, data were separated into two groups; 80% training data and 20% testing data. These data were tested by using non-linear time series analysis and Levenberg-marquardt algorithm for error adjustment of the network.

The results show that non-linear time series analysis can be employing for predicting the stock price by the best network which had two hidden layers. The first hidden layer has 22 nodes and the second one has 1 node. Mean squared error (MSE) of training data and testing data were 0.0195 and 0.0087, respectively. Therefore, non-linear time series analysis can be used to predict stock price in real situation by using fundamental analysis.

Published
2017-09-17
Section
Engineering and Technology Articles

Most read articles by the same author(s)

<< < 1 2 3 4 5 6 > >>