MODELLING AND PREDICTING STOCK PRICES OF NIGERIAN STOCK EXCHANGE USING SOME MACHINE LEARNING TECHNIQUES AND TIME SERIES MODEL

Authors

  • Uzoaga G.A. Department of Statistics, Nasarawa State University, Keffi,
  • Adenomon M.O. Department of Statistics, Nasarawa State University, Keffi,
  • Nweze N.O. Department of Statistics, Nasarawa State University, Keffi,
  • Bilkisu Maijama Department of Statistics, Nasarawa State University, Keffi,

Abstract

Nigeria is an emerging stock market in Africa. The Nigerian stock market's potential and growth prospects can be further explored by increasing trading volume proportionate to the size of the economy. The stock market's liquidity will increase, and therefore, demand for predictions into the future will be of help to investors. Machine learning algorithms produce more exact predictions and find the future value of financial assets traded on an exchange. This study aims to model and predict Nigerian stock prices using machine learning techniques and ARIMA with exogenous variables and to investigate the important variables in predicting stock prices in Nigeria. The study utilized secondary stock market data, sourced from the website (www.investing.com), covering 11 years with total observations of 2773. The models were adjudged using key performance criteria metrics such as Root Mean Square Error (RMSE), R-Squared value ( ), and Mean Absolute Error (MAE).  In terms of training the models, Random Forest techniques performed better with low values of RMSE and MAE. Meanwhile, linear regression performed the worst with high values of RMSE and MAE. Although using the R-square criterion, SVM did very well. Lastly, for testing the models, Random Forest techniques performed better with a low value of RMSE. While the Decision Tree performed the worst with a high value of RMSE. Although using the R-square and MAE criteria, SVM did very well. This study concluded that Random Forest and Support Vector Machine have the potential to effectively predict stock prices in Nigeria.

Downloads

Published

2025-06-30

Issue

Section

ARTICLES