Predicting the Performance of Gold Companies on the Victoria Falls Stock Exchange Using Machine Learning.
- Author
- Zengeya, Deleen R.
- Title
- Predicting the Performance of Gold Companies on the Victoria Falls Stock Exchange Using Machine Learning.
- Abstract
- The forecasting of the performance of stock markets in the emerging economies waged serious problems as a result of erratic macroeconomic conditions and chaotic market dynamics. The natural setting of Victoria Falls Stock Exchange in Zimbabwe was hyperinflation, currency instability, and incredible economic uncertainty not easy to be captured by the traditional econometric models. The gold mining companies listed on the VFEX experienced some special problems where the linearly-oriented methods of prediction were not appropriate in predicting the interaction of global commodity rates, local currency floats and hyperinflating effects that acted collectively in the stock performance of the companies. This study compared how random forest machine learning models compared with conventional linear regression models in forecasting the returns on the VFEX stock of gold companies. The study used the stock returns data of the major 4 companies dealing in gold namely; Padenga Holdings Limited, Caledonia Mining Corporation, Bindura Nickel Corporation, and Karoi Consolidated Gold Mine across a period of 24 months between January 2023 and December 2024. Descriptive statistics, diagnostic tests, model implementation, and a complete validation process involved to measure prediction accuracy and model performance were used in a quantitative approach. It was proven that random forest models performed far better than linear regression in all performance measurements with the differences between MAE values being between 19.2-23.6 percent and RMSE between 19.0-26.8 percent. R² gains as high as 41.6 percent and 54.1 percent were attained in random forest models, with exception of the hyperinflationary periods when random forest models recorded R² greater than 0.35 and linear lower than 0.20. the price of gold became a dominant predictive variable with 33.7 percent point of predictivity followed by the company revenue with 24.0 percent. The research gave a solid indication that machine learning strategies were important in accurate estimation of stock returns in the volatile emerging market situations because it had better reliability in making investment decisions within problematic economic settings with high macro-economic uncertainty.
- Date
- June 2025
- Publisher
- BUSE
- Keywords
- Gold Companies
- Stock Exchange
- Supervisor
- MS. P. Hlupo
- Media
-
Zengeya, Deleen R..pdf