The study employs the Box-Jenkins Methodology to forecast South African gold sales. For a resource economy like South Africa where metals and minerals account for a high proportion of GDP and export earnings, the decline in gold sales is very disturbing. Box-Jenkins time series technique was used to perform time series analysis of monthly gold sales for the period January 2000 to June 2013 with the following steps: model identification, model estimation, diagnostic checking and forecasting. Furthermore, the prediction accuracy is tested using mean absolute percentage error (MAPE). From the analysis, a seasonal ARIMA(4,1,4)×(0,1,1)12 was found to be the “best fit model” with an MAPE value of 11% indicating that the model is fit to be used to predict or forecast future gold sales for South Africa. In addition, the forecast values show that there will be a decrease in the overall gold sales for the first six months of 2014. It is hoped that the study will help the public and private sectors to understand the gold sales or output scenario and later plan the gold mining activities in South Africa. Furthermore, it is hoped that this research paper has demonstrated the significance of Box-Jenkins technique for this area of research and that they will be applied in the future.
The current study aims to model the South African crude oil prices using the hybrid of Box-Jenkins autoregressive integrated moving average (ARIMA) and Neural Networks (NNs). This study introduces a hybrid approach to forecasting methods aimed at resolving the issues of lack of precision in forecasting. The proposed methodology includes two models, namely, hybridisation of ARIMA with artificial neural network (ANN)-based Extreme Learning Machine (ELM) and ARIMA with general regression neural network (GRNN) to model both linear and nonlinear simultaneously. The models were compared with the base ARIMA model. The study utilised monthly time series data spanning from January 2021 to March 2023. The formal stationarity test confirmed that the crude oil price series is integrated of order one, I(1). For the linear process, the ARIMA (2,1,2) model was identified as the best fit for the series and successfully passed all diagnostic tests. The ARIMA-ANN-based ELM hybrid model outperformed both the individual ARIMA model and the ARIMA-GRNN hybrid. However, the ARIMA model also showed better performance than the ARIMA-GRNN hybrid, highlighting its strong competitiveness compared to the ARIMA-ANN-based ELM model. The hybrid models are recommended for use by policy makers and practitioners in general.
The distributional specification in Markov-switching GARCH models has historically been driven by empirical convention rather than statistical theory. This paper derives the two-regime MS-GARCH specification from the Maximum Entropy Principle, providing an information-theoretic motivation for Student-t regime-conditional innovations in cryptocurrency volatility modelling. The framework is applied to five major cryptocurrencies, Bitcoin, Ethereum, Ripple, Litecoin, and Bitcoin Cash, over the period January 2017 to March 2026, comprising 15,834 daily observations spanning six complete market cycles. Three principal findings emerge. First, a Calm-Phase Fragility pattern is identified: four of five assets exhibit calm-regime half-lives below one trading day (0.48 to 1.16 days), with turbulence the dominant long-run state (stationary turbulent probability in [0.451, 0.771] across all assets), establishing turbulence rather than calm as the structural baseline of the cryptocurrency ecosystem. Second, the Maximum Entropy derivation yields endogenous Student-t degrees of freedom, with heavy-tailed turbulent innovations (degrees of freedom approximately 4.5) confirmed across all assets, validating the MaxEnt constraint framework empirically. Third, near-unity turbulent GARCH persistence drives MS-GARCH point forecasts toward the persistence ceiling, consistent with an information-theoretic bound on predictability when the calm half-life collapses below one trading day; HAR-RV achieves the lowest QLIKE loss for three of five assets under these near-critical conditions. Cross-asset consistency is confirmed across seven statistical indicators including Hill tail exponents in [2.31, 3.26], Hurst exponents in [0.543, 0.577], and Wald tests rejecting parameter homogeneity at p < 0.001 for all assets. The framework is formalised as a deployable expert system for real-time regime monitoring and risk management.
This study investigated the forecasting performance of the South African gold sales series using the seasonal autoregressive integrated moving average (SARIMA), generalised autoregressive conditionally heteroscedastic (GARCH), general regression neural network (GRNN) and artificial neural network (ANN)-based extreme learning machine (ELM). This study employed traditional methods and a recently developed ML method for single hidden-layer feed-forward neural networks (SLFNs). The findings revealed that was considered the best model for the gold sales series. The empirical findings demonstrated that the SARIMA model outperforms neural network-based models, providing the South African government and its lenders with a more reliable and cost-effective tool for predicting foreign exchange earnings from gold. This study contributes to the literature by providing one of the first comparative evaluations of traditional time-series models and advanced neural network methods for forecasting South African gold sales. This study is novel as it is a first-of-its-kind comparative application of traditional SARIMA and GARCH models alongside GRNN and ANN-based ELM methods to South African gold sales, revealing the superior forecasting performance of a traditional SARIMA model over advanced ML approaches. Future research should explore the development and application of hybrid models that integrate the strengths of linear SARIMA frameworks with the pattern-recognition capabilities of nonlinear ANN-based ELM models.
Forecasting equity returns remains challenging because financial markets exhibit nonlinear dynamics, volatility clustering, and complex temporal dependencies that are difficult to capture using a single modelling approach. Traditional statistical models can capture dependence and volatility dynamics, while deep learning models are capable of learning nonlinear temporal patterns. This study evaluates the forecasting performance of traditional statistical models (ARIMA and GARCH), deep learning models (TCN and GRU), and hybrid models (ARIMA-TCN, ARIMA-GRU, GARCH-TCN, and GARCH-GRU) for forecasting the FTSE/JSE Top 40 index in South Africa. Daily closing prices comprising 4159 observations from 2010 to 2026 were analysed, with model performance evaluated on log returns using MSE, RMSE, and MAE, with the Naïve model used as a benchmark. The results show that all competing models substantially outperformed the Naïve benchmark, while the TCN outperformed the standalone GRU. GARCH-based models also demonstrated strong forecasting performance, highlighting the relevance of volatility dynamics in forecasting equity returns. Although GARCH-GRU recorded the lowest MSE, RMSE, and MAE among the models evaluated, its performance was very similar to that of the GARCH (2,2) model. The Diebold-Mariano test further showed no statistically significant difference in predictive accuracy between GARCH-GRU and (DM = 0.9495; p = 0.3424). These findings indicate that GARCH-GRU and provide statistically comparable forecasting performance, suggesting that the additional complexity of the hybrid framework does not necessarily result in a significant improvement in predictive accuracy. This study contributes to the financial forecasting literature by providing empirical evidence from an emerging African equity market and demonstrating the importance of volatility modelling and complementary statistical–deep learning approaches for forecasting FTSE/JSE Top 40 log returns.
The study utilized two machine learning (ML) techniques namely Random Forest (RF) and Support Vector Machines (SVM) with the aim of understanding the implications of missing data on RF and SVM's ability in forecasting accuracy. The study employed a time series data sourced from South African Reserve Bank website from January 1960 to June 2021 with a total of 738 observations for variables of government expenditure as a dependent variable and government revenue as independent variable. The study employed RMSE, MSE, MAE and MAPE for measuring forecasting accuracy of these two machine learning models. Results of the study revealed that SVM outperformed RF by giving the highest prediction accuracy. Furthermore, the results show that missing data does not negatively impact the accuracy of these models since the prediction error values decreases from modelling complete data to incomplete data. The study recommends that for future research purposes other machine learning models to be explored and that other scenarios in the data including bias and outliers to be considered.