Comparing various methods of forecasting stock index prices in a shock-affected market: based on data covering the COVID-19 pandemic
The recent years have been economically challenging, with financial markets worldwide facing turmoil. From late 2020 to mid-2022, global economies were heavily impacted by the COVID-19 pandemic due to the implementation of lockdowns and austere quarantine measures that crippled world trade. As the p...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Universiti Kebangsaan Malaysia
2025
|
| Online Access: | http://journalarticle.ukm.my/25914/ http://journalarticle.ukm.my/25914/1/SMT%2018.pdf |
| Summary: | The recent years have been economically challenging, with financial markets worldwide facing turmoil. From late 2020 to mid-2022, global economies were heavily impacted by the COVID-19 pandemic due to the implementation of lockdowns and austere quarantine measures that crippled world trade. As the pandemic abated with increased vaccination rates, global economic growth was expected to return to normalcy. However, commodities (especially oil), exchange rates, and stocks remain greatly devalued, continuing to restrict financial progress. The primary goal is to identify the most effective models for predicting the impact of international trade during COVID-19 pandemic. Accurate stock index forecasting is crucial in such uncertain economic conditions. Using Malaysia, Indonesia, and Singapore as the research targets, this study compares time series linear regression (TSLR), Bayesian regression, and support vector regression (SVR) in predicting major stock indices during pre- and post-vaccination periods. The models are evaluated based on root mean square error (RMSE), mean absolute error (MAE), and adjusted R² to determine their effectiveness. Results show that Bayesian regression outperforms other models in the pre-vaccination period due to its ability to incorporate prior information, whereas SVR performs better in the post-vaccination period, capturing complex market dynamics more effectively. These findings suggest that Bayesian regression is particularly useful during high-uncertainty periods, while SVR is better suited for stable market conditions. This method should be utilized extensively in future research with other machine learning methods to enhance forecasting accuracy, while additional macroeconomic variables such as inflation, interest rates, and geopolitical factors should also be considered. Furthermore, the findings of this study, shows that by incorporating Bayesian regression and machine learning can provide valuable insights for policymakers, investors, and financial analysts in navigating financial risks during economic crises. |
|---|