TLDR: A new study by Imen Mahmoud and Andrei Velichko demonstrates that incorporating COVID-19 health data, especially vaccination rates, significantly improves the accuracy of Bitcoin return forecasts. Using a LightGBM model optimized with a genetic algorithm, the research found that COVID-19 indicators increased the model’s ability to explain Bitcoin price variance by 40% and reduced prediction errors by 2%, particularly in capturing extreme market fluctuations. This suggests that public health signals can provide valuable insights for investors and policymakers navigating financial markets during systemic crises.
A recent study delves into the complex relationship between the COVID-19 pandemic and Bitcoin’s market behavior, proposing a new way to predict Bitcoin returns by integrating public health data. The research, titled “Evaluating COVID-19 Feature Contributions to Bitcoin Return Forecasting: Methodology Based on LightGBM and Genetic Optimization,” was conducted by Imen Mahmoud and Andrei Velichko.
The global health crisis caused by COVID-19 significantly impacted financial markets, leading to unprecedented volatility and a surge of interest in alternative assets like Bitcoin. While Bitcoin is often seen as a hedge against traditional economic risks, its interaction with public health crises has been less understood. This study aimed to determine if incorporating COVID-19-related health data could significantly improve the accuracy of Bitcoin return predictions, especially during periods of market uncertainty.
The researchers developed a novel framework that combines a LightGBM regression model with a genetic algorithm (GA) for optimization. They built a comprehensive dataset that included daily Bitcoin returns alongside various COVID-19 metrics, such as vaccination rates, hospitalization figures, and testing statistics. To ensure robust statistical assessment, predictive models were trained both with and without these COVID-19 features, and their performance was optimized over 31 independent runs.
The findings revealed a significant improvement in model performance when COVID-19 indicators were included. Specifically, the model’s ability to explain the variance in Bitcoin returns (measured by R²) increased by 40%, and the Root Mean Square Error (RMSE) decreased by 2%. These improvements were statistically highly significant, indicating that pandemic-related factors help the model better capture the complexities of market behavior during unstable periods, such as sudden downturns or spikes. Interestingly, while R² and RMSE showed clear benefits, the Mean Absolute Error (MAE) showed only a marginal, non-significant improvement. This suggests that COVID-19 data is particularly effective in forecasting extreme price movements rather than gradual, continuous price shifts.
Among the various COVID-19 features, vaccination metrics emerged as the most dominant predictors. The 75th percentile of fully vaccinated individuals, in particular, proved to be highly influential. This highlights the market’s sensitivity to critical vaccination thresholds, which often correspond to significant shifts in public health policies and economic reopening strategies. The study also emphasized that raw COVID-19 data had limited predictive power; instead, statistical transformations like moving averages, percentiles, and range statistics over rolling 7-day windows greatly enhanced their informational value.
The methodology employed in this research extends existing financial analytics tools by incorporating public health signals. This provides investors and policymakers with refined indicators to navigate market uncertainty during systemic crises. The findings also contribute to the ongoing debate regarding market efficiency, suggesting that the significant predictive improvements achieved by incorporating COVID-related data align with the Adaptive Market Hypothesis, which posits that markets are adaptive rather than strictly efficient, especially during crises.
Also Read:
- Unraveling Time Series Causality Across Frequency Bands
- Artificial Intelligence and Green Finance Converge to Foster Climate Resilience
This research offers a fresh perspective on how public health dynamics can influence financial markets, particularly the volatile cryptocurrency space. By understanding these complex linkages, investors can potentially refine their risk management and hedging strategies, leading to more resilient investment approaches during future global crises. For more details, you can refer to the full research paper available at this link.
