Cryptocurrency Forecasting after the 2024 Bitcoin Halving Yields New Insights

The world of cryptocurrencies continues to evolve rapidly, driven by factors such as growing institutional participation and tighter global macroeconomic conditions. The 2024 Bitcoin halving was a critical turning point in the cryptocurrency market, leading to an imbalance between supply and demand that contributed to price appreciation and volatility.

Despite its significance, there is still a lack of research on multivariate cryptocurrency forecasting after this period. To address this gap, a recent study compared two popular multivariate forecasting approaches: Vector Autoregressive (VAR) models and multivariate Fourier series estimators. The goal was to determine which method provides the most accurate predictions.

The study used daily closing price data from April 2024 to August 2025 for Bitcoin, Ethereum, and Litecoin. The results of the Variance Decomposition analysis showed that Bitcoin's own innovations drove its price movements during this period, accounting for approximately 98.84% of forecast error variance. In contrast, Ethereum and Litecoin were more susceptible to cross-asset shocks.

The study found that the Fourier Series Estimator method provided more accurate predictions than VAR models, with a Mean Absolute Percentage Error (MAPE) value of 3.768%. This finding suggests that nonparametric methods can be effective in modeling complex data dynamics, particularly in financial analysis contexts.