The National Bank of Ukraine has released new research on cryptocurrency price modeling. The study, covering data from Binance for Bitcoin, Ethereum, Ripple, and Dogecoin, spans from July 6, 2020, to April 1, 2023.
This research pioneers the use of LSTM neural networks in predicting crypto prices, outperforming traditional models like Naive, ARIMA, and FB Prophet. LSTM models showed lower error rates, indicating their effectiveness in managing the high volatility of crypto prices.
Key market dynamics include supply and demand, market stability, and regulatory changes. Unique features like 24/7 trading and high volatility distinguish the crypto market from traditional financial markets.
The findings promise advancements in trading algorithms and portfolio management tools, potentially enhancing market prediction accuracy. This research highlights the critical role of advanced machine learning in navigating the complex cryptocurrency landscape.
