A Novel Multi-Window Time-Series Forecasting of Major Cryptocurrencies Using Hybrid 1D-CNN–LSTM Framework

Predicting cryptocurrency prices is challenging due to high market volatility and nonlinear dynamics of digital asset markets. This paper proposes a novel hybrid deep learning model combining framework that integrates one-dimensional Convolutional Neural Networks (1D-CNN) and Long Short-Term Memory (LSTM) networks to forecast the closing prices of Bitcoin (BTC), Ethereum (ETH), and BinanceCoin (BNB). The 1D-CNN component captures short-term local patterns captures local price patterns, while the LSTM component models long-term temporal dependencies. Using historical daily transaction data and a sliding window approach, we evaluated the hybrid model against standalone 1D-CNN and LSTM baseline deep learning models across multiple window configurations and various window sizes. Experimental results demonstrate that the proposed hybrid model consistently outperforms the baseline models, especially in medium- and long-term forecasting horizons, achieving lower prediction errors (MSE, MAE, RMSE) and higher R² scores. This highlights the effectiveness of integrating CNN and LSTM for improved cryptocurrency price forecasting and provides practical value enhanced cryptocurrency price prediction accuracy, offering valuable insights for investors and researchers navigating these complex markets.