Deep Temporal Convolutional Networks for High Frequency Cryptocurrency Price Forecasting

Short-Term Price Prediction in Cryptocurrency Ecosystems Using a Hybrid CNN–LSTM Model

This study investigates short-term predictive relationships between major cryptocurrencies—specifically Ethereum (ETH) and Solana (SOL)—and their respective sub-tokens (DYDX, UNI, GRT, JUP, RAY, PYTH) by employing a hybrid deep learning framework that integrates Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) models. Using high-frequency data across different time intervals (15 minutes, 1 hour, and 4 hours), the study examines whether price movements in major tokens are associated with enhanced short- and medium-term predictability of ecosystem-based sub-tokens. The empirical results indicate that the hybrid CNN-LSTM model achieves strong forecasting performance at shorter time horizons, while prediction accuracy declines as the time interval increases, reflecting the limiting role of market volatility. The results indicate statistically significant short-term predictive relationships and pronounced co-movement patterns between major token movements and sub-token pricing behavior, particularly at higher-frequency intervals, suggesting that Ethereum and Solana provide informative signals related to ecosystem-wide liquidity and trading activity.

MSPCIFormer: A Multi‐Scale Patching Channel‐Independent Transformer for Cryptocurrency Price Forecasting

MSPCIFormer is proposed, a novel Transformer‐based architecture that integrates multi‐scale patching with channel‐independent (CI) modelling to capture heterogeneous temporal dynamics while mitigating noise from time‐varying inter‐asset correlations.

Joint prediction of post-overreaction price movements across cryptocurrencies using multi-source and multi-output deep learning models

Accurately forecasting cryptocurrency price movements following market overreactions is crucial for traders, investors, and risk managers operating in highly volatile environments. This study presents a novel multi-source, multi-output deep learning framework designed to predict the direction of price changes in four major cryptocurrencies — Bitcoin (BTC), Ethereum (ETH), Binance Coin (BNB), and Ripple (XRP) — immediately after overreaction events. By jointly modeling these assets, our approach captures their interconnected market dynamics, enhancing predictive accuracy. We compile an extensive dataset with over 656 features from diverse sources, including historical trading data, on-chain metrics, technical indicators, and social sentiment data from Google Trends, collected at both daily and intraday frequencies. To improve model interpretability and performance, we introduce two engineered features — price change magnitude and price variation speed — that effectively represent intraday volatility. Feature selection using a Random Forest approach reduces the feature set to 30 key variables, ensuring robustness and avoiding overfitting. Using three advanced deep learning architectures — LSTM, RNN, and CNN — we train models to classify the next-day price movement as upward or downward. Empirical results demonstrate that the multi-output LSTM achieves an F1-score of 73.42%, outperforming both single-asset models (62.95–68.25%) and alternative architectures. These findings highlight the benefits of joint modeling, leading to more reliable forecasts during turbulent market conditions. Our framework offers a practical tool for algorithmic trading, portfolio management, and risk mitigation in the dynamic cryptocurrency landscape.

Comparison of neural network models for prediction cryptocurrency price volatility in trading pairs

A Comparative Analysis of GRU and LSTM Networks for Interval-Valued Foreign Exchange Market Time Series Forecasting

The Foreign Exchange (FX) market’s extreme volatility and non-linearity pose significant challenges for traditional point-valued forecasting models. This study proposes a comprehensive framework for interval-valued time series (ITS) forecasting, which captures intraday volatility by predicting the daily low and high bounds of exchange rates. We conduct a rigorous comparative analysis between two prominent deep learning architectures: Gated Recurrent Units (GRU) and Long Short-Term Memory (LSTM) networks. To enhance predictive accuracy, the proposed framework leverages a multi-input configuration to model interval-valued time series, using GRU and LSTM networks to learn the non-linear dynamics and complex temporal relationships between the daily price bounds. Empirical evaluations are performed on the EUR/USD exchange rate and simulated datasets characterized by cyclic means and time-varying variance. The numerical results demonstrate that the GRU architecture consistently outperforms the LSTM, achieving a superior Nash-Sutcliffe Efficiency (NSE) of 0.9526 compared to 0.7507 for real-world low-bound prediction. Furthermore, the GRU exhibits greater robustness in simulation experiments and lower computational complexity. These findings suggest that the simplified gating mechanism of GRU is more effective at capturing the structural integrity of FX price intervals, offering a more reliable decision-support tool for market participants navigating high-uncertainty financial environments.

Forecasting Economically Significant Bitcoin Moves: A Multi-Scale TCN with Profit-Optimized Thresholds

Bitcoin's future fluctuations are a substantial concern for investments and risk management. Investors and financial institutions require accurate forecasts of these price movements to hedge and optimize portfolios. This study aims to answer the question of whether Bitcoin's price will rise beyond 5% within the next 7 days by utilizing on-chain, market, and sentiment data from February 2018 to December 2025. The proposed model consists of a multi-scale temporal convolutional network with InceptionTCN blocks, CNN channel attention, adaptive average pooling, and a pairwise ranking loss. Dilated convolutions with bottleneck and fusion layers are employed to efficiently capture features over horizons from 1 to 4 days. Given the class imbalance in the dataset, AUC is used instead of accuracy and other classification metrics to reflect the model's performance better. Subsequently, a profit-optimized decision threshold is also applied to align model selection with financial objectives. The proposed model is compared with 5 other baselines: ImprovedTCN_GRU, LSTM, TCN, XGBoost, and Random Forest. Results indicate that the proposed model achieved an AUC of 0.6316 and a profit of 1.703, outperforming all baseline models. Using a novel deep learning model would assist investors in making better financial decisions.