• Asif Ali Ansari
• Sarita Gulia
• Rashmi Gupta

The extreme volatility of cryptocurrency markets creates difficulties for traders, investors, and financial analysts due to their unpredictable nature. Traditional models like ARIMA and GARCH perform well for linear pattern detection whereas LSTM, GRU, and Transformer networks excel at modeling complex nonlinear dependencies. These models individually fail to deliver a complete and accurate depiction of cryptocurrency volatility. The research proposes a hybrid method that merges time-series forecasting models with deep learning methods to improve forecast precision. The analysis needs historical price data from CoinGecko while it evaluates technical indicators that include Simple and Exponential Moving Averages (SMA, EMA), Bollinger Bands, Relative Strength Index (RSI), Moving Average Convergence Divergence (MACD) and volatility signals such as GARCH-based volatility, Average True Range (ATR), and rolling standard deviation. The study applies ensemble learning techniques to evaluate forecast reliability across different models including ARIMA, GARCH, LSTM, Transformer, and CNN-LSTM hybrid models. Metrics such as RMSE, MAE, MAPE, and the Sharpe Ratio are used in evaluating the models performance. Hybrid models produce better results than conventional statistical methods and deep learning techniques when forecasting short-term cryptocurrency volatility. The application of meta-learning and residual learning techniques alongside XGBoost as a meta-model leads to enhanced prediction accuracy. The study advances financial forecasting methods by demonstrating superior cryptocurrency volatility predictions with hybrid models to aid risk management and improve algorithmic trading and investment tactics.

ARIMA, Cryptocurrency Volatility, Deep Learning, GARCH, GRU, Hybrid Modeling.

"Hybrid Modeling for Cryptocurrency Volatility Prediction using Time-Series and Deep Learning", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 5, page no.i285-i293, May-2025, Available :http://www.jetir.org/papers/JETIR2505929.pdf

"Hybrid Modeling for Cryptocurrency Volatility Prediction using Time-Series and Deep Learning", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 5, page no. ppi285-i293, May-2025, Available at : http://www.jetir.org/papers/JETIR2505929.pdf