COMPARISON OF NEURAL NETWORK MODELS FOR PREDICTION CRYPTOCURRENCY PRICE VOLATILITY IN TRADING PAIRS

Abstract

The increasing digitization of financial markets and the rise of Web3 trading pairs have significantly impacted asset trading, introducing new challenges in price volatility forecasting. Traditional risk management strategies often fail to adapt to the unpredictable nature of cryptocurrency markets. This study explores the application of advanced machine learning models for forecasting price volatility in cryptocurrency trading pairs. We compare the performance of Temporal Fusion Transformer (TFT), Temporal Convolutional Network (TCN), XGBoost, Random Forest, and a hybrid CNN-LSTM model. Our findings indicate that TFT and CNN-LSTM hybrid models outperform traditional recurrent neural networks in capturing complex market dynamics, enhancing risk management strategies in high-frequency cryptocurrency trading.