Abstract: Bitcoin, the most widely used cryptocurrency, is characterized by extreme price fluctuations, making its prediction is a complex and crucial task in the financial domain. This research paper presents a comparative analysis of various machine learning and deep learning models for Bitcoin price forecasting. Traditional approaches such as Linear Regression, Decision Trees, and Support Vector Machines are compared with advanced deep learning architectures like Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU), and Transformer-based models. The study evaluates these models using key performance metrics, including Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R²-score. The results demonstrate that Transformer-based models outperform other techniques due to their ability to capture long-term dependencies and complex patterns in sequential data. This paper further discusses hyperparameter tuning, trading strategy implications, and limitations of each model, providing a comprehensive perspective on Bitcoin price forecasting.

| DOI: 10.17148/IJIREEICE.2025.13439