This research paper investigates the application of machine learning algorithms for forecasting Bitcoin price movements, addressing the cryptocurrency's complexity and volatility in financial markets. The study employs four distinct machine learning approaches—Support Vector Machines (SVM), Artificial Neural Networks (ANN), Naive Bayes (NB), and Random Forest (RF)—alongside logistic regression as a benchmark model. A key methodological innovation involves testing these algorithms on both continuous price data and a specially created discrete dataset, allowing for comparative analysis of different data representations.

The research employs comprehensive evaluation metrics including F statistic, accuracy measures, Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Root Absolute Error (RAE), supplemented by t-tests for statistical comparison between algorithm performances. Empirical findings reveal that Random Forest demonstrates superior forecasting capability on continuous data, while Artificial Neural Networks excel when applied to discrete datasets. Notably, the study demonstrates that discrete data representation consistently enhances forecasting performance across all tested algorithms, suggesting that data preprocessing and representation significantly impact prediction accuracy in cryptocurrency markets. These results contribute to the growing literature on quantitative finance applications in cryptocurrency trading and provide practical insights for algorithmic trading strategies.