Cryptocurrency price prediction is an essential aspect of navigating the volatile nature of digital currencies like Bitcoin, Ethereum, and others. With the unprecedented growth and influence of cryptocurrency markets, accurate price forecasting models can significantly benefit traders, investors, and financial analysts. In this paper, we evaluate state-of-the-art machine learning models, including Linear Regression, Support Vector Machines (SVM), and Long Short-Term Memory (LSTM) networks, to predict cryptocurrency prices using historical time series data. The focus is on comparing the performance of these models through their ability to predict price trends.Our dataset comprises historical Bitcoin price data from 2017 to 2021, processed through multiple steps of feature engineering, normalization, and cross-validation. Linear regression achieves a Mean Squared Error (MSE) of 0.0045, making it a simple yet effective model for short-term predictions. SVM, primarily used for price classification (increase or decrease), produces a classification accuracy of 82% with a precision of 80%. The LSTM model, known for capturing long-term dependencies in time-series data, outperforms both with a lower MSE of 0.0031 and provides more accurate long-term forecasts, achieving an 87% accuracy over a test set.These results demonstrate that deep learning models like LSTM are more robust for time-series predictions, especially in highly volatile environments like cryptocurrency markets. However, simpler models like Linear Regression and SVM still offer competitive performance in specific scenarios, such as short-term forecasts and classification tasks. The findings underscore the need for adaptive, datadriven models in the ever-evolving world of cryptocurrencies.

Cryptocurrency Price Prediction, Machine Learning, Time-Series Forecasting, LSTM, XGBoost, Regression Models, Support Vector Machines (SVM), VolatilityFeature Engineering, Model Evaluation