Abstract

Nowadays, blockchains and cryptocurrencies have become very popular among investors and hedge fund managers, showing great potential for the future. As cryptocurrencies become more integrated into trading with the advancement of websites and computers, their price fluctuations are increasingly important to the public. This paper focuses on using three machine learning models—XGBoost, LightGBM, and a Linear Model—to predict the prices of three cryptocurrencies: Bitcoin, Dogecoin, and Ethereum. The study uses data from January 1, 2020, to December 7, 2022, including close price, open price, high price, low price, and trading volume. The analysis shows that the Linear Model performs best, accurately predicting trends and prices. While the other models also make good predictions, they do not surpass the Linear Model. These findings can help others predict cryptocurrency prices and gain a better understanding of both cryptocurrency markets and machine learning techniques.