Testing Lasso Regression and XGBOOST on Crypto-Currency Volatility and Price Prediction

Cryptocurrencies have gained popularity as investment and speculative assets, with a surge in daily trading volume exceeding $100 billion. Predicting cryptocurrency prices has become a topic of academic interest, but there is still no consensus on the best predictors due to the high volatility of cryptocurrency prices. Anticipating price volatility is crucial for investors and venture capitalists in the cryptocurrency market. This research examines the application of two machine learning models, Lasso and XGBoost, to predict the short-term returns of leading cryptocurrencies. A high-frequency time series dataset containing 26.75 million rows of trade data from multiple cryptocurrencies dating back to 2018 is used to evaluate the models using various validation metrics. The results of this study provide insights into the suitability of different machine learning models, such as Lasso Regression and XGBOOST, for cryptocurrency volatility and price prediction. We evaluated the models using different validation metrics such as mean absolute percentage error, mean squared error, mean absolute error, and R squared score.