Predictive analysis of cryptocurrency volatility using the random forest algorithm: the impact of macroeconomic indicators on Bitcoin and Ethereum

Abstract

This study explores the macroeconomic factors driving cryptocurrency price fluctuations, focusing on Bitcoin and Ethereum using the Random Forest machine learning model. It analyzes daily data from 2015 to 2025, incorporating key economic and financial indicators such as the S&P 500, NASDAQ, Treasury yield spreads, Federal Funds Rate, long-term Treasury rates, inflation, Brent oil prices, major exchange rates, and the Volatility Index. Separate predictive models for Bitcoin and Ethereum achieved high accuracy (R² = 0.999 and R² = 0.995, respectively), demonstrating the model's strong forecasting capability. The findings reveal that cryptocurrencies are increasingly influenced by traditional financial markets, particularly US stock indices, highlighting their integration into the global economic system. Bitcoin emerged as relatively stable and less sensitive to monetary policy shifts, functioning as a speculative hedge. In contrast, Ethereum showed greater sensitivity to liquidity and interest rate variables due to its linkage with decentralized finance applications. The study concludes that machine learning methods, combined with traditional macroeconomic indicators, can effectively explain and predict cryptocurrency market behavior, offering a foundation for developing new hybrid economic models tailored to the unique nature of digital assets.