Decoding the crypto crowd: how social media sentiment predicts Ethereum’s price
摘要
Many cryptocurrency investors invest for reasons other than maximizing their expected profits. Traditional financial models are frequently at odds with the behavior of cryptocurrency investors compared to investors in different asset classes. To address this misalignment, we have developed a random forest model using features derived from sentiment analysis of data gathered from social media sources like X, Google Trends, and discussions on Reddit among cryptocurrency (Ethereum) enthusiasts. Our findings underscore the substantial role played by sentiment analysis features in understanding investor behavior and resolving the discord between predictive financial models and actual investor actions. Notably, the modeling presented in this study outperforms the most advanced models currently available in the literature. As a result, our findings hold significance for academic research on financial forecasting models via machine learning and user-generated data and for industry professionals seeking a better understanding of financial markets, particularly cryptocurrency markets.