Financial News for Exchange Rate Predication

Currency is a critical medium of exchange that connects individuals to the global economy. However, the exchange rate of national currencies is dynamic and unpredictable. Previous research has demonstrated that financial news plays a key role in forecasting market trends. Therefore, a considerable amount of financial data is used in exchange rate prediction studies. However, these data sources are often limited or outdated, and the potential of using multisource financial news data remains underexplored. This study utilizes news from multiple United States financial news networks to identify the optimal combination of multisource data by integrating various text representations, sentiment features, and machine learning (ML) techniques for exchange rate prediction. The findings reveal that combining support vector machine (SVM) and term frequency-inverse document frequency (TF-IDF) yields the best accurate in ML-based exchange rate forecasting using multisource financial news.