A News Sentiment-Driven Framework for Forex Trend Prediction

The foreign exchange (Forex) market is highly sensitive to macroeconomic announcements, making fundamental information a key driver of price movements. This paper proposes a framework for trend prediction in EUR/USD by integrating currency-specific economic surprise indices derived from macroeconomic news with historical price data. Macroeconomic announcements are processed to compute daily surprise scores, which quantify the deviation of actual released values from market consensus forecasts. These currency-specific surprise indices are then incorporated as exogenous features in a binary classification framework, predicting upward or downward movements in the EUR/USD exchange rate. Experimental results demonstrate that including surprise-based features improves trend prediction accuracy compared to models using only historical prices, highlighting the effectiveness of the proposed approach in capturing fundamental pressures in the Forex market.

Keywords:

Forex trend prediction , Economic surprise index , Macroeconomic news , Fundamental analysis