Impact of News Sentiment on Foreign Exchange Rate Prediction

Foreign exchange market is a global market place to exchange national currencies. It is highly volatile and depends on economic positions of the two countries. Like any other financial market, the FOREX market rates are influenced by the market sentiment. Market sentiment is a metric to gauge the mood and trend of prices in the market. This includes various influencing factors such as other asset markets like gold and crude oil, currency indices. Information about the two countries of exchange pair in the form of news, blogs, social media messages, etc. also contributes towards the market sentiments. This work proposes a foreign exchange rate forecasting model which uses sentiment score as a parameter along with econometric factors. The work is primarily focused on the USD-INR currency pair. For text mining, FinBERT, a language model based on Bidirectional Encoder Representations from Transformers (BERT) fine-tuned for financial sentiment classification is used. The proposed model uses Long-Short Term Memory (LSTM) network, since it performs well in analysing and forecasting time series data. The outcome of this work will be to accurately forecast the foreign exchange rate for the Indian Rupee against the US Dollar. The performance of the model is evaluated based on Root Mean Square Error (RMSE) and R
²
score. The hybrid prediction model with market indicators and news sentiment outperforms the baseline models.