Exchange Rate Forecasting with Twitter Sentiment Analysis Technology

This paper attempts to introduce sentiment analysis technology into the task of exchange rate prediction, and studies the impact of sentiment factors on short-term fluctuation of exchange rate based on the theoretical support of behavioral finance. Firstly, the relevant network social media data generated during the new round of trade war between China and the United States were obtained, and the sentiment analysis technology was used to quantify the data to form the sentiment score sequence. Then the nonlinear model LSTM based on machine learning is used to model and predict the high-frequency exchange rate sequence. The empirical results show that the accuracy of the exchange rate forecasting model with the sentiment factors of public opinion is improved, which provides a new idea for the exchange rate forecasting method based on the technical analysis method.