A neural network model for predicting price changes of the exchange rate EUR/USD

Abstract

With the rapid development of the global economy, economic exchanges and cooperation between countries have become increasingly frequent, and the importance of the foreign exchange market has become increasingly prominent. As a link between international trade, foreign exchange plays a crucial role in formulating relevant financial policies and the risk management of multinational enterprises. However, accurately predicting foreign exchange changes is very challenging because the foreign exchange market is characterized by non-linear and dynamic changes and exchange rate fluctuations are also affected by historical data. The analysis method based on time series has always been an effective solution for studying exchange rate issues, and the model trained by the deep learning model based on time series data can further simulate and predict the exchange rate. Therefore, this article uses the BP neural network algorithm based on the time series model to predict the price change of the exchange rate EUR/USD. This article first uses the historical exchange rate data of the euro against the US dollar as a training set and performs data preprocessing. Secondly, the activation function and learning rate of the model are tuned, and the performance of the model is compared by evaluating indicators to improve prediction accuracy. The final experimental results show that the neural network model is highly accurate and predictive in predicting EUR/USD exchange rate price changes.