RMB Exchange Rate Forecasting Analysis - Based on BP Neural Network

Abstract

This paper proposes to combine factor analysis regression with time series to construct a BP neural network for exchange rate forecasting, and to analyze the better way of medium-term and short-term exchange rate forecasting by comparing the model constructed according to time series and according to random series. The results show that the BP neural network model with exchange rate influencing factors as input variables and RMB/USD mid-price as output variables is better for short-term and medium-term forecasting of RMB, and the model based on time series is more suitable for medium-term forecasting, while the model based on random series is more suitable for short-term forecasting, but both approaches are more accurate for exchange rate forecasting with less error.