Prediction of USD Exchange Rate Against CNY and RUB Using Support Vector Regression and Neural Network

Abstract

Major currency exchange rates have been impacted by the escalation of global trade volatility brought on by the trade war between the United States and China and economic sanctions imposed on Russia. USD dominance in global trade exposes developing countries to economic risks. BRICS seeks to reduce reliance by boosting local currency trade and diversifying reserves. This study analyzes BRICS exchange rate movements, specifically USD-RUB and USD-CNY, using Support Vector Regression (SVR) and Neural Network (NN). Statistical analysis of 2009-2025 data shows USD-RUB's high volatility due to oil prices and sanctions, while USD-CNY remains more stable but is influenced by monetary policy and global conditions. The results show that the SVR method is superior to NN in prediction accuracy. For USD-RUB, SVR with a sigmoid kernel achieves MSE 6.1596, MAE 1.8808, and MAPE 1.95%, while for USD-CNY, SVR with a Radial Basis Function kernel achieves MSE 0.0014, MAE 0.0322, and MAPE 0.45% Thus, the use of SVR-based prediction models is recommended to analyze the exchange rate to reduce the risk of volatility. Additionally, diversifying reserves, enhancing bilateral trade in local currencies, and considering external factors like commodity prices and global policies can improve exchange rate stability and economic resilience.