Predictions models of Taiwan dollar to US dollar and RMB exchange rate based on modified PSO and GRNN

Abstract

Exchange rate adjustment is an important means for a country’s monetary policy, especially for countries that are dependent on the import and export trade, exchange rate policy has a strong impact on the profits of import and export traders, and therefore various countries are very prudent for the adjustment of the exchange rate, and even an exchange rate forecast system is provided for government policy-makers to make decisions. So, this paper proposes a new exchange rate prediction modeling technology. First, it collects data about Taiwan dollar (TWD) to USD and RMB exchange rate, and draws a trendency chart to explore the characteristics of both the exchange rate and possible future trends; then separately calculates technical indexes of TWD to USD and RMB exchange rate referring to stock technical index; finally, smoothing parameter $$\sigma $$ of two kinds of modified PSO-GRNN models are used to construct TWD to USD and RMB exchange rate prediction models, and compared with the other two kinds of prediction models. Research results demonstrate that TWD to USD and RMB exchange rate trends have different characteristics; we can see from four indexes that RWPSO-GRNN smoothing parameter $$\sigma $$ has better prediction ability than other three models.