Font Size: a A A

Forecasting Exchange Rate Based On Smoothing Spline Filtering And RBF Neural Network

Posted on:2012-08-04Degree:MasterType:Thesis
Country:ChinaCandidate:F YangFull Text:PDF
GTID:2249330374495916Subject:Management Science and Engineering
Abstract/Summary:PDF Full Text Request
With China joining World Trade Organization and reforming the exchange rate system, the behavior of RMB exchange rate becomes more and more complicated. Because of this, traditional technical analysis and fundamental analysis were hard to catch all the characteristic of exchange rate and forecast it accurately. Besides, considering that the European Union has become the biggest trading partner of our country for years and the volume of trade between the two sides has been increasing each year, this paper proposes a combinative model based on Smoothing Spline filtering and RBF neural network to catch the characteristic of RMB/EUR exchange rate better and incease the precision of forecasting model.This paper firstly reviews the development of exchange rate forecasting theory and techniques, specifically analyzes the application of neural network and non-parametric estmation, especially in exchange rate forecasting field. Then proposes a multiple levels decomposition aglorithm based on the theory of Smoothing Spline filtering, and combine this aglorithm with RBF neural network to construct the combinative forecasting model. Finally the combinative forecasting model is tested by empirical research with the RMB/EUR exchange rate. The empirical results show that the proposed model outperforms the BP neural network model and single RBF neural network model.This thesis applies Smoothing Spline filtering and RBF neural network to forecast the RMB/EUR exchange rate. The results not only has important theoretical value and practical meanings for policy maker to deeply understand the behavior characteristic of exchange rate and to make the macroeconomic policy related to exchange rate, but for enterprise and investor to operate and avoid risk and for investor to make better investment strategies.
Keywords/Search Tags:Exchange rate forecasting, Time series analysis, Smoothing splinefiltering, RBF neural network
PDF Full Text Request
Related items