Change search
ReferencesLink to record
Permanent link

Direct link
A comparison of the prediction performances by the linear models and the ARIMA model: Take AUD/JPY as an example
Umeå University, Faculty of Social Sciences, Umeå School of Business.
Umeå University, Faculty of Social Sciences, Umeå School of Business.
2007 (English)Independent thesis Advanced level (degree of Master (One Year)), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

With the development of the financial markets, the foreign exchange market has become more and more important for investors. The daily volume of business dealt with on the foreign exchange markets in 1998 was estimated to be over $2.5 trillion dollars (the daily volume on New York Stock Exchanges is about $20 billion). Today (2006) it may be about $5 trillion dollars. More and more people notice the foreign exchange market, and more and more sophisticated investors research such markets. The purpose of this thesis is to compare different methods to forecast the exchange rate of the money pair AUD/JPY. Firstly we studied the relationship between the AUD/JPY exchange rate and some economic fundamentals by using a regression model. Secondly, we tested whether the AUD/JPY exchange rate had any relationship with its historical records by using an ARIMA model. Finally, we compared the two model forecasting performance. A secondary purpose is to test whether the Market Efficiency Hypothesis works on the money pair AUD/JPY. In the study, data from January 1986 to June 2006 were chosen. To test which method produces better forecasts, we chose data from January 1986 to December 2002 to build up the prediction functions. Then we used the data from January 2003 to 2006 June to evaluate which predicting method was closer to the reality. In the comparison of the forecasting performances, two approaches dealing with the unknown future fundamentals were used. Firstly we assumed that we could do perfect predictions of these regressors, that was, our predictions of these regressors were the same as the actual future outcomes. So we put the real data for the fundamentals from January 2003 to June 2006 into the regression function. Secondly we assumed that we were in real life situation, and we had to predict the regressors first in order to get the predictions of the exchange rate. The results of the comparison were that the AUD/JPY exchange rate could to some extent be predictable, and that the predictions by the ARIMA model were more accurate.

Place, publisher, year, edition, pages
Umeå: Handelshögskolan vid Umeå universitet , 2007. , 56 p.
Keyword [en]
Foreign exchange market, AUD/JPY, regression model, ARIMA model
National Category
Economics and Business
URN: urn:nbn:se:umu:diva-1047OAI: diva2:140033
Available from: 2007-03-20 Created: 2007-03-20Bibliographically approved

Open Access in DiVA

fulltext(1062 kB)2250 downloads
File information
File name FULLTEXT01.pdfFile size 1062 kBChecksum MD5
Type fulltextMimetype application/pdf

By organisation
Umeå School of Business
Economics and Business

Search outside of DiVA

GoogleGoogle Scholar
Total: 2250 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 1012 hits
ReferencesLink to record
Permanent link

Direct link