Forecasting exchange rates with linear and nonlinear models

Rakesh Bissoondeeal*, Jane M. Binner, Muddun Bharuth, Alicia M. Gazely, Veemadevi P. Mootanah

*Corresponding author for this work

Research output: Contribution to journalArticle

Abstract

In this paper, the exchange rate forecasting performance of neural network models are evaluated against the random walk, autoregressive moving average and generalised autoregressive conditional heteroskedasticity models. There are no guidelines available that can be used to choose the parameters of neural network models and therefore, the parameters are chosen according to what the researcher considers to be the best. Such an approach, however, implies that the risk of making bad decisions is extremely high, which could explain why in many studies, neural network models do not consistently perform better than their time series counterparts. In this paper, through extensive experimentation, the level of subjectivity in building neural network models is considerably reduced and therefore giving them a better chance of performing well. The results show that in general, neural network models perform better than the traditionally used time series models in forecasting exchange rates.

Original languageEnglish
Pages (from-to)414-429
Number of pages16
JournalGlobal Business and Economics Review
Volume10
Issue number4
DOIs
Publication statusPublished - 2008

Fingerprint

Network model
Exchange rate forecasting
Neural networks
Autoregressive conditional heteroskedasticity
Forecasting performance
Random walk
Experimentation
Autoregressive moving average
Subjectivity
Time series models

Keywords

  • ARIMA
  • Autoregressive Integrated Moving Average
  • exchange rates
  • forecasting
  • GARCH
  • Generalised Autoregressive Conditional Heteroskedasticity
  • linear and nonlinear models
  • Neural Networks
  • NNs

Cite this

Bissoondeeal, Rakesh ; Binner, Jane M. ; Bharuth, Muddun ; Gazely, Alicia M. ; Mootanah, Veemadevi P. / Forecasting exchange rates with linear and nonlinear models. In: Global Business and Economics Review. 2008 ; Vol. 10, No. 4. pp. 414-429.
@article{183310e910074cd7b99d942855348114,
title = "Forecasting exchange rates with linear and nonlinear models",
abstract = "In this paper, the exchange rate forecasting performance of neural network models are evaluated against the random walk, autoregressive moving average and generalised autoregressive conditional heteroskedasticity models. There are no guidelines available that can be used to choose the parameters of neural network models and therefore, the parameters are chosen according to what the researcher considers to be the best. Such an approach, however, implies that the risk of making bad decisions is extremely high, which could explain why in many studies, neural network models do not consistently perform better than their time series counterparts. In this paper, through extensive experimentation, the level of subjectivity in building neural network models is considerably reduced and therefore giving them a better chance of performing well. The results show that in general, neural network models perform better than the traditionally used time series models in forecasting exchange rates.",
keywords = "ARIMA, Autoregressive Integrated Moving Average, exchange rates, forecasting, GARCH, Generalised Autoregressive Conditional Heteroskedasticity, linear and nonlinear models, Neural Networks, NNs",
author = "Rakesh Bissoondeeal and Binner, {Jane M.} and Muddun Bharuth and Gazely, {Alicia M.} and Mootanah, {Veemadevi P.}",
year = "2008",
doi = "10.1504/GBER.2008.020593",
language = "English",
volume = "10",
pages = "414--429",
journal = "Global Business and Economics Review",
issn = "1097-4954",
publisher = "Inderscience Enterprises Ltd.",
number = "4",

}

Forecasting exchange rates with linear and nonlinear models. / Bissoondeeal, Rakesh; Binner, Jane M.; Bharuth, Muddun; Gazely, Alicia M.; Mootanah, Veemadevi P.

In: Global Business and Economics Review, Vol. 10, No. 4, 2008, p. 414-429.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Forecasting exchange rates with linear and nonlinear models

AU - Bissoondeeal, Rakesh

AU - Binner, Jane M.

AU - Bharuth, Muddun

AU - Gazely, Alicia M.

AU - Mootanah, Veemadevi P.

PY - 2008

Y1 - 2008

N2 - In this paper, the exchange rate forecasting performance of neural network models are evaluated against the random walk, autoregressive moving average and generalised autoregressive conditional heteroskedasticity models. There are no guidelines available that can be used to choose the parameters of neural network models and therefore, the parameters are chosen according to what the researcher considers to be the best. Such an approach, however, implies that the risk of making bad decisions is extremely high, which could explain why in many studies, neural network models do not consistently perform better than their time series counterparts. In this paper, through extensive experimentation, the level of subjectivity in building neural network models is considerably reduced and therefore giving them a better chance of performing well. The results show that in general, neural network models perform better than the traditionally used time series models in forecasting exchange rates.

AB - In this paper, the exchange rate forecasting performance of neural network models are evaluated against the random walk, autoregressive moving average and generalised autoregressive conditional heteroskedasticity models. There are no guidelines available that can be used to choose the parameters of neural network models and therefore, the parameters are chosen according to what the researcher considers to be the best. Such an approach, however, implies that the risk of making bad decisions is extremely high, which could explain why in many studies, neural network models do not consistently perform better than their time series counterparts. In this paper, through extensive experimentation, the level of subjectivity in building neural network models is considerably reduced and therefore giving them a better chance of performing well. The results show that in general, neural network models perform better than the traditionally used time series models in forecasting exchange rates.

KW - ARIMA

KW - Autoregressive Integrated Moving Average

KW - exchange rates

KW - forecasting

KW - GARCH

KW - Generalised Autoregressive Conditional Heteroskedasticity

KW - linear and nonlinear models

KW - Neural Networks

KW - NNs

UR - http://www.scopus.com/inward/record.url?scp=84969452171&partnerID=8YFLogxK

UR - http://www.inderscience.com/offer.php?id=20593

U2 - 10.1504/GBER.2008.020593

DO - 10.1504/GBER.2008.020593

M3 - Article

VL - 10

SP - 414

EP - 429

JO - Global Business and Economics Review

JF - Global Business and Economics Review

SN - 1097-4954

IS - 4

ER -