Exchange rate forecasting is an important financial problem that is receiving increasing attention nowadays especially because of its difficulty and host of practical applications in globalising world of today. The paper presents an enhanced MIA-GMDH-type network, discusses its design methodology and carries out some numerical experiments in the field of exchange rate forecasting. The method presented in this paper is an enhancement of self-organizing polynomial Group Method of Data Handling (GMDH) with several specific improved features - coefficient rounding and thresholding schemes and semi-randomized selection approach to pruning. The experiments carried out include exchange rate prediction and hedging case study where the predictions were used for financial management decision simulation of a virtual company. The results indicate, that the method shows promising potential of self-organizing network methodology. This implies that the proposed modelling approaches can be used as a feasible solution for exchange rate forecasting in financial management.
[1]
Frank Lemke,et al.
Modeling tool wear in end-milling using enhanced GMDH learning networks
,
2008
.
[2]
Sung-Kwun Oh,et al.
Self-Organizing Polynomial Neural Networks Based on Genetically Optimized Multi-Layer Perceptron Architecture
,
2004
.
[3]
Kin Keung Lai,et al.
Foreign-Exchange-Rate Forecasting with Artificial Neural Networks
,
2007
.
[4]
M. Snorek,et al.
Hybrid inductive models: deterministic crowding employed
,
2004,
2004 IEEE International Joint Conference on Neural Networks (IEEE Cat. No.04CH37541).
[5]
F. Lemke,et al.
GMDH algorithms for complex systems modelling
,
1998
.
[6]
A. G. Ivakhnenko,et al.
Polynomial Theory of Complex Systems
,
1971,
IEEE Trans. Syst. Man Cybern..
[7]
Hitoshi Iba,et al.
Genetic programming using a minimum description length principle
,
1994
.