Intelligent Soft Computing on Forex: Exchange Rates Forecasting with Hybrid Radial Basis Neural Network

Table 4

Predictive comparison of tested models, best configurations (ex-post).

Model	Regressor(s)	Weights adaptation	MSE	sd
RBF	Autoregressive (1)	Backpropagation	0.0000183763	0.0000028765
		-means + backpropagation	0.0000168649	0.0000002319
		Genetic algorithm	0.0000176860	0.0000006219
RBF-SMA	Autoregressive (1)	Back-Propagation	0.0000136485	0.0000005710
		-means + backpropagation	0.0000130549	0.0000003013
		Genetic algorithm	0.0000136146	0.0000003816
AR(0)-EGARCH(1,1,1)	Conditional variance (1)	Marquardt	0.0000170651	—
		Berndt-Hall-Hall-Hausman	0.0000170651	—

: mean squared error; : standard deviation.