Modeling Markov Switching ARMA-GARCH Neural Networks Models and an Application to Forecasting Stock Returns

Table 3

MS-ARMA-GARCH models.

MS-ARMA-GARCH models
(1) MS-ARMA-GARCH	Garch	Sigma	Arch	Constant	𝑝_{0∣0}	𝑃_{1∣1}			log𝐿	RMSE
Regime 1.	0.966483; (0.01307)***	0.000333727; (1.916𝑒−006)***	0.03351; (0.005)	6.23008𝑒−005; (1.360𝑒−005)***	0.500244	0.5102			385.09	0.458911
Regime 2.	0.436124; (0.01307)***	0.0004356; (1.231𝑒−005)***	0.56387; (0.0098)	6.29344𝑒−005; (1.161𝑒−005)***
(2) MS-ARMA-APGARCH	Garch	sigma	Arch	Constant	Mean	𝑃_{0∣0}	𝑝_{1∣1}	Power	log𝐿	RMSE
Regime 1.	0.616759; (0.01307)***	0.000679791 (5.685𝑒−006)***	0.383241; (0.0102)	8.13394𝑒−005; (2.007𝑒−005)***	0.00021042	0.500227	0.50300	0.80456; (0.00546)	1756.5	0.42111
Regime 2.	0.791950; (0.01307)***	0.0012381; (3.451𝑒−004)***	0.20805; (0.0201)	8.20782𝑒−005; (1.835𝑒−005)***	2.83𝐸−03			0.60567; (0.0234)
(3) MS-ARMA-FIAPGARCH	ARCH; (Phi1)	GARCH; (Beta1)	d-Figarch	APARCH; (Gamma1)	APARCH; (Delta)	Cst	𝑝_{0∣0}	𝑝_{0∣1}	log𝐿	RMSE
Regime 1.	0.277721; (0.00)	0.67848; (0.00)	0.2761233; (0.0266)	0.220157; (0.0106)	0.123656 (0.001)	0.00135; (0.001)	0.50212	0.510021	1877.9	0.4222066
Regime 2.	0.309385; (0.002)	0.680615; (0.0001)	0.181542; (0.00005)	0.21083; (0.0299)	0.1448; (0.0234)	0.00112; (0.00984)