Forecast Markov switch probability

[Nicotrader87]

Hello everyone,

re-propose my problem as the previous discussion was closed without anyone helping me to solve the problem, much less giving an explanation of why.

As a user of the software I believe that the community and the employees of eviews should give the requested support, as an alternative it will be my concern to orient myself towards other software.

Qui trovate il modello da me stimato:

Estimation Command:
=========================
SWITCHREG(RNG=KN, SEED=406193532) SERIES01 C

Estimation Equation:
=========================
1: SERIES01 = C(1)

2: SERIES01 = C(2)

SIGMA = @EXP(C(3))

Substituted Coefficients:
=========================
1: SERIES01 = 0.126675229233

2: SERIES01 = 0.883723555728

SIGMA = @EXP(-2.6902295801)

Dependent Variable: SERIES01
Method: Simple Switching Regression (BFGS / Marquardt steps)
Date: 08/13/19 Time: 13:52
Sample: 3000 3312
Included observations: 313
Number of states: 2
Standard errors & covariance computed using observed Hessian
Random search: 25 starting values with 10 iterations using 1 standard
deviation (rng=kn, seed=406193532)
Convergence achieved after 2 iterations

Variable Coefficient Std. Error z-Statistic Prob.

Regime 1

C 0.126675 0.003990 31.74567 0.0000

Regime 2

C 0.883724 0.015273 57.86159 0.0000

Common

LOG(SIGMA) -2.690230 0.040226 -66.87790 0.0000

Probabilities Parameters

P1-C 2.684443 0.232604 11.54084 0.0000

Mean dependent var 0.175049 S.D. dependent var 0.197513
S.E. of regression 0.198149 Sum squared resid 12.17160
Durbin-Watson stat 0.066217 Log likelihood 323.5584
Akaike info criterion -2.041907 Schwarz criterion -1.994032
Hannan-Quinn criter. -2.022775



For this model I need to get a forecast of the smoothed probabilities of the series, no fitted series.
is there a command to enter?

[EViews Gareth]

You received an answer in your previous thread.

Further, this forum is for users to help each other out on a voluntary basis. If you wish to receive dedicated support you should contact the EViews support team (support@eviews.com) and provide your EViews serial number (Help->About EViews).