Bonjour,

Je souhaite prédire l'évolution d'un indice avec un modèle TGARCH, de la forme:

Nom : Capture.PNG
Affichages : 12
Taille : 5,0 Ko

A l'aide du package ruGarch j'ai réussi à obtenir ceci:

*---------------------------------*
* GARCH Model Fit *
*---------------------------------*

Conditional Variance Dynamics
-----------------------------------
GARCH Model : fGARCH(1,1)
fGARCH Sub-Model : TGARCH
Mean Model : ARFIMA(1,0,1)
Distribution : std

Optimal Parameters
------------------------------------
Estimate Std. Error t value Pr(>|t|)
mu 0.000238 0.000169 1.4104 0.15843
ar1 0.483180 0.091529 5.2790 0.00000
ma1 -0.514822 0.089517 -5.7511 0.00000
omega 0.000302 0.000057 5.3405 0.00000
alpha1 0.085278 0.009815 8.6888 0.00000
beta1 0.910145 0.010019 90.8385 0.00000
eta11 1.000000 0.115417 8.6642 0.00000
shape 8.498012 1.208720 7.0306 0.00000

Robust Standard Errors:
Estimate Std. Error t value Pr(>|t|)
mu 0.000238 0.000158 1.5031 0.13282
ar1 0.483180 0.023371 20.6747 0.00000
ma1 -0.514822 0.025607 -20.1044 0.00000
omega 0.000302 0.000059 5.0949 0.00000
alpha1 0.085278 0.008110 10.5151 0.00000
beta1 0.910145 0.009498 95.8296 0.00000
eta11 1.000000 0.086956 11.5001 0.00000
shape 8.498012 1.285378 6.6113 0.00000

LogLikelihood : 9334.539

Information Criteria
------------------------------------

Akaike -6.0562
Bayes -6.0405
Shibata -6.0562
Hannan-Quinn -6.0506

Weighted Ljung-Box Test on Standardized Residuals
------------------------------------
statistic p-value
Lag[1] 0.01134 0.9152
Lag[2*(p+q)+(p+q)-1][5] 2.21836 0.9017
Lag[4*(p+q)+(p+q)-1][9] 5.12908 0.4248
d.o.f=2
H0 : No serial correlation

Weighted Ljung-Box Test on Standardized Squared Residuals
------------------------------------
statistic p-value
Lag[1] 2.517 0.1126
Lag[2*(p+q)+(p+q)-1][5] 3.574 0.3122
Lag[4*(p+q)+(p+q)-1][9] 4.724 0.4711
d.o.f=2

Weighted ARCH LM Tests
------------------------------------
Statistic Shape Scale P-Value
ARCH Lag[3] 0.07818 0.500 2.000 0.7798
ARCH Lag[5] 0.23947 1.440 1.667 0.9559
ARCH Lag[7] 1.40048 2.315 1.543 0.8411

Nyblom stability test
------------------------------------
Joint Statistic: 5.0391
Individual Statistics:
mu 0.1084
ar1 0.2526
ma1 0.2536
omega 0.4163
alpha1 0.2666
beta1 0.3127
eta11 1.2181
shape 0.1862

Asymptotic Critical Values (10% 5% 1%)
Joint Statistic: 1.89 2.11 2.59
Individual Statistic: 0.35 0.47 0.75

Sign Bias Test
------------------------------------
t-value prob sig
Sign Bias 0.7642 0.444828
Negative Sign Bias 0.6410 0.521541
Positive Sign Bias 3.1536 0.001628 ***
Joint Effect 10.8494 0.012569 **


Adjusted Pearson Goodness-of-Fit Test:
------------------------------------
group statistic p-value(g-1)
1 20 35.17 0.01333
2 30 41.09 0.06762
3 40 56.68 0.03339
4 50 56.17 0.22416


Elapsed time : 2.108773

Je suppose que les paramètres à utiliser sont dans la partie "Optimal Parameters" et sont donc:
mu = 0.000238
ar1 = 0.483180
ma1 = -0.514822
omega = 0.000302
alpha1 = 0.085278
beta1 = 0.910145
eta11 = 1.000000
shape = 8.498012
mais pourquoi autant de paramètres, je n'en ai besoin que de 4 normalement. J'ai essayé de prendre omega alpha, beta et eta mais la valeur du eta est beaucoup trop importante et fait exploser la valeur prédite.

Quelqu'un pour m'aider ?

Merci