On Sat, 31 Mar 2012, Daniel Bencik wrote: Again, the skewed variant of the t density looks much better to me: <hansl> open http://eubie.sweb.cz/gretl_forum/allrng_allin.gdt include gig.gfn moo = gig_setup(allRng,1,const,null,7) gig_set_dist(&moo,3) gig_estimate(&moo) </hansl> yields: Model: GARCH(1,1) [Bollerslev] (Skewed T) Dependent variable: allRng Sample: 2007/11/20-2011/11/09 (T = 992), VCV method: Robust Conditional mean equation coefficient std. error z p-value ------------------------------------------------------- const 0.00358468 0.000457851 7.829 4.90e-15 *** AR1 0.129156 0.0313980 4.114 3.90e-05 *** AR2 0.121689 0.0311499 3.907 9.36e-05 *** AR3 0.0752985 0.0261992 2.874 0.0041 *** AR4 0.105514 0.0276418 3.817 0.0001 *** AR5 0.0667961 0.0273828 2.439 0.0147 ** AR6 0.0693248 0.0225564 3.073 0.0021 *** AR7 0.116931 0.0296532 3.943 8.04e-05 *** Conditional variance equation coefficient std. error z p-value --------------------------------------------------------- omega 6.65636e-07 4.21231e-07 1.580 0.1141 alpha 0.0373254 0.0165827 2.251 0.0244 ** beta 0.923552 0.0389155 23.73 1.68e-124 *** Conditional density parameters coefficient std. error z p-value ------------------------------------------------------ ni 10.3048 2.77794 3.709 0.0002 *** lambda 0.581766 0.0683239 8.515 1.67e-17 *** Llik: 4101.20099 AIC: -8176.40197 BIC: -8112.70557 HQC: -8152.18346 Riccardo (Jack) Lucchetti Dipartimento di Economia Università Politecnica delle Marche (formerly known as Università di Ancona) r.lucchetti(a)univpm.it http://www2.econ.univpm.it/servizi/hpp/lucchetti