On Mon, 2 Jun 2014, Daniel Bencik wrote: I think there's a "*" missing in the rightmost element of the loglikelihood. The following seems to work quite nicely here (of course, this isn't supposed to be right, it's just something you can take inspiration from). <hansl> function series loglik(series y, scalar theta, scalar c, scalar a, scalar b, scalar m0) series ret = NA if (a<=0) || (b<0) return ret endif series m = m0 series m = c + b*m(-1) + a*y(-1) series k = 0.5*ln(theta) - theta*m series ret = k + y*(ln(y)-1) - lngamma(y+1) + theta*y*(1 + ln(m/y)) return ret end function include yahoo_get.gfn nulldata 2500 setobs 5 2005-01-01 list X = yahoo_full("IBM") range = gap_filler(IBM_h - IBM_l, 2) scalar a = 0.1 scalar b = 0.8 scalar c = 0.01 scalar theta = 1 #scalar gamma = 1 scalar m0 = mean(range) mle ll = loglik(range, theta, c, a, b, m0)     params theta c a b end mle --robust --verbose </hansl> ------------------------------------------------------- Riccardo (Jack) Lucchetti Dipartimento di Scienze Economiche e Sociali (DiSES) Università Politecnica delle Marche (formerly known as Università di Ancona) r.lucchetti(a)univpm.it http://www2.econ.univpm.it/servizi/hpp/lucchetti -------------------------------------------------------