On Tue, 27 Feb 2018, Mike Lasarev wrote: These functions in effect assume stationarity of the data, so No, the parameters cannot take the form of series. But it's not difficult to write your own function for the non-stationary case (if it can't be handled just by differencing). Here's a trivial example: <hansl> function series ns_cdf (const series y, const series mu) series c loop t=$t1..$t2 -q c[t] = cdf(z, y[t] - mu[t]) endloop return c end function nulldata 10 setobs 1 2000 genr time series y = time + normal() series c = ns_cdf(y, time) print y c --byobs <hansl> In this case it would, of course, be easier just to do series c = cdf(z, y-time) but I imagine you can see how more complex cases could be handled in analogous fashion. Allin Cottrell