On Fri, 26 Dec 2008, Ivan Sopov wrote: If you want to set the sample for a specific year, with dated daily data, I think you'll have to give the year-start and year-end dates. Unless, that is, you add a variable to the data set that records the year of the observation; then you could sample on that. When you load daily data in the first instance, you have the option of giving dates for all the observations, in which case gretl will not automatically insert missing values but will accept your observations as the full data record. In a loop context it seems that it would be easiest to add a year variable to the data set. Allin Cottrell