A technique used for time series forecasting that begins with an exponentially-smoothed current forecast and adjusts for trend by using a second smoothed estimate.