Forecasting for many businesses is done more by gut feel than science. How much product to buy, how much revenue to expect, how many calls will come in — these questions often are answered by some quick Excel formula that doesn’t account for trends or seasonality. We will examine the Holt-Winters exponential triple smoothing (ETS) algorithm for forecasting, both for its excellent results and for its ease of implementation. We will also dive into optimization algorithms — genetic, particle-swarm, and Nelder-Mead — to best identify initial constants.