This forecast method is applicable to demand patterns that exhibit level, trend, and seasonality. It smooths average, trends, and seasonality and takes seasonal variations into account by assigning them a seasonal index, which is updated monthly, to each forecast slice. In turn, each seasonal index is smoothed. This produces a very accurate forecast. It is intended for implementations where some parts exhibit seasonal demand. Winters Multiplicative requires a minimum of 12 months of demand history. If less than 12 months of demand history are available, the application will use the Average forecast method and post a Review Board review reason.

The method is based on three smoothing equations, one for level, one for trend, and one for seasonality. Corresponding to each equation is a smoothing parameter. Due to the recursive nature of smoothing equations, an initialization method is used to generate the initial level, trend, and seasonality values. The initial values are then updated by applying the smoothing parameters as the actual demand in the latest period is observed.

The WINTERS_VERSION global setting controls the inputs to the calculation.