Demand Forecast Calculator

Each method weights history differently: the simple moving average treats the last three months equally, the weighted average favors the most recent month, exponential smoothing blends all history with a decaying weight set by alpha, and the trend forecast extrapolates a straight line through all six points. When demand is flat, the four converge closely; when it's trending or noisy, they diverge more.

A higher alpha, closer to 1, makes the forecast react quickly to recent months but also makes it more sensitive to one-off noise or a fluke month. A lower alpha, closer to 0, produces a smoother forecast that changes gradually, which suits demand that's genuinely stable but lags behind a real, sustained shift in demand.

It diverges most when demand shows a consistent upward or downward trajectory across all six months, since it's the only method that explicitly fits and projects a slope rather than averaging recent values. The moving-average methods will lag behind a real trend because they're always looking backward at recent averages instead of extrapolating a direction.

Six months gives enough history to compute a meaningful three-month moving average, a weighted average, a full exponential smoothing chain, and a linear trend, without requiring years of historical records that many smaller operations don't have readily available. More historical months would generally improve trend and seasonality detection, but this is a practical middle ground.

Not on their own — none of the four methods here explicitly models seasonal patterns, so a product that spikes every December or dips every summer will show that pattern as noise or a false trend rather than a recognized seasonal cycle. For strongly seasonal demand, compare the same calendar month across multiple years rather than relying on six consecutive months.