The methods behind your next order
Stock Perfect builds a forecast for each product, then turns that forecast into an order quantity using your stock position and supplier rules.
OptimizedTheta and AutoETS
For smooth and erratic demand, the engine combines forecasts from OptimizedTheta and AutoETS. OptimizedTheta models the level and trend of sales. AutoETS fits exponential-smoothing models. Both run on sales after the engine has accounted for applicable seasonal patterns. With shorter histories, the forecast also blends in a recent-sales average so a brief launch spike has less influence.
These models use the open-source StatsForecast library. Read its model documentation.
A method matched to the product
A product that sells occasionally has a different sales pattern from a product that sells every day. The engine routes intermittent and lumpy demand through methods built for that pattern, including time aggregation. New products use their own sales together with a prior estimated from comparable products. Inactive products are handled separately.
Seasonality and stockouts
When the history supports a seasonal pattern, the engine removes that pattern before fitting the underlying demand, then applies it to the future dates. Days when stock was unavailable are handled separately from genuine days without demand, so selling out is not treated as evidence that customers stopped buying.
From forecast to order
The order calculation combines forecast demand during the supplier lead time and review period, the safety buffer, stock on hand, committed stock and relevant incoming orders. Supplier minimums and case packs are applied to the final quantity. The product screen shows the calculation so you can review the inputs and the result.
Measured against actual sales
As forecast windows finish, Stock Perfect compares forecasts with actual sales and a recent-sales baseline. Product-level results include forecast error and interval coverage when enough scored history is available. Forecasts estimate future demand; the strength of a product's result depends on its history and changes in the business.