No forecast is a single number: it is a range.
The StockAI forecast comes with a band around it. This chapter explains where it comes from, what it means, and why it widens the further ahead you look.
The middle number and the band
The forecast line is the most likely value. But we have already seen that there is always noise: actual sales almost never land exactly on the line. The band shows the range within which it is reasonable for sales to fall.
In StockAI the band runs from 5% to 95%: if the past repeats itself, actual sales should land inside it about nine times out of ten, and above the top edge about one time in twenty.
It comes from real errors, not a formula
There are two ways to draw that band. One is to assume the errors follow a bell curve and calculate. The other, the one StockAI uses, is to look at the errors the model actually made on the hidden dates (chapter 2) and see how large they were.
If in those tests 90% of the one-week errors fell between −12 and +15 units, the one-week band is that range. It assumes nothing about the shape of the errors: if your product is wrong more on the high side than the low side, the band shows it.
Why it widens
Forecasting tomorrow is easier than forecasting a month ahead. That is why errors are measured separately for each distance: one, two, three periods ahead. And the ones further away are almost always larger, so the band opens up as it goes.
Move the simulator: with a more irregular product, the whole band widens. That is the information a single number hides.
The band is built from the errors
SimulatorAn invented eight-week history and a two-week forecast. The band comes from testing the forecast on the past and measuring how far off it was at each distance.
Simulation with invented numbers. Here the "model" is a simple average, so the example is clear; StockAI does the same with the winning model of each product.
What it is for
The band is not decoration. It tells you how much to trust each product. Two products with the same forecast of 100 units are not equal if one has a band from 90 to 110 and the other from 40 to 170. The second needs more cushion, and chapter 6 explains how that is calculated.
What you take away
- The forecast is a most-likely value with a 5% to 95% band around it.
- The band comes from the real errors the model made on the hidden dates, not from an assumed curve.
- It widens the further ahead you forecast.
- A wide band means more uncertainty and therefore more cushion.
Where you see it in StockAI
On each product chart in Forecasts, the shaded area around the line is this band.
Keep reading
- From forecast to purchaseService level, cushion, reorder point and the lead time learned from your receipts.
- LearnA guide for non-technical readers: what a time series is, how a forecast is measured, what each model does, intermittent demand, the forecast range, and how it turns into a purchase.
- How it is calculated, in briefThe best way to understand it is to see it with your own products.
Put your own AI to work forecasting your sales. Today.
Three ways to start, depending on how much you want to commit right now. All three lead to the same product.
Create your account
With the whole engine. Upload your sales file and see your first forecast, and with it your first list of what to order.
Create my accountLook before giving your email
An instant trial account with sample data. It lasts 24 hours and is then erased.
Try it without signing upTalk to us
For the corporate plan, the source code, or to see it with your data and someone beside you. We answer within 24 hours.
Sales and contact: contacto@stockai.es. Phone: +506 7186 2820.
Sales forecasting made in Costa Rica, for companies across Latin America.