Demand planning you can actually explain
Fintra forecasts demand from your real outflow history using transparent models - moving average, trend, seasonal, and Croston’s for lumpy items - auto-selected by walk-forward backtest.
Illustrative product view
Forecasts built from your history
Fintra builds a demand series from your actual outflow transactions - issues to WIP, sales, scrap, and negative adjustments - so the forecast reflects how an item truly moves, not a guess. It’s deterministic, not a black box: you can see which model was chosen and why.
- Demand is derived from real inventory outflows, per item.
- Multiple models are fit and compared, not one imposed.
- The forecast returns the method used, mean demand, demand variability, and a confidence signal.
The models and how one is chosen
| Model | Best for |
|---|---|
| Moving average | Stable demand |
| Trend (linear regression) | Steadily rising or falling demand |
| Naive seasonal | Repeating seasonal patterns |
| Croston’s method | Intermittent, lumpy demand |
Demand feeds the planning math
The forecast isn’t an end in itself - its mean demand and variability feed reorder points, safety stock, and stockout-risk analysis. A good demand model is what makes those recommendations trustworthy.
Why transparency beats a black box
A forecast you can’t explain is hard to trust and impossible to defend. Because Fintra shows the chosen model, the demand mean, and the backtest error, a planner can sanity-check a recommendation instead of taking it on faith - which is what makes people actually use it.
Frequently asked questions
How does demand planning work in Fintra?
Fintra builds a demand series from your real outflow transactions - issues to WIP, sales, scrap, and negative adjustments - then fits several transparent models and auto-selects the best by a walk-forward backtest. It returns the chosen method, mean demand, demand variability, and a confidence signal per item.
What forecasting models does Fintra use?
Moving average for stable demand, trend via linear regression for rising or falling demand, naive seasonal for repeating patterns, and Croston’s method for intermittent, lumpy demand. Smooth models are chosen by backtest error, and lumpy items are routed to Croston’s.
How is intermittent demand handled?
Items whose demand is zero much of the time - spare parts, slow SKUs - break ordinary averaging. Fintra detects that intermittency and uses Croston’s method, which separately models the size of demand and the interval between demands, for a more sensible forecast.
Is the forecast a black-box AI model?
No. The forecasting is deterministic and explainable: you can see which model was selected, the mean demand and variability, and the backtest error behind the choice. That transparency lets planners sanity-check the numbers rather than trusting an opaque model.
Stay in the loop
One practical finance briefing a week - new guides, checklists, and benchmarks.
Forecast demand you can trust
Start free, no card required. Let Fintra build a forecast from your own history.
Talk to us