Forecast and recommendation model

Forecast contract

• Horizon and creation time.
• Baseline and model version.
• Uncertainty interval.
• Inputs, exclusions, and history quality.

Recommendation

A recommendation links observed signal, suspected cause, possible action, expected effect, and risk. The user accepts, dismisses, or postpones with a reason.

Outcome

Actual results are compared with baseline after action. Without outcome measurement, usefulness cannot be claimed.

Synthetic demo interval semantics

The published demo keeps a daily value on a forecast date distinct from the sum of the full computed path over a period. Daily and cumulative intervals use completed historical errors for their matching horizon when sufficient calibration exists. The cumulative interval is not the sum of daily endpoints. Synthetic backtests demonstrate model behavior, not accuracy in a real seller business. A replenishment quantity remains an advisory draft with explicit supplier, inbound-stock and safety-stock assumptions; no order is placed.

Private lead-time comparison in the demo

In ProfitVena Intelligence, open the private what-if calculation and enable supplier lead time. Select a catalog model PV-001–PV-080 and 7, 14 or 30 days, then Calculate and save. The saved comparison uses current completed listing Forecast evidence: demand dates, stock, incoming supply, safety stock and supplier minimum/pack multiple. Positive order deficits are rounded up to those supplier terms. Each account has its own result and a link to its source Forecast.

The historical financial calculation and future replenishment comparison are separate. Changing supplier lead time does not recalculate past profit. Supplied demand bounds produce only a conditional order range, not a delivery guarantee or a recommendation to buy its upper bound. Missing intervals stay unavailable. The overlay is private to the visitor, expires after two hours, and changes no shared data, stored Forecast or marketplace order. AI checks saved inputs against current evidence before discussing quantities; recalculate if evidence becomes unavailable.