Prediction Results (`/predictions`) compares Historical observations to the forecast band for an Asset series.
· Fit language is qualitative (for example fitting vs insufficient). There is no public master score.
· You need Mode 1 Projection plus Historical points before fit can be scored.
· Customer risk reports may sample up to three Assets. That sample is not a full portfolio valuation.
If results say insufficient, add Historical observations or finish Projection — do not treat empty fit as “low risk.”