Interface: ForecastAbstention
@kortexya/reasoninglayer / ForecastAbstention
Interface: ForecastAbstention
Section titled “Interface: ForecastAbstention”Defined in: src/types/forecast.ts:329
An abstention — the model declined to forecast.
Remarks
Section titled “Remarks”reason is the machine-readable cause. The reasons the backend emits today
are:
model_unavailable— no trained model is loaded for this domain.slice_too_small— the relevant conformal slice had too few calibration examples.stale_data— the freshest inputs are older than the decision policy allows.low_confidence— the model’s confidence fell below the commit threshold.missing_market— the domain has a market but odds were missing at decision time.
It is typed as string rather than a closed union because the backend
publishes the forecast payload as a free-form JSON object: a reason added
server-side must not break clients built against an older spec snapshot.
Example
Section titled “Example”if (forecast.outcome === 'abstained') { console.log(`declined (${forecast.reason}): ${forecast.detail}`);}Properties
Section titled “Properties”detail
Section titled “detail”detail:
string
Defined in: src/types/forecast.ts:334
Context (e.g. which slice, how few samples).
reason
Section titled “reason”reason:
string
Defined in: src/types/forecast.ts:331
The machine reason (see the remarks for the reasons emitted today).