Joint-variance scoring across a portfolio of spots
Score N spots at once and get back not just per-spot verdicts but the joint distribution + per-spot variance contribution. Designed for operators running multi-location bookings and parametric underwriters pricing diversified covers.
Pro plan and above. Up to 50 spots per request. Heavier than running N independent single-score calls because the engine computes cross-spot correlation; budget ~2s per 10 spots in the request.
What it returns
Per-spot scores (same shape as the single-score endpoint) plus a portfolio-level summary: joint expected score, joint variance, per-spot variance contribution. The variance contribution tells you which spot drives the most uncertainty in your portfolio's joint outlook — useful for booking-engine UI that prioritises which spot forecasts to refresh fastest.
Request
{
"spots": [
{ "location": { lat, lng }, "activity": "kitesurfing" },
{ "location": { lat, lng }, "activity": "windsurfing" },
… up to 50 spots
],
"window": { from, to },
"correlationModel": "exponential",
"correlationLengthKm": 50
}Correlation models
none — treat spots as independent. Simplest, fastest; good when your spots are climatologically very different (Alps + Mediterranean + Atlantic).
exponential (default) — covariance decays exponentially with Haversine distance, parameterised by correlationLengthKm (default 50km). Good when your spots cluster in one region (a fleet of Tarifa schools, a Costa Blanca chain).
Use cases
- Multi-location booking — show users which spot in your network is best today + a "you're choosing between these" summary.
- Parametric pricing — feed the joint variance into a portfolio policy quote (the underwriting endpoints accept portfolio input directly; this is the unbundled version).
- Capacity planning — which spots can absorb traffic when the headline location is unfavorable?