Valuation range for a locality, no address needed (Pro+ tier)

Costs 2 credits. Give a suburb/postcode plus attributes (beds, baths, land size, property type) instead of a street address, and get back a value range and confidence. "What would a 4-bed house on ~600sqm in this postcode go for." Unlike `POST /v1/valuations`, which resolves a real address and takes ~30 minutes, this answers synchronously and doesn't touch anything outside our own data. With no address, we can't pin the estimate to a parcel, only a locality, so `confidence` is always `"Low"` (see `coordinate_basis`). **Branch on `precision`, not `confidence`.** `precision` bands how tight this particular locality's range is, and it moves: on an 8,000-sale temporal holdout, median absolute error ran ~7% for `high`, ~11% `medium`, ~14% `low` and ~20%+ `very_low`. A uniform master-planned suburb usually bands `high`; an architecturally mixed inner suburb bands `low` or `very_low`. If there's nothing nearby to compare against, `estimated_value` and its range come back null with `reason: "insufficient_comparables"` instead of a guess. Pro+ tier. If you're prototyping on a lower plan before launch, ask us and we can enable it on your key.

curl -X POST "https://api.propradar.com.au/v1/valuations/estimate" \
  -H "X-API-Key: pr_live_..." \
  -H "Content-Type: application/json" \
  -d '{
  "suburb": "Pascoe Vale",
  "postcode": 3044,
  "state": "VIC",
  "property_type": "House",
  "bedrooms": 3,
  "bathrooms": 1,
  "land_size": 550
}'

{ "query": { "suburb": "Pascoe Vale", "postcode": 3044, "state": "VIC",
              "property_type": "House", "bedrooms": 3, "bathrooms": 1, "land_size": 550 },
  "estimated_value": 1031701, "estimated_value_low": 832464, "estimated_value_high": 1259106,
  "confidence": "Low", "precision": "medium", "comp_count": 40,
  "coordinate_basis": "suburb_centroid", "reason": null }