Estimate safeguards
Automated tests check input handling, comparable-sale selection contracts, null-distance behaviour, report access and release safeguards.
Current position: the product is suitable for early research and screening. It is not a certified valuation and no public claim of a fixed accuracy percentage is made.
What is currently validated
Opportunity contract
Public scores use a consistent 0–100 display, retain missing values as unavailable and identify the model horizon and non-forecast status.
Environment isolation
Database environment guards are designed to fail closed and keep Preview and Production selection rules separate.
Release checks
Syntax, contract, regression and production-build checks are run before a release is approved.
What is not yet proven
- No single accuracy rate applies to every suburb, property type or evidence level.
- Opportunity scores have not been presented as calibrated probabilities of capital growth.
- Planning signals do not replace a current property-specific planning review.
- Historical market relationships may change with rates, policy, supply and buyer behaviour.
How to interpret confidence
Confidence describes evidence availability and model coverage. It is not the probability that a sale falls inside the displayed range. A high-confidence result can still be wrong when condition, renovation quality, title constraints or off-market terms are missing.
Next validation standard
Future accuracy reporting should use time-separated holdout sales, report median absolute percentage error by property type and geography, disclose sample coverage, and compare results with simple suburb-median and recent-comparable baselines. Results should be published only after leakage and survivorship checks.
Safe use
Use the output to form a shortlist and identify questions. Confirm decisions using current contracts, title and planning material and appropriately qualified property, legal and financial professionals.