SA2s, not SA1s
The dashboard map is drawn from ABS ASGS Edition 3 SA2 2021 boundaries. The price target, forecast ranking and driver stack are all reported at SA2 / property-type level.
It is a 10-year scenario forecast
The model starts from the latest SA2 median sale-price seed and compounds quarterly forecast growth through the configured horizon. It is a screening model for relative structural growth, not a property valuation.
Forecast Flow
How raw data becomes the map ranking.
1Sales target
Valuer General transactions are cleaned, geocoded, assigned to SA2 and aggregated to quarterly median prices.
2Local features
SA2 population, workforce, household demand, supply, rent proxies, census baselines and event overlays are aligned to quarter.
3Historical validation
Factors are screened against future 1Q, 4Q and 8Q price growth and event exposures are compared against control areas.
4Forecast overlay
Macro scenario anchor plus local structural overlay plus local price momentum produces quarterly SA2 growth.
5Map output
Quarterly growth is compounded into 10-year growth, annualised growth and forecast median price for each SA2.
Model Logic
The current SA2 tool is deliberately transparent and driver-based.
forecast_price_t+h = seed_price_t * exp(sum quarterly_growth)quarterly_growth = macro_anchor + demand + jobs + supply + tightness + valuation + momentum + liquidity + geo_eventhousehold_demand = population / average_household_sizehousehold_gap = household_demand - dwelling_stockdwelling_stock_t = dwelling_stock_t-1 + completions - demolitions_assumptionConfigured Weights
These are the live settings used by the dashboard build.
Live Feature Inputs
Every model feature group currently present in the SA2 feature matrix.
| Feature group | Features | Source level | Source | Frequency / treatment |
|---|
Source Register
Live configured sources, their geography and caveats.
What Is Still Approximate
Important interpretation boundaries.
Not SA1
The dashboard does not rank individual SA1s or parcels. It ranks SA2s, which are larger neighbourhood-scale statistical areas.
Some inputs are projections
TfNSW and NSW CPA projection inputs are planning assumptions. They are flagged separately from observed historical SA2 series.
Rent is not fully SA2-native yet
The optional rental-bond SA2 index exists separately; the active core rent feature is still an LGA proxy unless switched on in config.
Geo-event effects are controls/priors
Flood and bushfire are risk controls. Transport and project catalysts are not automatically positive everywhere and are checked against historical evidence.