CMA:
matching comparable sales on 30+ features.
A comparable sale isn't a property nearby. It's a property nearby with similar specs, sold recently enough that the market hasn't moved since. SuburbIQ's structural adjustment engine matches each comp on 30+ features and reweights the sale price accordingly.
The features we match on.
Land area, frontage, depth, lot shape (battle-axe, corner, irregular). Bedrooms, bathrooms, parking spaces, carport vs garage. Internal floor area, number of living rooms, kitchen layout. Construction era (decade band), construction material (brick, weatherboard, hebel, concrete). Roof type, foundation type. Pool, granny flat, deck, fence type. School catchment ID. Distance to train station, distance to CBD. Flood zone, bushfire zone, noise band. Heritage overlay, zoning code. Solar, smart-meter. Renovation flag, condition band. That's 30+ — the exact count varies by region and data availability.
How adjustment works.
For each comp, we compute a vector of differences from the subject property — for example, the comp has one extra bedroom, 50m² more land, no pool. The HTAG structural-adjustment endpoint applies a regression-derived $/feature coefficient to each difference and reweights the comp's sale price to "what would this comp have sold for if it matched the subject exactly?" The final fair-value range is the median of three such adjusted comps, with an inter-quartile band as the confidence interval.
What we don't match on.
Aesthetic and condition factors not in the dataset — kitchen quality, bathroom updates, finishes, curb appeal, listing photography. These can affect realised sale prices by 5-10%. We disclose this band in the report as the "presentation premium/discount" and don't adjust for it.
Why this beats raw median.
Suburb median is a 12-month rolling average of all sales, regardless of property type. In a suburb with mixed stock — 2-bed units alongside 5-bed houses — the median bears no relationship to any specific parcel. Comp-led structural adjustment respects the actual parcel economics: similar land, similar improvement value, similar exposure.
When CMA returns "MISSING."
If the structural-adjustment endpoint finds fewer than three comps within 1 km that meet a minimum similarity threshold, the report degrades gracefully — we widen the radius to 2 km, drop the strictest filters, and re-run. If still under three matches, the report hedges the verdict and flags the missing data prominently. We do not silently substitute a hedonic estimate when comp data is thin.