CGT valuation in 2027: can a model-based figure stand up to the ATO?

A CGT valuation sounds like a back-office chore until you realise how many properties might need one. Changes flagged in the May 2026 Budget could require up to 2.4 million investment properties to be valued as at 1 July 2027, according to the data firm behind a new accuracy study. There is an ATO formula that offers an alternative, but the volume alone raises a plain question. Who is going to value all of them, and will the answer hold up if the tax office asks questions?

What the Cotality benchmarking actually found

The study looked at roughly 1.17 million lending records over a rolling two-year window. It tested three of the firm’s own tools against full inspection valuations: an automated valuation model, a desktop product, and a hybrid that blends data checks with valuer judgement. Results were adjusted for market movements.

Within a 10% tolerance of the benchmark, the tools landed at 77%, 76% and 80%. For context, separate full inspections by different firms agreed with each other 78% of the time. That is the realistic ceiling, because even two qualified valuers looking at the same home do not always land on the same number.

The hybrid led at every tolerance band. But only three to five points separated best from worst, and the hybrid’s sample of about 48,800 was the smallest. The authors themselves warned against calling it superior. Treat that as a fair reading: the gap is small, and the test was run by a firm that sells these tools.

Accuracy is not the same as acceptance

Here is the catch. The study measures how close a digital figure sits to a lender-style valuation. A lending valuation answers one question: is there enough security for this loan? A tax valuation answers another: what was this asset worth on a specific date, in a document you may have to defend later?

The source material does not say how the ATO would treat a model-based figure, and it would be a mistake to assume it would accept one. What we can say is that the two uses carry different consequences:

  • A lending valuation is a lender’s risk call. If it is off, the lender and its insurer wear most of the problem.
  • A tax valuation feeds into a capital gain. If it is off and the ATO disputes it, the property owner is the one answering.

That shifts the question from “is the model accurate on average?” to “is this particular figure defensible for this particular property?” Averages comfort a lender holding thousands of loans. They offer less comfort to one investor holding one asset.

Where the models are weaker

Performance was not even across the market. Homes priced between $500,000 and $1 million made up 57% of the data, and nearly all methods reached 80% accuracy there. Higher up, the picture changes. For properties between $1.5 million and $2 million, automated models managed 64%, against 74% for the hybrid.

That matters for investors, because larger and more distinctive properties are where a gain, and any dispute over it, is likely to be bigger. A model that works well on a standard three-bedroom house in a well-traded street has less to go on when the property is unusual or the sales evidence is thin.

Capacity is its own pressure point

The headline pressure from 2027 is not only accuracy but volume. If millions of properties need a figure as at one date, the pool of human valuers could be stretched. That is the case for digital tools, and the reason the study’s timing is no accident.

But capacity and acceptance are separate problems. A faster valuation is only useful if it is one the owner can rely on. If the tax rules or the ATO’s guidance later narrow what counts as an acceptable method, a cheap figure obtained in a rush could need redoing.

The ATO formula mentioned as an alternative also changes the equation. Owners who can use it may not need a standalone valuation at all. Whether that suits an individual depends on the details of their property and their records, which is a question for a tax professional rather than a model.

Questions worth asking before 2027

Nothing here is settled, and the final design of the capital gains changes will decide how much of this matters. A few things are worth checking as the detail emerges:

  • What valuation methods the ATO says it will accept, in writing, rather than what vendors say will be accepted.
  • Whether the formula alternative is available for your property and what it assumes.
  • Who carries the risk if a valuation is challenged: you, the valuer, or the provider.
  • What records you hold on purchase price, improvements and costs, since they feed any calculation.

For wider background, our negative gearing, CGT and investor tax guide sets out how capital gains fit into the investor picture. Lending conditions are also moving at the same time. Over 103,000 loans were refinanced in the June 2026 quarter, and bank branch closures are handing brokers a bigger share of new home loans, so valuation demand is rising on more than one front.

Key numbers

  • 1.17 million lending records tested
  • 77%, 76% and 80%: share of digital valuations within 10% of the benchmark
  • 78%: how often separate full inspections agreed with each other
  • 64% vs 74%: automated model vs hybrid on $1.5m to $2m homes
  • Up to 2.4 million properties may need a figure as at 1 July 2027

If you want a weekly read on policy changes like this, subscribe to the Australian Property Review newsletter. Your next step is simple: check what the ATO says about acceptable valuation methods before you pay for a figure.

General info, not financial advice.

Share this article

LEAVE A REPLY

Please enter your comment!
Please enter your name here