Can GPT Pick Growth Suburbs?
We asked 7 versions of GPT to name Australia's top growth suburbs. Then we measured what actually happened over 38 months. Here is what we found.
You would have been better off picking suburbs at random.
Six of seven GPT versions picked suburbs that grew slower than the market average. On a typical $875,000 property, GPT would have cost you around $4,600 per year in missed growth compared to picking a suburb at random.
The Prompt We Used
We sent the same prompt to every version, for every state and territory in Australia:
What GPT-3.5 Said (NSW, up to 2024)
1. Parramatta - 20% growth: Parramatta is undergoing significant urban renewal and infrastructure development.
2. Penrith - 18% growth: Penrith is benefiting from its proximity to Sydney and ongoing investment in transport infrastructure.
3. Liverpool - 15% growth: Liverpool is experiencing population growth and improved amenities.
4. Castle Hill - 16% growth: Castle Hill is a popular suburb with strong demand.
And so on for 100 suburbs per state.
Notice the generic reasoning. "Infrastructure development", "proximity to Sydney", "strong demand". These phrases appear thousands of times in real estate articles. GPT is reciting training text patterns, not analysing live market signals.
How Each Model Performed
Each bar shows total outperformance or underperformance versus the market. Red means GPT's picks grew slower than a random selection.
On average, GPT cost investors around $4,600 per year on a typical $875,000 property
Performance by State
GPT's biggest miss was Western Australia, where it cost investors $84,028 on the median property. Only Victoria saw GPT picks slightly outperform.
| State | Picks | Median Price | vs Market | Dollar Cost | Beat % |
|---|---|---|---|---|---|
| WA | 1,036 | $969k | -8.67pp | -$84,028 | 24% |
| NSW | 1,111 | $1,051k | -1.57pp | -$16,487 | 46% |
| Qld | 877 | $951k | -1.56pp | -$14,836 | 47% |
| SA | 1,024 | $788k | -0.36pp | -$2,875 | 47% |
| Vic. | 1,084 | $803k | +0.70pp | +$5,652 | 54% |
WA was the worst by far. GPT gravitated to Perth's most expensive, most-written-about suburbs. The real growth was in affordable outer suburbs that GPT ignored. Only 24% of GPT's WA picks beat the state median.
Real Properties. Real Losses.
These are not hypotheticals. GPT recommended these suburbs. Anyone who bought where GPT told them to lost money, before transaction costs.
GPT predicted +6% growth. Actual result: -13.4%.
GPT's reasoning: "Increased investment in infrastructure and amenities."
3 year growth to March 2026
GPT predicted +8% growth. Actual result: -15.0%.
GPT's reasoning: "Prestigious suburb with strong demand for quality housing."
3 year growth to March 2026
GPT predicted +10% growth. Actual result: -18.1%.
GPT's reasoning: "Popular snow destination with limited housing supply."
3 year growth to March 2026
GPT predicted +10% growth. Actual result: -16.2%.
GPT's reasoning: "Proximity to Melbourne CBD and affordability."
3 year growth to March 2026
Launceston was GPT-3.5's second strongest recommendation for Tasmania. Not a throwaway pick buried at the bottom of a list. It was number two. GPT predicted 6% growth because of "increased investment in infrastructure and amenities." Three years later, the median house price has fallen 13.4% to $817,000. Anyone who bought at the time and followed GPT's advice has lost $126,000. That is before stamp duty, conveyancing, and agent fees.
The Bottom Line
You would have been better off picking suburbs at random.
Across 6,888 scored predictions from 7 GPT versions, six underperformed the market. The only version that beat the market (GPT-4.1) gained $3,599 on an $861,000 property over 9 months. That is not a strategy. That is noise.
At the other end, GPT-4.1-mini cost investors $26,729 on an $884,000 property. GPT-4-turbo cost $15,945 on a $950,000 property. GPT-3.5 cost $14,994 on an $882,000 property. In Western Australia alone, GPT cost investors $84,028 on the median property.
A random selection of suburbs in the same state would have outperformed GPT's recommendations in six of seven cases. Do not ask ChatGPT where to buy property.
Read the Full Research Whitepaper
All 7 GPT versions. All 8 states and territories. Animated growth charts. State-by-state breakdown. Data validation against CoreLogic and Domain.
Read the WhitepaperProperty Forecasting That Uses Real Data
Microburbs analyses 12.8 million data points across every suburb in Australia. Not text predictions. Actual supply, demand, and price signals updated weekly.