The claim everyone repeats
“Pick the right suburb and you've done 80% of the work.” You've heard this one. Podcasts repeat it. Buyer's agents repeat it. Your uncle at Christmas repeats it. Nobody seems to have checked if it's true.
So we tested it. Not with opinions. With 10,874 property resales.
How we tested it
We looked at properties bought and sold without renovation, held for 3 years or less. Pure single-cycle market resales. We calculated their annual capital growth and grouped them into cohorts. Same area. Same purchase year. Same sale year. The same market forces hit every property in a cohort.
The question was simple. Did each property beat or miss the overall median return? And did properties in the same cohort move together? If they did, that geographic level is a real market. If they didn't, it's noise.
We tested every level. National down to individual streets.
The results
Here's what the data shows. The smaller the market, the more reliable it is. National, state, and city-level markets are terrible at predicting individual property outcomes. The signal lives at suburb and street level.
At the national level, only 64% of properties matched their cohort's direction. With large cohorts, you'd expect about 53% by pure chance. So the ‘national market’ removes just 24% of the uncertainty. When a podcast host says “now's the time to buy,” that's barely better than a coin flip.
What does that look like in practice? Take the 2021-to-2022 national cohort. The overall dataset median was 6.5% per year. Some properties in that cohort returned +25%/yr. Others returned -15%/yr. Same country, same buy year, same sell year, 40 percentage points apart. Properties deviate from their national cohort median by an average of 10.34 percentage points per year. Over a three-year hold, that gap compounds to 30 points or more.
one “market”
Four properties in the same national cohort. Same buy year, same sell year, same “market”. The national narrative told you nothing about which outcome you would get.
State level? 72%. Better. But still only 36% of uncertainty removed. You can't just ‘buy in NSW’ and expect it to determine your outcome. SA4 city regions manage 38%. These large geographic buckets are too blunt. They tell you almost nothing about what your specific property will do.
Suburbs hit 90%. Sounds good. But here's the catch. At the suburb level, cohorts are small. 2 to 3 properties. And small groups agree by chance about 75% of the time. The real signal? Suburb selection removes 61% of the uncertainty. Not 80%.
| Market Level | Move Together | Uncertainty Removed | Verdict |
|---|---|---|---|
| National | 64.3% | 24% | Mostly noise |
| State | 71.7% | 36% | Weak signal |
| SA4 (Region) | 77.7% | 38% | Moderate |
| Suburb | 90.3% | 61% | Strong |
| Street | 97.3% | 89% | Very strong |
Smaller markets are more reliable. Every time.
National, state, and city-level (SA4) markets remove less than 40% of uncertainty. Suburb markets are strong. Street-level markets are the strongest of all.
Streets: the real edge
What if the real answer was one level below where everyone stops looking?
97.3% of properties on the same street, bought and sold in the same years, moved in the same direction. That removes 89% of the uncertainty. The highest of any level we tested.
Two properties bought and sold on the same street in the same years have a 97.3% chance of producing the same outcome. That's not a market trend. That's close to a guarantee.
The ‘80% rule’ is wrong. And it stops one level too soon.
Suburb selection does 61% of the heavy lifting. Street selection does 89%. The classic advice points in the right direction. But it overstates the case and misses where the real predictability lives. Worse, it keeps investors focused on national and state narratives that are basically noise.
Two streets. Two stories. Same pattern.
Abstract statistics only go so far. Here are two real street pairs from the dataset. Same street, same buy year, same sell year. Different buyers, different agents, different prices. Same outcome.
Both properties were purchased in October 2019 at the start of the Queensland growth cycle. Neither was renovated. One sold in August 2021, the other in November 2021. 95 days apart. Both returned well above the overall median.
Both outperformed. Different prices, different sell dates, same direction. The street ran hard. 24.9%/yr and 18.8%/yr against a dataset median of 6.5%/yr.
Both properties were bought during the ACT price peak in late 2021. Both sold in the first half of 2022 as the market corrected. Buy dates were 59 days apart. Sell dates were 116 days apart. Independent transactions by different buyers. But the street told the same story for both.
Both underperformed. Independent buyers, independent transactions, same outcome. The street fell sharply against a rising national median.
These are not cherry-picked outliers. The typical street pair differs by about 1.7 percentage points per year. The average property deviates from its street cohort median by just 0.87 pp/yr. These two examples show wider spreads (4 to 6 pp/yr) because the direction effect is easier to see. But across 855 street-level cohorts, 97.3% of properties matched their street's direction.
What this means for you
Go deeper than the suburb
Microburbs gives you street-level data. Median prices, rents, demographics, and turnover rates. So you can invest where the data actually matters.