Does the Proportion Sold at a Loss Predict Future Growth?

Key findings

• Nationally, about one in six repeat sales (15.4%) resulted in a nominal loss over the full period from 1990 to 2026. That share peaked at 22% in 2012 and fell to under 6% in early 2026.
• The raw relationship between a suburb’s loss rate and its future growth is near zero. Knowing that 20% of sales in a suburb lost money tells you almost nothing about whether that suburb will grow faster or slower than the national average over the next three years.
• The exception is at the extremes. Suburbs where more than two in five resales lost money grew roughly 1 to 3 percentage points per year slower than the national median in the years that followed (2022 to 2026).
• The strongest signal is the trend. Suburbs whose loss rate fell from elevated (above 15%) to low (below 15%) returned about 2.6 percentage points per year more than suburbs whose loss rate moved in the other direction (2022 to 2026).
• Units lose money roughly 50% more often than houses (one in five vs one in seven, 1990 to 2026). Short holds (under two years) are the riskiest, with more than one in four resulting in a loss.

Methodology

We paired consecutive sales of the same property using address-matched records across multiple transaction sources. Each pair produced a buy price, sell price, and hold period. We computed the annualised nominal return for every pair and classified any pair with a return below zero as “sold at a loss”.

After applying quality filters to remove likely land-to-house conversions, extreme price outliers, and suspect renovations, the final dataset contained 12.5 million paired sales across 6.0 million unique properties in 13,807 suburbs.

To test the predictive value of loss rate, we split the data by time. For each suburb, we calculated the loss rate using all sales before a cutoff year. We then measured the median annualised return for sales in that suburb after the cutoff. We repeated this across five different cutoff years (2019 through 2023) to test robustness.

The trend analysis classified suburbs into four groups based on how their loss rate changed between two periods (pre-2019 vs 2019-2021). We then measured each group’s post-2022 performance against the national median.

Results

Static loss rate does not predict future growth

We divided 5,400 suburbs (each with at least 50 pre-2022 sales and 20 post-2022 sales) into bins by their pre-2022 loss rate, then measured their post-2022 median annualised return relative to the national median.

Suburbs with moderate loss rates (15 to 30%) actually outperformed low-loss suburbs in the years following 2022. This is most likely mean reversion: suburbs that had been beaten up during the 2018-2020 downturn bounced back harder during the recovery.

To illustrate: Noosa Heads (QLD, Sunshine Coast) had a 22% loss rate before 2022, but it collapsed to just 5.5% as demand surged during the sea-change boom. Post-2022, Noosa Heads returned about 5 percentage points per year above the national median (2022 to 2026). Contrast this with Docklands (Melbourne), where the 45% pre-2022 loss rate worsened to 50%. Post-2022, Docklands returned about 6 percentage points per year below the national median.

The trend matters more than the level

We grouped suburbs by whether their loss rate was improving or worsening between two measurement windows (pre-2019 and 2019 to 2021), then checked how each group performed after 2022.

The spread between the best-performing group (improving suburbs) and the worst-performing group (worsening suburbs) was about 2.6 percentage points per year over 2022 to 2026. That is a material difference for a buy-and-hold investor.

Property type and hold period

Units lost money roughly 50% more often than houses over the full period from 1990 to 2026 (one in five vs one in seven). The median annual return for units was about 2.4 percentage points lower than for houses over the same period. Suburbs with high loss rates tend to have more unit stock.

Short holds amplify risk. Owners who sold within two years of buying lost money 28% of the time (1990 to 2026). For holds of 10 to 20 years, the loss rate dropped to 6%. For holds over 20 years, it was 1.5%. Time is the most reliable hedge against selling at a loss.

Defence against criticism

“This is just mean reversion”

Partly true. The moderate-loss suburbs (15 to 30%) outperforming low-loss suburbs in the 2022-2026 recovery is consistent with mean reversion. We do not claim the static loss rate predicts growth. Our claim is narrower: the trend in loss rate is predictive, and extreme loss rates (above 40%) are a reliable warning signal. Mean reversion explains the middle of the distribution but not the tails or the trend.

“Nominal returns ignore inflation”

All returns in this study are nominal. A property that “grew” 2% per year in a 3% inflation environment lost purchasing power. The true proportion of real losses is higher than the 15.4% we report. However, because we compare suburbs against a national median (also nominal), inflation cancels out in relative comparisons. The trend signal and extreme-loss findings hold regardless of the inflation adjustment.

“You are measuring the past, not predicting the future”

We used a strict out-of-sample test. Loss rates were calculated using only sales before a cutoff year. Future growth was measured using only sales after that cutoff. We repeated this across five different cutoff years (2019 through 2023) and found consistent results. The trend signal held across all test periods.

Limitations

• The study covers only properties that sold at least twice. Properties bought and never resold are invisible. If owners avoid selling at a loss (holding through downturns), the true loss rate may be higher.
• Transaction costs (stamp duty, agent fees, holding costs) are not included. A property that sold for the same price after five years actually lost money in real terms after costs.
• The post-2022 recovery period is unusually strong. The trend signal may weaken during a sustained downturn. Testing through a full market cycle would strengthen the finding.
• The extreme loss rate signal (above 40%) applies to only 28 suburbs nationally. The sample is small, though the effect is large and consistent.

Conclusion

The proportion of properties sold at a loss is a widely cited risk metric, and it already appears in Microburbs suburb and property reports. But the raw number alone tells investors almost nothing about future growth. What matters is whether that number is getting better or worse.

Suburbs where the loss rate is falling, where fewer owners are losing money than before, outperformed those where it is rising by about 2.6 percentage points per year over 2022 to 2026. That is a stronger signal than the static loss rate, the absolute price level, or the median hold period.

For investors, the practical takeaway is simple. Do not avoid a suburb just because it has a high historical loss rate. Check whether the loss rate is improving. If it is falling, the suburb may be recovering. If it is still rising, the structural problem has not resolved.