Strategy
Property Investment Strategy AustraliaBuild Around What Survived Backtesting
Most property investment strategies in Australia share the same flaw: they’re built on metrics that sound rigorous but failed when actually tested against outcomes. Infrastructure pipelines. Population growth. Building approvals. Five-year momentum. All present in buyer’s agent slide decks. All discarded when we ran the numbers.
After backtesting five versions of a suburb-scoring formula across 12,360 postcode-months, two signals survived. This post is about how to build a decision framework around those two signals — and why everything else is noise at best, misleading at worst.
Why Most Strategies Are Built on the Wrong Foundation
The conventional property investment strategy in Australia follows a predictable template: find suburbs with strong population growth, infrastructure projects in the pipeline, low vacancy, rental yield above a threshold, and days on market tightening. Add a narrative about a regional hub “catching up” to the capital or a city corridor undergoing renewal. Buy. Wait.
The problem is not that these metrics are meaningless. Some of them are real. The problem is that they’ve never been stress-tested at the level of granularity where you actually make decisions: individual suburbs, over investment-relevant time horizons, with publicly available data.
We tested that. The v1 version of our formula incorporated seven factors including infrastructure spend, population pressure, and building approvals. The result was 55% accuracy — a coin flip with a longer checklist. Infrastructure spending affects corridors over decades. Population growth is too slow-moving to time an entry. Building approval data is state-level at best, with almost no usable suburb-level signal from free sources.
Coin-flip accuracy. Discarded entirely before v2.
The lesson from the failed formula wasn’t “we need more data.” It was a structural one: prediction doesn’t work at suburb level with publicly available data.The variables that drive suburb booms operate at timescales and geographies that don’t fit investment-horizon suburb analysis.
What this means for your strategy
If your current approach relies on infrastructure pipelines, population growth projections, or building approvals as primary signals, the backtest says those signals have near-zero predictive power at the suburb level over investment-relevant time horizons.
We scored 393 suburbs using only the signals that survived
Strong / Good / Fair / Weak signal labels updated fortnightly. Join the wishlist to get access.
The Two Signals That Actually Work
After abandoning prediction and pivoting to detection — asking “is this suburb booming now?” instead of “will it boom?” — accuracy jumped dramatically. The current formula (v2.3) achieves 85.7% accuracy with 0% false positives across 78 suburbs (28 that boomed, 50 controls), with a 20.2-point separation gap between real booms and false signals.
But detection alone doesn’t tell you whichsuburb within the detected universe will outperform. We needed a ranking signal — something that, after cancelling the market tide, discriminated between suburbs.
We tested everything: annual growth rate, acceleration ratio, five-year momentum, days on market, vacancy, rental yield, repeat boomer history. We computed excess returns — each suburb’s growth minus the median across all peers in the same period — and checked which features predicted outperformance.
One signal survived. Combined with detection timing, it becomes a two-axis system for suburb selection.
Affordability headroom — the ranking signal
Boom timing — the entry signal
That’s the complete framework. Not seven factors. Not a sophisticated model. Two signals, both grounded in backtested data, both publicly checkable. The strategy question becomes: is this suburb affordable relative to its city, and is it early in a detected boom?
What Didn’t Survive
The hardest part of building a data-driven strategy is discarding metrics you’ve been told to watch. Several well-regarded signals failed the backtest outright.
Five-year momentum — failed
Mean reversion dominates. Past outperformers tend to underperform going forward. Buying a suburb because it “has been performing well” is backing the opposite of what the data supports.
Growth phase / acceleration ratio — failed
Growth phase does not predict relative outperformance. The market tide lifts all boats. A suburb accelerating faster than its peers today does not systematically outperform those peers going forward. The forecaster version of our formula had a pooled Spearman correlation of 0.42 — which looked strong until we computed the within-date figure: −0.08, worse than random.
Population growth — failed
Too slow-moving and too geographically coarse to time a suburb-level entry. Population flows affect regions over years; boom timing operates over months.
The statistical illusion is worth dwelling on. Our forecaster looked like a genuine breakthrough — 0.42 pooled Spearman correlation, 28,049 scored rows in the walk-forward backtest. We nearly shipped it. Then we split the correlation within each scoring period and got −0.08. The model was ranking time periods, not suburbs. Boom years had both higher predictions and higher realised returns, generating the pooled signal from era effects alone.
If you’re evaluating any property research tool, model, or algorithm: ask whether accuracy is measured pooled or within-period. Pooled statistics are easy to inflate with era effects. Within-period discrimination is the real test.
The pattern to watch for
Any strategy or tool that uses momentum, growth phase, or acceleration as a primary ranking signal is doing something the backtest doesn’t support. These signals detect booms after the fact — they don’t discriminate which suburb outperforms within a boom cohort.
Two signals. 393 suburbs. Fortnightly labels.
BoomAU applies the backtested framework across every suburb under $800K. Strong / Good / Fair / Weak signal, filtered to your budget.
How to Build Your Decision Framework
A property investment strategy built on these two signals has three layers: filter, rank, time.
Layer 1 — Filter
Set hard boundaries that eliminate suburbs where the signal doesn’t apply. The v2.3 formula uses four hard filters: annual growth at or above 5%, days on market at or below 45, vacancy at or below 2%, and a median price cap of $800K. Suburbs that don’t pass all four are excluded from scoring entirely. This is not a ranking exercise — it’s a gate. Currently 393 suburbs pass these filters nationally.
Layer 2 — Rank by affordability
Within the filtered universe, rank by affordability headroom. A suburb priced well below the capital city median has more room to run than one already trading above it. The backtest says this is the only cross-suburb ranking signal that survived tide cancellation. Budget bands matter here — 35 suburbs nationally score under $400K, 149 under $600K, 204 under $800K.
Layer 3 — Time the entry
Detection catches booms 6–12 months after they start, still capturing 60–85% of total gains. The timing signal is how much of the affordability gap has already been consumed. Early detection — less than 30% of the gap closed — means the majority of the headroom remains available to the next buyer. This is where the tier system becomes actionable.
The tiers from the walk-forward backtest across 12,360 postcode-months show what this framework produces in practice:
| Tier | Excess return | Beat market | n |
|---|---|---|---|
| Strong | +7.5pp | 71% | 2,103 |
| Good | +1.3pp | 55% | 3,349 |
| Fair | −0.7pp | 47% | 5,788 |
| Weak | −6.4pp | 28% | 1,120 |
Walk-forward backtest, 12,360 postcode-months, no lookahead. Excess return = suburb 12-month growth minus market median growth. Full methodology →
The discrimination is perfectly monotonic. Strong Signal outperforms Good Signal outperforms Fair Signal outperforms Weak Signal. No tier crossovers. A 13.9 percentage-point spread between the top and bottom tiers, all from the same asset class, the same country, the same macro environment. The spread comes entirely from suburb selection.
That’s the case for a data-driven strategy: not that you can predict which suburb will boom, but that you can systematically put yourself in the upper tier and avoid the lower one. The difference between 71% beat-market and 28% beat-market is the difference between a strong investment portfolio and a poor one — and it comes down to which signals you built your strategy around.
One More Thing: Era Dependency
One finding from the backtest deserves its own section because it contradicts a lot of conventional strategy advice.
Boom size is era-dependent. The pre-2015 median boom in the dataset was 1.3%. Post-2020, the median boom was 16.2%. Same framework, same signals, dramatically different magnitudes depending on the macro environment.
What this means for strategy: the framework tells you where to berelative to the market, not how big the market move will be. Strong Signal suburbs in a low-growth era will outperform their peers by roughly 7.5pp — but if the market only moves 2%, that means 9.5% vs 2%. In a high-growth era, the same signal might mean 23% vs 16%. The relative position is consistent. The absolute return is not.
Any strategy that promises specific absolute returns is making claims the backtest doesn’t support. Era effects are real, they dominate absolute returns, and they’re not predictable from suburb-level data. What you can control is your relative positioning within the available opportunity set.
The honest version of property investment strategy
Filter to suburbs with genuine affordability headroom and boom detection signals. Enter early in the cycle when headroom consumption is below 30%. Don’t chase past momentum. Don’t build strategy around population growth or infrastructure pipelines. The data says those are category errors.
What You Can Check Yourself
You don’t need a tool to apply this framework. The data is free. The calculation is straightforward.
Check affordability headroom
Domain publishes capital city median house prices quarterly. Look up the suburb’s current median on YIP or Domain. Divide suburb median by city median. Below 1.0 means headroom. Above 1.5× means the backtest says it’s less likely to outperform. The effect is monotonic — the further below city median, the stronger the signal.
Check the boom detection filters
Four signals together form a boom signature. Annual growth at or above 5% (YIP publishes this per suburb). Days on market at or below 45 (YIP). Vacancy rate at or below 2% (SQM Research — free postcode charts, 16 years of history). Median price at or below $800K. If all four pass, you’re looking at a suburb that qualifies for scoring.
Assess entry timing
Compare the suburb’s current median to its median from 12–24 months ago. A suburb that has moved from 60% of the city median to 75% has consumed 15 percentage points of headroom. If the city median is $1M and the suburb was at $600K, moved to $750K, and still has room to run toward the median — that’s early detection. Closer to the city median means later in the cycle and less remaining upside.
The hard part is doing this at scale. Checking one suburb takes 10–15 minutes across four data sources. Checking the full investable universe — 8,417 Australian suburbs — requires a scraping and scoring pipeline that runs fortnightly. That’s what we built. Currently 393 suburbs pass the hard filters nationally, broken into budget bands: 35 under $400K, 149 under $600K, 204 under $800K.
The full backtest methodology — formula versions, 78-suburb validation, walk-forward tier discrimination, and the forecaster we abandoned — is published on our proof page. No gating, no email required.
Join the Wishlist
We'll email you when BoomAU launches — starting with the budget range you care about.
Be first in line
- ✓Fortnightly Strong / Good / Fair / Weak signal labels per suburb
- ✓Filtered to your budget band
- ✓Built on a backtest of 12,360 postcode-months