Why Property Hotspot Lists Are Wrong — And What Actually Works

The first thing to understand about property hotspot predictions is their business model. Most hotspot lists exist to sell something — a subscription, a buyer’s agent service, a course, a development package. The list is the lead magnet. The accuracy of the list is irrelevant to the revenue model.

This doesn’t mean the people writing them are dishonest. Many genuinely believe in their methodology. But the incentive structure explains why almost nobody publishes their historical accuracy. If your 2023 hotspot list underperformed the market, you don’t publish a retraction — you publish a 2024 list.

Think about it this way: if a hotspot list were genuinely accurate, the publisher would lead with their track record. “Our 2022 picks averaged 14% growth versus 6% for the broader market.” You’d see that on every landing page. Instead, you see methodology descriptions — impressive-sounding inputs with no output verification.

We know the standard methodology well because we built it ourselves. When we started BoomAU, our first formula used the same inputs everyone uses: population growth, infrastructure spending, building approvals, rental yield, days on market, vacancy rate, and vendor discounts.

The logic is intuitive. More people moving in means more demand. New infrastructure means better amenity. Low vacancy means landlords have pricing power. It all makes sense in a slide deck.

Then we did something most hotspot publishers never do. We backtested it.

Fifty-five percent. Barely better than flipping a coin. With seven inputs, months of data sourcing, and a methodology that sounds bulletproof to anyone who hasn’t tested it.

Each hotspot input fails for a specific reason. Understanding why is more useful than just knowing that they fail.

It gets worse. Even when hotspot models appear to produce impressive statistics, the numbers can be deeply misleading.

We learned this the hard way. After our prediction formula failed at 55%, we built a forecaster — a model that tried to rank which suburbs would outperform others. We backtested it across 28,049 scored rows and got what looked like a breakthrough result.

A 0.42 rank correlation is enormous in quantitative research. We were ready to ship it. Then we split the statistic within each scoring period — asking not “does the model rank well across all time?” but “within any given month, does the model rank suburbs correctly?”

The pooled 0.42 was a statistical illusion. Boom years had both higher predictions and higher realised returns. The model was ranking time periods, not suburbs. Within any given month, it couldn’t tell you whether Suburb A would outperform Suburb B. It was useless for the decision that actually matters.

This is the trap. Aggregated statistics pool data across time and space. They can produce genuinely impressive numbers that collapse the moment you ask the question an investor actually needs answered: “Right now, which suburb should I choose?”

Any hotspot model that publishes pooled accuracy without within-period accuracy is likely hiding this problem — or, more charitably, has never checked.

Many hotspot lists lean heavily on momentum: suburbs that have been growing strongly are predicted to keep growing. “This suburb grew 12% last year, so it’s a hotspot.”

The problem is mean reversion. In our backtesting, past outperformers tended to underperform going forward. Five-year momentum — which sounds like exactly the kind of robust, long-term signal you’d want — failed as a ranking signal after we cancelled the market tide.

This makes intuitive sense once you think about it. A suburb that has already boomed has consumed its affordability headroom. The gap between its median price and the city median has narrowed. The easy gains are behind it. Momentum-based hotspot lists are often recommending suburbs at the end of their run, not the beginning.

The opposite signal — affordability — turned out to be far more predictive. But recommending cheap suburbs doesn’t sell courses nearly as well as recommending the suburbs everyone is already excited about.

After our prediction formula hit 55%, we made a fundamental shift. Instead of trying to predict booms before they start, we built a formula to detectbooms after they’ve begun.

At first this felt like giving up. Surely the whole value is in being early? But when we studied actual boom trajectories across Australian suburbs, we found that booms are multi-year events. Detecting one 6–12 months after it starts still captures 60–85% of the total gains — with dramatically higher confidence.

78-suburb backtest: 28 that boomed, 50 controls that didn’t. Walk-forward validation, no lookahead bias. Full methodology →

85.7% accuracy with zero false positives. The detection formula correctly identified real booms and — crucially — never flagged a stable market as booming. False positives are what cost investors money, and our detection approach produced none across 78 tested suburbs.

We tested every feature we could find: growth rate, acceleration ratio, repeat-boomer history, days on market, vacancy, yield, rental trend, momentum. We computed tide-cancelled excess returns — each suburb’s growth minus the median growth across all peers in the same period — and checked which features predicted outperformance.

Two signals survived.

That’s the honest answer. Out of dozens of metrics tested across 12,360 postcode-months, two survived. Everything else either failed outright or produced statistical illusions that collapsed under within-period analysis.

You don’t need our formula. The two signals that matter are public knowledge. Here’s how to check them yourself:

The hard part isn’t knowing the signals — it’s doing this across thousands of suburbs fortnightly, catching booms within weeks of starting, and filtering out noise from thin markets. That’s what we automate.

Full backtest methodology, the 78-suburb validation dataset, and the walk-forward tier discrimination results are published on our proof page. No gating, no email required. Check the maths yourself.