In this episode, Damien interviews Jeremy about how he approaches property market analysis, inspired by an old interview Jeremy did with Ryan McLean on the OnProperty podcast around eight years earlier, which Damien credits as his own first introduction to Jeremy's work. Damien revisits several of the same questions from that original interview to see how Jeremy's answers have evolved.
How Jeremy Got Into Property Analysis
Jeremy explains that his starting motivation was simply wanting to identify some of the best areas in Australia to invest in. The key realisation was that capital growth is driven by supply and demand, which led him to look for data capable of representing supply and demand, then combine that data into a single overall score for each suburb. In the early years, he was largely interpreting general property investing experience rather than working from formal data science methods, using proxies like how many people turned up to an open inspection (demand), how few properties were listed for sale (supply), or a high auction clearance rate, as informal signals of underlying market conditions.
Jeremy notes that even from the earliest version of the DSR, certain metrics carried more weight than others, since some correlate to capital growth more strongly than others. The tool started as something he built in his own time, alongside a full-time job, initially just to compare individual suburbs he was personally curious about, before realising the value of comparing suburbs Australia-wide on a consistent basis.
What Qualifies Jeremy to Give This Advice?
Asked directly what qualifies him to advise people building potentially million-dollar portfolios, Jeremy is upfront that he has no formal background in data science, describing his own coding ability (in Python) as enough "to get myself in real trouble." His actual background is in electrical engineering, and he attributes his genuine expertise to the practical, on-the-ground experience gained from personally purchasing 16 investment properties, rather than formal data training. He describes the early DSR as "human intelligence" rather than artificial intelligence.
Jeremy recounts his well-known early sanity check from around January 2010: after ranking Australian suburbs by his new scoring system, Airlie Beach units scored at the very bottom, and when he called local agents there, they responded eagerly and enthusiastically, keen to have him visit. By contrast, Heathcote houses (in Sydney's Sutherland Shire) scored at the very top at the time, and the one agent who returned his call simply offered to add him to a waiting list for when something became available, reflecting a market where demand clearly exceeded supply.
Is Rent Growth a Precursor to Capital Growth?
Jeremy explains this can work in either direction: sometimes rental yield leads capital growth (since securing a lease is generally faster and easier than securing a mortgage, making renters more agile than buyers in responding to an emerging market), and sometimes capital growth pulls rents up afterward. He cautions that relying purely on current yield still carries its own form of uncertainty, since future rent isn't guaranteed either, a property could face a vacancy period or downward rent pressure if the local rental market becomes oversupplied, so it's just as important to check surrounding vacancy rates and stock-on-market trends as it is to research capital growth potential.
Is the DSR a Perfect Science?
Jeremy is direct that the DSR is not 100% accurate, and that there have been embarrassing failures along the way, but argues it remains a considerable improvement over pure guesswork or "gut feel," which he associates with an older style of property advice. He describes the current period as still early in the broader application of data and AI to property investing, an ongoing area of improvement rather than a finished, perfected system, with further data expansion planned over the following 12–24 months.
On timeframes, Jeremy notes forecasting reliability drops off meaningfully beyond about four or five years, and that historically, markets tend to rebalance within around three years of a supply-demand imbalance emerging, after which growth tends to return to a more typical, ordinary rate for that area. He cites Hobart and Adelaide as examples where the data flagged strong conditions that subsequently played out, and Brisbane as an example where a high DSR score persisted for some time before growth genuinely took off, partly due to lingering investor caution following Perth's earlier downturn tied to the resources sector.
Can Future Supply Be Predicted?
Jeremy notes that while development application data exists at the council level, it doesn't provide certainty, since a new application could be lodged the day after a purchase is made, regardless of how thorough prior research was. His preferred, more reliable approach is simply buying an established house in an already built-up area, since genuinely limited remaining vacant land constrains how much future oversupply is even possible, in contrast to units or vacant land, which remain more exposed to this risk.
Reviewing an Underperforming Property
Jeremy and Damien discuss what to do when an existing property has underperformed for years. Jeremy explains that with reasonable data, it's possible to assess whether a market is likely about to turn a corner or remains at genuine risk of further oversupply, rather than defaulting to a blanket "always hold" mentality, which he considers outdated advice from an earlier, less data-informed era. He notes that selling and re-entering a market is harder and slower than trading shares, and that being willing to admit a past purchase was a mistake is a difficult but sometimes necessary step. Jeremy shares that he personally sold out of New Zealand property holdings (despite those specific properties having performed well) largely because he found reliable New Zealand market data much harder to access than Australian data, and judged the resulting uncertainty too risky to hold through long-term.
Population Growth: Indicator of Demand or Supply?
Jeremy pushes back on the popular idea that rising population growth signals strong demand and therefore future capital growth. He explains that population growth actually tends to follow the construction of new housing supply, since people generally can't move into an area faster than dwellings become available for them to occupy, meaning population growth is often better understood as an indicator of new supply rather than genuine demand at the suburb level. He shares a related experience of contacting a population forecasting organisation, who confirmed their figures are based on planned development activity discussed directly with councils, rather than a survey of aspirational buyer demand.
Technical vs. Fundamental Analysis
Jeremy agrees with a framing of DSR-style analysis as broadly analogous to "technical analysis" (numbers and data-based), compared to what used to be called "fundamental analysis," historically involving more subjective, on-the-ground assessment (driving around a suburb, observing garden upkeep, car types, and so on). He notes the industry has increasingly moved trust toward data-driven analysis, and expresses scepticism toward anyone describing their approach as "part science, part art," suggesting that framing often signals a lack of genuine underlying data rigor.
Can Data Be Trusted?
Jeremy is candid that no data set should be trusted implicitly, since every data set he's worked with has contained some anomalies or quirks. His view is that trust should be built empirically, by observing whether a data source's historical predictions have actually played out over time. He notes that in the DSR's early days (from 2010), there was little trust in data-driven approaches generally, and the tool was explicitly positioned only as a shortlisting aid; today, he believes data (and increasingly AI) plays a much larger role in decision-making, to the point where he cautions that manually overriding an algorithm's output with personal judgement risks undoing its value rather than improving it.
The Property Clock
Asked about the commonly referenced "property clock" concept (where different positions on a clock face represent different phases of a market cycle, such as boom, peak, or decline), Jeremy expresses scepticism, noting he doesn't know its original source, that different commentators define its phases inconsistently, and that mapping actual demand-and-supply conditions onto to a fixed clock position is inherently difficult, since a market can show the same score on the way up as on the way down. Overall, he says he isn't a strong believer in the concept as a genuinely useful analytical tool.
Monitoring Existing Property Performance
Jeremy explains there's a general tendency for most suburbs to eventually grow at broadly similar long-term rates, but the timing of when that growth occurs can vary significantly, meaning there can be genuine merit in holding an underperforming property if the data suggests conditions are turning, rather than assuming permanent underperformance. He references a specific metric, market cycle timing, which looks at historical growth patterns over periods from six months up to roughly a decade, checking whether a suburb's recent growth trend resembles the pattern typically seen just before a boom. This connects to the broader concept of mean reversion: an area that has underperformed for a long period may now be comparatively affordable and due for a correction upward, while an area that has already boomed significantly may be due for a softer growth period ahead, essentially the opposite of assuming recent strong performance will simply continue.
Why No Single Metric Can Be Relied On Alone
Jeremy stresses that no individual metric should be relied upon in isolation, since there will always be exceptions where a metric looks favourable but growth doesn't follow, or vice versa. He notes that even using the original, simpler eight-variable version of the DSR, he never found a single market where all eight variables sat above benchmark simultaneously, reinforcing that a combined, weighted score is far more reliable than any single data point.
Statistical Reliability
Jeremy explains that data volume is a major factor in how much confidence can be placed in a given suburb's figures, for example, an auction clearance rate based on just one or two auctions in a month is far less reliable than one based on a larger sample. Contradictions between independent data sources reduce confidence, while agreement between sources increases it. He also notes that modern data science techniques (such as engineered or "synthetic" variables) can combine many base metrics into new, sometimes less intuitively explainable combinations that still correlate meaningfully with capital growth, and that determining the ideal number of variables to include (avoiding both under- and over-fitting) is itself a data science problem, one Jeremy defers to specialists on rather than claiming expertise in himself.
Income Growth Is Not a Reliable Indicator
Jeremy states plainly that, based on the analysis he's seen, income levels and income growth don't show a strong enough correlation with capital growth to be a useful metric, citing decades of higher income in Sydney's eastern suburbs without those suburbs consistently outperforming areas like Sydney's western suburbs, as an example.
Data Lag and Currency
Jeremy explains that different data sources carry very different lag times, Census data (collected only every five years, sometimes with a further year or more delay before publication) can be several years out of date in a worst-case scenario, while other metrics can be updated within a week of being collected. He notes that broad, national-level metrics could theoretically be updated close to daily, given the volume of transactions occurring across the country each week, but that highly localised analysis (down to a specific street or suburb) will always involve a trade-off between data recency and statistical reliability, since transaction volume at that granular level is inherently limited.
Volatility as a Reliability Signal
Jeremy acknowledges that DSR scores can and do drop noticeably from one month to the next, occasionally shortly after someone has made a purchase decision based on a prior month's score. His approach to managing this is to always check a market's historical volatility: a suburb whose score bounces around significantly from month to month is treated as inherently less reliable and generally best avoided, whereas a consistent, stable trend (even with occasional single-month dips) can still be reasonably relied upon. Damien adds that checking whether neighbouring suburbs show a similarly elevated score provides useful confirmation that a signal reflects a genuine area-wide trend rather than an isolated, potentially unreliable data point.
Closing Thoughts
Damien and Jeremy close by reiterating that this kind of data analysis represents just one part of a broader investment decision-making process, and that continuing to expand and improve the underlying data set remains an ongoing priority.