Inside Sarah Kaye & Co Research — the in-house analytics practice we’ve built over the last six months, and the reason every piece you’ll read from us this year carries a different kind of evidence behind it.
BY SARAH KAYE & CO RESEARCH · MAY 2026
Through 2025 we noticed something quietly unsettling about the property commentary written for the suburbs we work in.
The same suburb would appear in three reputable property reports in the same week with three different median prices. The same prestige sale would be cited at one figure in agency commentary and another in the registered Torrens record. The same forecaster who told the country in March to expect 9 per cent growth in 2025 told us in April to expect 2.5 per cent growth in 2026 — across an entire nation, as if every postcode behaved the same way. Each individual disagreement was small. The pattern was not.
We’ve used those sources for years. We know how they’re built. They are produced by people we respect — journalists at the Australian Financial Review, researchers at Cotality and Domain, analysts at major banks. The disagreements are not sinister. But they are systematic, and they all run in the same direction. Every published number we’d been quoting on behalf of our clients was filtered through someone with a commercial interest in how it read. Agencies want buyers to feel urgency. Portals want to be cited. Forecasters want their brand built. None of those motivations are bad. None of them serve the buyer we represent.
We represent buyers exclusively. So six months ago we made a structural change. We stopped using intermediated property data as the basis for any client analysis or any public commentary. We went to the most direct source available — the NSW Valuer General’s Property Sales Information bulk feed, which is the registered Torrens-record data, drawn from Land Registry Services, maintained by the state government, and free under the NSW Open Data Policy.
Every property settlement registered with NSW Land Registry Services — every legal change of ownership — is published in this dataset. It is the ground truth of who paid what for which property. Other commentators read this data after it has passed through their own filters. We read it directly.
A taste of what changes when you change the inputs
Two examples will land the point.
In our own market commentary on Palm Beach, published in February, we wrote that in early 2026 “reputable portals simultaneously reported a median of $6.0 million, $5.75 million and $4.46 million” for the suburb. Three respected sources. Same period. A 35 per cent gap between the highest and the lowest. The methodology behind each number was defensible in isolation. None of them was wrong. But none of them was useful, either, to a buyer trying to make a multi-million-dollar decision.
In the same window, agency commentary about Mosman cited an approximately $50 million residential sale on Kirkoswald Avenue as the suburb’s new record. The registered sales data — the legal Torrens record — shows that Mosman’s top settlement in our 24-month window is $43.5 million at 4 McLean Crescent, settled in August 2025. The $50 million claim may yet register; conveyancing on prestige sales sometimes runs long. But as of today it is not in the registry. The $43.5 million sale is. The two numbers are different in ways that matter.
These aren’t isolated cases. They show a structural pattern that runs through almost every sentence written about our patch.
What we built
Sarah Kaye & Co Research is the firm’s in-house analytics practice.
It is director-led by Mike Kaye — ex-Accenture global partner with formal qualifications in property law, valuation, economics, and engineering (including a Master’s in both Engineering and Economics), and over two decades of advising global executives on data-led decisions. Mike has also lived in Sydney’s Northern Suburbs for over twenty years. The research practice draws on twenty years of on-the-ground market knowledge built up by the firm’s co-founder Sarah Kaye, who has worked the Northern Beaches and North Shore market every day for two decades — on the phones, in the cars, with the agents and buyers and houses themselves.
The research practice operates in close partnership with Anthropic’s Claude — among the most advanced analytical AI systems available, developed by Anthropic, the AI research and safety company whose customers include some of the world’s largest corporations, governments, and research institutions. Day to day, Claude functions as a senior analytical partner available without time constraint — running queries across our 20,000-plus-row registered-sales master in seconds, pattern-matching across years of data with a consistency a single human analyst cannot match, and supporting the methodology behind every claim we publish.
It is not a fifty-person research department. It is a director-led practice with AI partnership, plus external analyst support when a question demands it. What it lacks in headcount it makes up for in three things — corporate-grade analytical methodology that most buyers’ agencies are not built around; the ability to run queries in seconds that traditional research teams would need weeks for; and a depth of domain knowledge in our specific 57-suburb area — the Northern Beaches, the Lower North Shore, and the Upper North Shore — that nobody else is matching.
How we work now
We pull every NSW Valuer General registered settlement in our coverage area weekly, automatically. Every claim we make is traceable to a specific row of registered data. We never quote a price without checking the registry. We never extrapolate from a portal headline. We never amplify an agency-commentary number we haven’t tested ourselves.
As of May 2026, the master file contains 20,204 arm’s-length residential sales across 57 suburbs in Sydney’s Northern Suburbs — a 24-month window. Sample sizes for any region-and-property-type comparison run in the hundreds. For an individual suburb-and-property-type cell, in the tens to low hundreds. This is enough to do honest work.
On freshness — and why our limit is everyone’s limit
There is one constraint worth being open about. The NSW Valuer General publishes its weekly bulk feed every Monday morning, containing settlements registered with Land Registry Services in the prior week. The lag between an actual property settlement and that settlement appearing in the public bulk feed is, on average, six days. Sixty-seven per cent of settlements appear within seven days. Ninety-four per cent appear within fourteen days. The freshest week of data in our master is therefore typically one to two weeks behind real-time settlement activity.
This is not a Sarah Kaye & Co Research limit. It is a public-data limit that applies to everyone. There is no fresher complete dataset available outside the small group of approved bulk-data licensees — Cotality, Pricefinder — and even their data flows through the same Land Registry Services registration pipeline, with their own commercial layer added on top. Selling agents have private knowledge of recent transactions in the days immediately after exchange, before settlement, but only for transactions they personally have been involved in. That coverage is partial and patchy by definition. The Valuer General bulk feed is the freshest complete dataset that exists for the suburbs we cover. We use it because it’s the best available.
Why nothing can be hidden from this dataset
One more point, on completeness. The NSW Land Registry Services record is the legal record of property ownership in this state. Under the Real Property Act 1900, registration of a transfer is mandatory — without registration, the legal title does not change hands. The Torrens register is, by design, a public document.
There is no legal mechanism, and no practical one either, by which a buyer, a vendor, or a selling agent can suppress an individual residential sale price from this dataset. Some buyers prefer privacy in publicity. Some agents are pressed by vendors not to disclose the result. Some readers assume that a celebrity buyer or a discreet sale somehow doesn’t appear. None of that affects the registry. Every settled residential transaction in our coverage area, regardless of the buyer, regardless of the agent, regardless of the commercial preferences of either party, lands in the public record. Tom Cruise could buy a Mosman waterfront tomorrow and the registered settlement price would appear in our master file one to two weeks later. So would the price of a quiet off-market deal struck between two old friends in Wahroonga, or a deceased estate that never went to the open market in Avalon Beach. The public commentary may go silent on these transactions. The registry never does.
This matters because much of the property data quoted in mainstream commentary depends on which sales agents have chosen to disclose. Industry disclosure patterns tend to favour the prices that flatter a campaign; quieter treatment falls on the ones that don’t. The Valuer General’s bulk feed has no such filter. It is the unedited record of every settled sale.
What we won’t claim
The dataset doesn’t carry bedroom counts, days on market, or buyer demographics. When a question requires those, we source them separately and say so. The most recent month is always slightly understated because of the registration lag. Some related-party transfers slip through arm’s-length filters and have to be cleaned manually. There are no perfect datasets. There is just the most direct one available, used carefully, with its limits stated up front.
What’s coming
Over the next several months, Sarah Kaye & Co Research will publish:
- A finding from the Upper North Shore, now published as part of our autumn 2026 Sydney’s Northern Suburbs price tracker, on three rail-line suburbs whose unit prices have risen meaningfully — in excess of 10 per cent over the recent settlement window — while their neighbours’ detached-house markets have softened.
- A methodological piece that underpins everything else — why the headline phrase “Sydney property is up x per cent” is structurally misleading for buyers in our coverage area, and why we believe regional analysis at the Sydney level has stopped being fit for purpose for our cohort.
- A public correction of one of our own February 2026 reports, where the registered data has now disproved a specific claim we made.
- Several pieces that, calmly and with sources, push back on specific claims made by leading national forecasters and on the national-median framing common in mainstream property coverage about our patch.
- Six suburb-specific deep dives — Manly, Avalon Beach, Wahroonga, Killara, Mosman, and Cammeray — each surfacing something that surprised us when we ran the data.
- A trophy macro piece comparing the Sydney-wide property narrative against what is actually happening in our 57-suburb patch, built on third-party national data on one side and our 20,000-plus settlement master on the other.
Why this matters
We work for buyers exclusively. Buyers deserve evidence, not commentary. When the registered data tells us we got something wrong, we publish that too — see the Mosman correction we’ll run shortly. The most useful number for a buyer in our coverage area is the one that came from the registry, not the one that came from a press release.
Sarah Kaye & Co Research exists for that reason. It is a small practice. It is director-led. It is built around the most direct data source available. And it is fiercely on the side of the buyer.
Welcome aboard.
— — —
Methodology and sources
Source dataset: NSW Valuer General Property Sales Information bulk feed, accessed under the NSW Open Data Policy and the Creative Commons BY-NC-ND 4.0 licence. Refreshed weekly via automated pipeline. Coverage: 57 suburbs across the Northern Beaches, Lower North Shore, and Upper North Shore. Time window: May 2024 to May 2026. Total observations: 20,204 arm’s-length residential sales (units priced ≥ $300,000; houses priced ≥ $500,000; portfolio-style same-day same-price clusters of three or more sales excluded).
Settlement-to-publication lag distribution measured from 2026 weekly bulk-data files (n=10,740): median 6 days, mean 11 days, 67 per cent within 7 days, 94 per cent within 14 days.
External references cited or implied: Cotality (RP Data) Home Value Index April 2026; Australian Financial Review property coverage 19 April 2026 and 13 April 2026 (Lucy Slade, Michael Bleby); PEXA Property Insights calendar 2024 and FY25 reports; ANZ Australian Property Outlook April 2026; NSW Real Property Act 1900.
— — —
About the author
Mike Kaye is Co-Founder and Director of Sarah Kaye & Co., a boutique, director-led buyers agency working across Sydney’s Northern Suburbs — the Northern Beaches, the Lower North Shore, and the Upper North Shore. Mike leads Sarah Kaye & Co Research, the firm’s in-house analytics practice. A former Accenture Global Partner and Australian Army Officer, he studied property law, valuation, and economics at UNSW and is a Graduate of the Australian Institute of Company Directors. His commentary has been profiled in the Australian Financial Review, the Sydney Morning Herald, and AICD Magazine. As a Northern Beaches buyers agent Mike works exclusively for buyers — fixed-fee, no kickbacks, with over 75% of clients’ homes secured off-market.