Reading a micro-market from filings: supply, absorption, competitor density

Consultancy reports read markets at city scale; land is bought at lane scale, and the two disagree constantly: Pune can carry rising unsold inventory while three of its wards starve for supply. The register is the only dataset granular enough to referee, because every project files its own numbers at its own address. This is the complete method for reading one micro-market from filings, the working core of everything this series has built toward.

Step one: draw the catchment honestly#

A micro-market is where your buyer actually shops: typically a 2 to 3 kilometre radius around the parcel, cut by real barriers, highways, rivers, rail lines, that divide school runs and commutes. Resist locality names: "Baner" in a brochure now covers three distinct markets. Geometry keeps the analysis honest, and repeatable when you re-run it next quarter.

Step two: the supply table#

Every registration inside the catchment, from the register: sanctioned units, unsold units, the residential against non-residential split, promised completion dates, current status, promoter entity. This is the pipeline, sworn: what exists, what is coming and when, at a fidelity no listings portal matches, because registrations cannot exaggerate without consequences.

Know the table's edge before you build on it. MahaRERA files no unit configuration, no carpet area per flat and no price. The building-level table splits residential from non-residential and stops, so a supply table built from the register is a count of units and not a product mix, and any BHK-level or carpet-band view of a catchment has been modelled from transaction data and listings by whoever is showing it to you. That is a real limit and worth stating out loud, because the method below is otherwise strong enough to be trusted past where its data reaches. Note the vintage structure while you build it: a catchment whose supply is all 2023-registered is a different market from one carrying 2019 stock still unsold. That distinction does most of the work, and the Pune read is the worked example: most of the district's filed unsold stock sits in projects under 30 percent built, which reads as alarming until the registration dates show nearly four fifths of it is simply new.

Step three: absorption from the QPRs#

This is the step where most descriptions of this method, including the first draft of this one, promise something the register cannot give. MahaRERA publishes a project's CURRENT state, not its history. Open a project today and you get the sold count as it stands; open it tomorrow and you get tomorrow's. The portal keeps no archive of last quarter's figure, so a "trended absorption series over eight quarters" cannot be derived retrospectively by anyone, however patient. It can only be accumulated by somebody who started recording.

So the step splits in two, and both halves are real.

The cross-section, available today. Take every project in the catchment and read sold-against-sanctioned beside its age on the register. Many projects of different vintages, photographed on one day. That is not a velocity series, and it must never be labelled as one, but it answers the question a land buyer actually has: in this catchment, how sold is a two-year-old project, and how sold is a five-year-old one? A catchment where the 2019 stock is still 40 percent unsold is telling you something no absorption chart would tell you faster.

The movement, if you have been watching. We have been recording every project's filed figures since the end of July, which is long enough to say what movement in this register actually looks like, and it is not what a quarterly series implies.

Three things follow for the method. Movement is rare and lumpy: 1.6 percent of projects in seven weeks, with a median riser gaining six units, so a catchment of twenty projects may show two or three moves in a quarter and a chart of it will look like noise until you have a year. A catchment is the right unit precisely because of that: individual projects are too quiet to trend, and summing across the catchment is what turns lumpy filings into a readable number. And 120 projects moved DOWN, which is the trap: a falling sold count is almost never a sale being unwound, it is a promoter re-filing the table differently, and a method that diffs naively will publish that as negative absorption and be confidently wrong.

Price none of this from the register. It gives quantity and pace; the Inspector General of Registration's transaction values and the consultancy series give rupees.

Before scoring anything, know what the surrounding region reads, or a catchment number has nothing to sit against. For MMR that baseline is in the five-district dashboard; for Pune, the pincode table puts 48 lanes on one scale and shows the spread a district average hides, 32 percent unsold at one end and 69 at the other. The same table for any other market is one district page away.

Step four: the three scores#

An analogy: this is fisheries science against fishermen's gossip. The gossip (brokers, launch parties) tells you where boats cluster; the catch data tells you where fish remain. Fleets that fish on data outcatch fleets that fish on gossip, not because gossip is always wrong, but because it is always late: by the time a locality is famous, its months-of-inventory number has usually already turned.

What the method catches#

Anita Rao's standing example (illustrative, as ever): two adjacent Pune catchments, identical reputations, city-level data lumping both into one healthy ward. The filings split them: catchment A at 11 months of inventory with absorption accelerating and one distressed project; catchment B at 31 months, velocity concentrated in a single discounter, and three silent filers. Her firm's land broker had priced both identically. The register disagreed by twenty months of inventory, and the launch that followed went where the fish were. A lender reading the same two catchments sees the other side of it: where a credit book's risk concentrates.

Where to run it free, and what a report adds#

The supply table is assemblable by hand today at no cost. The free district pages list every registration in a district with its status and promised dates, and each project page carries the units, the progress, the extensions and the disclosed disputes. For a single catchment of a dozen projects that is an afternoon's work, and it is the right way to learn what the filings do and do not say before paying anybody for a summary of them.

What costs money is repetition and memory. The area market report is a flat Rs 2,999 per pincode and assembles this method's output for a Maharashtra micro-market, and it carries the second half of step three as far as it honestly goes: where ReraGenie's own record, kept since July 2026, is deep enough for the pincode, the report shows the measured movement the register itself does not retain, and it labels every other demand figure as a cross-section.

The one-line summary#

Draw the catchment with geometry, table the supply from registrations, read absorption as a cross-section and accumulate the series yourself, then score inventory-months, density and distress: one page, refreshed quarterly, and the lane-level truth that city reports average away becomes the cheapest edge in the market.

Methodology and sources#

• Booking movement over the observation window: ReraGenie trend ledger, 15 observation dates between 31 July and 17 September 2026, covering 58,573 MahaRERA projects carrying a filed sold-unit count. 960 projects (1.6 percent) filed a changed count; 840 rose and 120 fell; net movement +14,789 units; median gain among risers 6 units, 90th percentile 41.
• Unit-table coverage: only about a third of registered projects file the building-level unit table the unsold count is derived from, so any months-of-inventory figure is computed over the projects that file it and should carry that denominator.
• The register holds current state only. MahaRERA publishes no archive of a project's earlier filings, which is why a retrospective absorption series cannot be constructed from the portal by anyone and why the movement figures above exist only because they were recorded as they changed.
• Transaction values and price per square foot come from the Inspector General of Registration, Maharashtra, and from private consultancy series. No price figure in this method comes from RERA filings, because RERA carries none.
• Nor does RERA carry unit configuration or carpet area per flat: the building-level table splits residential from non-residential units and records nothing finer. Any configuration-level or carpet-band view of a catchment is modelled from outside the register, and this method does not produce one.