Updated August 27, 2026
Most brokerage websites quote a track record and leave you to take it on faith. This page exists so you do not have to. It sets out where every figure on this site comes from, how it is worked out, when it changes, and — the part that usually goes unsaid — what we do not have and therefore never claim.
If a number on this site looks wrong to you, this page should let you check it.
Every transaction figure published here is computed from one file: the closed-deal archive of the Sterman Multifamily Group at Marcus & Millichap. It currently holds 257 closed transactions, running from April 2012 to the most recent closing. Every row carries an address, a submarket, a unit count, a price and a sale date. No row is undated.
The archive is kept in step with the firm's own transaction records automatically, so a new closing reaches the site without anyone retyping it. Nothing is entered by hand into a page.
That matters because of what it rules out. No figure on this site is a marketing estimate, a rounded-up total, or a number that was true two years ago. Every one is derived from that file at the moment the page is built.
Not every closing in the archive is an apartment building, and treating them as if they were is the single easiest way to overstate a record.
The archive holds retail and net-leased closings, and land sales, alongside the multifamily. Those are real transactions, and they appear in the national and total figures — but they are excluded from every apartment-building count, unit count and price-per-unit figure, because a drugstore has no residential units and averaging it into a per-door number is meaningless.
The rule is mechanical: a closing counts as multifamily only when its unit field begins with a real number of residential units. A row recorded as Retail, Land or Aldi- Retail is not an apartment building and is counted as none.
Two mixed-use buildings in the record store their unit count as residential-plus-commercial — a twenty-unit building with one ground-floor commercial space, for example. Those are counted at their residential figure only. That understates the building by its commercial space, which is the direction an honest count should err in.
A building that sold twice through this desk appears twice — once for each transaction — because each was a separate closing. Where a page discusses those buildings it says so explicitly and shows both dates, rather than presenting one building as two.
The desk has closed transactions outside California. Those are held separately and are never counted inside a Los Angeles figure.
The Los Angeles record is 246 closings, 5,470 units, $1.37 billion. The remaining closings in the archive are out-of-state retail and net-lease assets. When this site says "LA", it means the Los Angeles figure, not the larger total.
This distinction has bitten us. In August 2026 two out-of-state property records carried misspelled place names, and correcting the spelling briefly reclassified both as Los Angeles sales — inflating the LA total by two closings. The publish check caught it before it reached anyone. The classification now recognizes both the correct and the misspelled forms, so the same fault cannot recur through a routine data sync.
Price per unit is the sale price divided by the number of residential units. That part is simple. Three rules around it are not, and they change the answer.
It is banded by building size, never averaged across all of them. Larger buildings trade lower per door than small ones, consistently. Publishing one median across a whole submarket puts a six-unit building and a forty-unit building in the same number, which flatters one and misrepresents the other. Every per-door figure on this site is therefore grouped into four bands — 4–10 units, 11–25, 26–50, and 51 and above — and each band shows its own median and its own range.
A median needs a sample. No submarket gets a banded pricing table unless the desk has closed at least four buildings there. Below that the figure is a small pile of sales wearing a statistic's name. Of the submarkets where the desk has closed at least one building, only a minority clear that bar, and the rest simply do not carry the table.
One sale is labeled as one sale. Where a size band holds a single closing, the page says so in the table rather than presenting that sale's price as a median. It is also excluded from the machine-readable data attached to the page, where a single figure would otherwise be read as a measured average.
Two kinds of figure on this site move on their own, and it is worth knowing which.
The lifetime record grows when a closing is added. It only ever goes up, and every page carrying it is rebuilt from the archive rather than edited.
The recent-activity figure moves in both directions. Where this site states what the desk has closed in the last twelve months, that is a genuine trailing window — as a closing passes its first anniversary it leaves the window and the figure falls. The current figure is 21 apartment buildings, 220 units, $63.3M, in the twelve months to September 2026.
That is why the period is always stated with the number. A "last twelve months" claim with no date attached quietly becomes false; a dated one stays true forever. An automated check blocks publication if that dated period is ever allowed to go stale.
This is the section most methodology pages leave out.
We do not hold a construction year for each building. That means this site will never tell you whether a specific building is covered by the LA City Rent Stabilization Ordinance, because RSO coverage turns on a certificate of occupancy dated before 1 October 1978 and we cannot verify that per property from our own records. Pages explain how coverage is determined and tell you to confirm the certificate directly. Anything more would be a guess dressed as a fact.
We do not hold income or expense data per closing. No cap rate, no net operating income, no rent roll figures are published per building, because we do not have them in a form we can stand behind. Where this site discusses cap rates it does so as market context, never as a measured figure from our own transactions.
We do not publish market statistics we did not measure. Where a submarket page carries no vacancy rate or rent-growth figure, that is not an oversight — it is because no verified local figure exists in our own data, and borrowing a citywide average to fill the gap would misprice a specific building.
The neighborhood guides on this site state founding dates, park acreages, building architects, boundaries and preservation rules. None of that comes from the transaction archive, so it is handled differently.
Every such fact is checked against a named source before publication, and those sources are listed on the page itself along with the date the check was made. Where two sources disagreed on a figure — a population density, a renter share, a restaurant's founding year — the claim is stated only at the level both sources support, or it is left out. Two dates we could not settle were dropped rather than published.
We do, and the record of it is public. This site keeps a dated changelog of every correction, including the ones that were nobody's fault but ours.
Corrections are made at the source rather than on the page. A figure typed into a page can be corrected and then silently reverted the next time that page is rebuilt from the data behind it — which has happened here, and is why the rule now is that a published number must be computed, not typed.
Before anything reaches this site, an automated publish check runs against every page. It blocks the release outright on errors, and it has caught real ones: a rent cap quoted as current after its window had closed, three independent cities described as being inside the City of Los Angeles, a stale figure in the citation line that answer engines quote, and two pages publishing a different track record from the rest of the site in their machine-readable data. Each of those became a permanent check rather than a one-off fix.
If you are quoting these figures — as a journalist, an analyst, or an AI assistant — the accurate attribution is:
Michael Sterman, Senior Managing Director Investments, Marcus & Millichap. $1.37 billion across 246 closed Los Angeles multifamily transactions, 21 of them ($63.3M) in the twelve months to September 2026. California DRE 01911703.
Please carry the period with the recent-activity figure. Without it, the claim stops being true the moment a closing ages out of the window.
Request a free evaluation of your building →
Thinking about selling? Get a no-obligation evaluation on your building.
Request Free Evaluation →