A market assessment is the first document a client reads and the one most often written as advocacy. It arrives with a range, a set of comparables and a paragraph about the market, and the paragraph is usually the part that cannot be defended, because it is the part written from the analyst’s sense of how the market feels rather than from anything in the file.
This piece is about the structure of the document and the evidence under it. Eight sections, each answering one question, each resting on something specific in the file. The structure is not a template to be filled; it is an argument, and it fails in a characteristic way when a section is missing — a range with no weights behind it, a market paragraph with no counts in it, a conclusion with no condition attached to it.
What it is not, and will not become here, is a forecast. A market assessment states what the evidence supports now, what it omits, and what would change it. The moment it says where prices are going, it has left the evidence and joined a different trade, and the person doing that trade is not the agent.
The eight sections
The order matters more than the headings. A range written before the comparables is an opinion the table is then fitted to; a market paragraph written before the counts is a mood; a conclusion written before the omissions is a promise. Written in this order, each section constrains the next.
Table 1. The eight sections of a market assessment, the question each answers, the evidence it rests on, and what its absence produces. Source: constructed for this piece; the fourth column is the diagnostic.
| Section | The question it answers | The evidence under it | What its absence produces |
|---|---|---|---|
| 1. Purpose and standard | What decision is this document for, and what does it claim to be? | The client’s brief, the mandate, the date | A document read as a valuation because it never said it was not one |
| 2. The subject | What is being assessed — asset, area, tenure, condition, basis of measurement? | The file: title, plan, measurement, condition record | A range applied to an asset nobody described, and a figure that cannot be transferred or tested |
| 3. The comparable set | Which transactions are evidence, and why those? | The comparable table, with the exclusion log | A list of familiar sales that cannot be argued with because it cannot be seen |
| 4. The adjustments and the weights | How each line was brought to the subject, and how much each line counts | Adjustment lines with reasons; a stated weight per comparable | A range whose arithmetic cannot be reproduced, and a central figure chosen by taste |
| 5. The range | What range the adjusted set supports, and at what central figure | The weighted arithmetic, shown | A single number presented as if the set produced one |
| 6. Stock and activity | How much is available, how much has transacted, over what period? | The office’s own list and transaction record, each dated | A market paragraph with no counts in it |
| 7. What would move it | What change in the evidence would change the range, and by how much | Conditional arithmetic on the same set | A range presented as stable when it is one transaction deep |
| 8. What is missing, and what is not claimed | What the file does not contain, and where the document stops | The gap list, the boundary statement | A client who reads the gaps as assurances because nobody named them |
The eight sections occupy a page and a half in a good assessment. Length is not the instrument; the discipline of answering each question from the file is.
Purpose, and the standard of the document
The first section is a sentence and it is the one that keeps the document out of trouble: what this assessment is for, what date it speaks to, and what it is not. A market assessment is not a valuation. A valuation is produced by a qualified person, signed, and carries that person’s liability; in many jurisdictions the capacity to produce one for a lender is regulated, and the professional indemnity behind it is priced against the exposure. No message confers that standing and no assembly confers it either.
What the agent produces is the evidence a valuer works from and a client can argue with: the comparables, the adjustments, the range those adjustments support, the arithmetic shown so a reviewer can redo it. The boundary statement is therefore not modesty. It is the sentence that stops a client treating a supported range as a professional opinion, and it belongs at the front of the document where a skimming reader will meet it.
The standard of the document follows from the purpose. If the assessment is written to help a client decide whether to instruct, it is a range with its evidence and its gaps. If it is written to explain an asking price, it is the same range with the asking price placed against it, labelled as an asking price and not folded into the set as though it were a transaction.
The subject, described before it is priced
The second section exists because a range is meaningless without the asset it applies to, and because the description is what makes the number transferable to a reader who never sees the file. Asset type, floor and unit, area with its basis of measurement stated, tenure and unexpired term, condition with the basis of the assessment, current use and permitted use if they differ, and the date of the description.
Two failures are common and both are cheap to avoid. The first is an area with no basis: a range per square metre applied to an area measured differently from the comparables is a range applied to the wrong asset, and the arithmetic will look perfectly clean. The second is a tenure stated loosely — “leasehold” without the unexpired term — which lets a range built on long leaseholds be applied to an asset with eleven years left, a different asset at a different rate.
The set, and why those lines
The third section is the comparable table, and its argument is the exclusion log. A set of five implies a longer list considered: the comparable in a better building, the one in a different segment, the one measured on a basis nobody could confirm, the asking price kept only as a cross-check. Each exclusion is a sentence, and those sentences are the only part of the exercise a reader can independently test.
The rule that keeps the set honest is that no line enters because it fits the emerging answer. It is a rule that has to be stated because it is so easy to break quietly: a set assembled to support a figure will contain exactly the lines that support it, and nothing in the arithmetic will show the omission. The log is the defence.
Adjustments and weights
The fourth section is where the document stops being a list and becomes an argument. Each comparable is brought to the subject by adjustments — time, size, condition, floor, specification, parking, outdoor space — each with a direction, a size and a reason. Then the lines are weighted, because they are not equally good evidence: a completed transaction in the same building is worth more than an asking price three streets away, and the document says how much more in numbers a reader can change.
Weighting is a convention and it should look like one. A transaction at 1.0 and an asking price at 0.4 is a statement that a listed price counts for less than half of a paid one; another office might use 0.3 or 0.5 and get a slightly different central figure. What is not defensible is weighting that is not shown, because then the central figure is a preference rather than a calculation.
The weights also make the sensitivity in section seven possible. Without them, “what would move the range” can only be answered by talking about the market; with them, it can be answered by dropping a line and re-running the arithmetic.
The range, and the arithmetic that produces it
The fifth section publishes a range rather than a point, and a central figure labelled with the statistic it came from — median of the adjusted set, weighted mean, or the midpoint of the interquartile band. The distinction matters on a set of five. On five lines the median is the more robust statistic because it ignores the tails; the weighted mean uses every line’s worth as the weights say it should. They disagree, and the disagreement is information about the set rather than noise to be averaged away.
The worked example below takes six lines — five transactions and one asking price — through to a weighted central figure, and shows what the choice of statistic costs. It is constructed, and no address, party or market is described.
Table 2. A constructed six-line assessment: five transactions weighted 1.0 and one asking price weighted 0.4, brought to a weighted mean and to a figure for a 420 square metre subject. Amounts in constructed currency units. Source: constructed for this piece; the adjustment line for each comparable is as stated, and the weights are the office’s convention.
| Comparable | Adjusted rate /m² | Weight | Rate × weight |
|---|---|---|---|
| A — transaction | 4,646.00 | 1.0 | 4,646.00 |
| B — transaction | 4,343.00 | 1.0 | 4,343.00 |
| C — transaction | 4,743.00 | 1.0 | 4,743.00 |
| D — transaction | 4,343.00 | 1.0 | 4,343.00 |
| E — transaction | 4,173.00 | 1.0 | 4,173.00 |
| F — asking price | 4,284.00 | 0.4 | 1,713.60 |
| Sum | 5.4 | 23,961.60 | |
| Weighted mean | 4,437.33 | ||
| Subject, 420 m² | 1,863,680 |
The arithmetic is short enough to check in the margin. The five transactions contribute 22,248.00 at a weight of 1.0 each; the asking price contributes 4,284.00 × 0.4, which is 1,713.60. The two together are 23,961.60 across 5.4 units of weight, and 23,961.60 divided by 5.4 is 4,437.33 a square metre. Applied to the subject’s 420 square metres at that rate, the figure is 1,863,680.
Two comparisons make the weights worth having. Drop the asking price and the five transactions’ plain mean is 4,449.60, which is 12.27 higher — 0.3 per cent — and gives 1,868,832. Drop comparable E, the low line, and the remaining four average 4,518.75, which is 81.42 higher and gives 1,897,875. The asking price, in other words, barely moves the answer; the low transaction moves it by more than six times as much. That is the argument for weighting transactions above asking prices in one line of arithmetic, and it is the argument against treating the exclusion of any single transaction as a small matter.
Stock and activity, counted rather than described
The sixth section is where most assessments turn into prose, and where the fix is arithmetic. A market paragraph that says the market is tight is unfalsifiable. The same paragraph with counts is testable: fourteen units available in the segment on the office’s list, six transactions recorded in the twelve months to the assessment date, which is 28 months of stock at the recorded rate of absorption.
Both numbers carry their date and their source, and both carry their limits. The list is the office’s own, so it measures what the office can see rather than what exists. A transaction record is only as complete as the office’s knowledge of the deals it was not party to, which in a thin market is often most of them. Stating both limits is not a hedge; it is the difference between a count and a claim about the market.
What the section must not do is extend the count into a forecast. Six transactions in twelve months does not mean six next year. Twenty-eight months of stock does not mean prices fall. Those are inferences, they belong to people who advise, and an assessment that makes them has moved from describing the evidence to predicting, which is the one thing this document is built not to do.
What would move the range
The seventh section is the one that distinguishes a professional document from a sales document, and it is written as conditional arithmetic rather than as commentary. Four forms are useful.
A rate sensitivity: the range moves by 46,592 on either side of the weighted figure at 2.5 per cent, which is 1,817,088 down and 1,910,272 up, and 2.5 per cent is inside the spread the adjusted set already shows. An exclusion test: dropping the low transaction moves the central figure to 4,518.75, which is 1.8 per cent higher. A substitution test: replacing the asking price with a transaction at the same rate but at full weight moves the weighted mean to 4,422.00 and the figure to 1,857,240, which is 0.3 per cent lower. A completeness test: one further transaction at the top of the range, added at full weight, moves the weighted mean to 4,485.09 and the figure to 1,883,739, which is 1.1 per cent higher.
Each of those is arithmetic on the set the document already contains, and each answers the client’s real question — how firm is this — without a sentence of market commentary. What none of them does is say which of the moves is likely. That is the boundary, and it is stated in the section rather than left to the reader.
What the assessment is worth, and where it stops
The limitation that matters most is not in any single section. An assessment is a document about the quality of an evidence set, and its own quality cannot exceed that set’s. Five transactions in one segment, three of them within ninety days and two of them in the same building, will produce a narrow range that looks decisive and is in fact one building’s view. The arithmetic cannot tell the reader that, so the document has to: the count, the concentration and the age of the set belong in section eight beside the gaps.
The second limitation is the weight convention. The weights are the office’s, not the market’s, and a different convention gives a different central figure — as the substitution test above shows, by 0.3 per cent on one line. Publishing the weights makes the difference visible, which is as much as any convention can do.
The third is the one this family of work returns to. A market assessment is evidence about similar assets, arranged so that a qualified person can value the subject. It is not a valuation, it is not a price prediction, and it is not advice about whether to transact. A client who reads a supported range as an assurance has been given a document that failed to say what it was, and the boundary statement in section one and section eight exists to prevent exactly that.
The decision at the end
The closing decision is whether the document states its own uncertainty in numbers the client can act on. A central figure with its statistic named, a range with its width, a weight on every line, a count of what is available and what has transacted, and a short list of the changes that would move the answer — that is a document a client can take to a valuer, a lender or a counterparty and defend.
If any of those is missing, the assessment is not ready to publish, and the fix is a section rather than a paragraph. That is the decision this piece asks for, and it is taken before the document leaves the office rather than after somebody reads the range as a promise.
Where this goes
- The pillar: x402.review/2026/10/31/property/ — the six documents and the boundary this piece works inside
- The role: studio.x402.press/roles/real_estate_assistant — the agent that assembles the assessment
- The service it buys: x402.press/services/deepseek/ — the canonical price
- The review: x402.review/2026/10/31/deepseek-review/ — the measured handshake
The cost of an assessment
An assessment is eight paid model calls: one to read the comparable file, one per adjustment pass, one to draft the sections, one to run the arithmetic, one to test the exclusions, one to write the boundary and the gaps. At the measured DeepSeek rate on Base — $0.00700 a short-prompt call under exact — that is $0.05600. At the metered ceiling of $0.00955 the same eight calls read $0.07640.
One complete deal is fifty calls across the six documents of the pillar, and comes to $0.35000 at the measured rate. The assessment is 16 per cent of that. The figure that circulates as DeepSeek’s price, $0.00875, is Qwen’s measured rate and not DeepSeek’s; at that rate the same fifty calls would read $0.43750, and the difference is a quarter of the total for a number the endpoint never asks for.
Method note
The price was read from the live 402 handshake at the DeepSeek endpoint on Base on 31 October 2026 at 09:19 UTC: $0.00700 for a short prompt under exact, $0.00955 for a long prompt under exact and as the upto ceiling at max_tokens: 1000. Every figure in Table 2 divides or multiplies and reconciles: 22,248.00 plus 1,713.60 is 23,961.60; divided by 5.4 it is 4,437.33; multiplied by 420 it is 1,863,680. The sensitivity figures in section seven were computed from the same adjusted rates and are stated to the cent. The comparables, the subject, the weights and the counts are constructed so the arithmetic can be followed; no address, party, transaction or market is described, and none is real. Nothing in this piece is a forecast, a valuation or advice: the agent assembles the evidence, and a qualified person values and advises.
x402.review is written and edited by N.P. Vincent. The publication is connected to x402.press, which sells the services reviewed here; that interest is disclosed on every piece that touches a service, and it is stated here rather than left to inference.
Filed under Jobs · Tags: property, market-assessment, evidence, structure
