Selling models and market structure

What the algorithm can price, and what only the block knows

6 min read

The records stop at the parcel line. The price is decided by everything around it.

The short version

Every automated valuation reads the same inputs: beds, baths, square footage, lot, year built, prior sales. Those describe the parcel. The sale price is set by the parcel plus its context, the traffic on that stretch, the school attendance line, the drainage that floods one side of the street, the neighbor whose yard tells you everything. Almost none of that context exists in a public record, which is why two homes with matching data sheets can sell tens of thousands of dollars apart. The gap between what the record captures and what the block knows is Context Blindness, and it is the permanent limit of pricing by data sheet.

The parcel is not the market

A public record describes a structure: its size, its rooms, its age, its prior prices. A buyer purchases a situation: that structure, on that street, inside that school line, next to those neighbors, with that light. Hedonic pricing research has documented for decades that micro-location factors move sale prices, school attendance boundaries and road noise among the most studied. The record and the situation overlap, but they are not the same object, and the price follows the situation.

The inventory

The variables below decide real sale prices every week. None of them appears in the data an automated model reads.

Block-level variable	Where it lives	What it does to price
Road exposure and noise	Standing on the curb at 5 pm	Discounts one side of a street against the other
School attendance line	District maps and local knowledge	Steps value at a border no record shows
Drainage and flood history	Neighbors and old repair bills	A quiet discount the listing never mentions
View and light lines	The upstairs window	A premium no field in a record encodes
The immediate neighbors	The walk from the car to the door	Moves buyer willingness in either direction
The block price break	Closed sales two streets apart	The market boundary a model averages across

Same records, different streets

The mechanism is averaging. A model prices a home from comparable sales chosen by rule, and the rule reaches across the invisible boundaries, the school line, the busy road, the block where prices step down, pulling in sales from the wrong micro-market. The estimate that results is an average of two different markets, correct for neither. Which sales make the set is the whole question, the subject of comp selection is the whole game, and the reason the wrong comps produce a confidently wrong number. The published off-market median error near 7.5%, about $45,000 on a $600,000 home, is what averaging across such boundaries looks like at national scale.

Why more data does not close the gap

Some of the inventory digitizes over time. Flood maps improve, noise maps exist, school boundaries are published. What does not digitize is the composite judgment: how the buyer pool for this specific home weighs this specific corner, this spring. That judgment is formed by walking the block, sitting the open houses, and watching which homes move, and it lives with the people who do those things. The data gap narrows at the edges. The judgment gap is structural.

What this means for an agent

The inventory above is your listing presentation. Walk the seller through the variables the record never captured, the road, the line, the light, the block, and attach each one to the comparable sales that prove it. That is the difference between disputing a number and explaining a market, the same ground covered from the model side in what a Zestimate cannot see. The appointment-ready version, built on what the automated number missed about the specific home, is our home-value accuracy page, and the honest-range consumer estimate is the home-value page.

Frequently asked questions

Why do identical homes sell for different prices?

Because the records are identical and the situations are not. Road exposure, school attendance lines, drainage history, view, and the immediate neighbors all move price, and none of them appears in the public data the two identical records came from.

What is Context Blindness?

It is the gap between what a record captures and what the block knows: the structural inability of data-sheet valuation to see the micro-location and condition context that sets real sale prices. A model prices the parcel; the market prices the situation.

Do school boundaries change home value?

Yes. Published research on school attendance boundaries finds price steps at the border between assignment zones, homes that are otherwise similar selling for different prices because they feed different schools.

Can an algorithm price road noise?

Not reliably. Noise varies within a single street and by time of day, and it reaches no public record a model reads. It shows up in closed prices, which is why comps from the quiet side of a street misprice the loud side.

How do agents price what the data cannot see?

By selecting comparable sales from the same micro-market, the same side of the boundary, the same block character, and adjusting for the context they can see standing on the curb. The comp set carries the block knowledge into the price.

Sources: Published hedonic pricing research on micro-location factors including school attendance boundaries, road noise, and flood exposure, appraisal guidance on location adjustments, and standard descriptions of the inputs available to automated valuation models. Magnitudes vary block to block; the mechanism is the finding.

This article is general information and analysis, not financial, lending, or appraisal advice. Verify any home value with a licensed professional before acting.