Comparable sales and asking-to-final patterns

Abstract

This note examines how final prices differed from asking prices in a reviewed, Stockholm-focused cohort. It first describes 7,397 eligible sale rows, then compares each sale with similar sales using observed area, size, room count, and timing characteristics. The public result is deliberately narrower than an explanation of why prices differ: it publishes distributions and anonymous group summaries, not identities or rankings.

The median asking-to-final uplift was 6.7%. Across 30 anonymous groups, matched estimates appeared in both reviewed bands: below zero and zero or more. These patterns describe the reviewed cohort; they do not establish market value, intent, or a causal broker effect.

Central result

Asking and final prices often differed, but the reviewed aggregates cannot explain why they differed.

1. Dataset Overview

The source period runs from 5 June 2013 to 27 January 2026. The figures below describe the eligible rows retained for this analysis, not every sale in Stockholm.

Eligible sale rows: 7,397
Median uplift: 6.7%
Mean uplift: 7.69%
Directional comparisons: 23,958

Unit of observation

One eligible sale row contributes one asking-to-final percentage. A directional comparison is a separate analytical row that contrasts one focal sale with one matched sale; it is not a distinct property sale.

2. Methodology

2.1 Asking-to-final uplift

The descriptive measure is the percentage difference between the recorded final price and asking price. It treats asking price as a reference point, not as an estimate of the property’s market value.

Asking-to-final uplift (final price − asking price) ÷ asking price × 100

2.2 Comparable-sale design

Each focal sale is compared with sales carrying a different recorded broker label when the properties share an area and room count, differ by no more than 10% in size, and occur within 90 days. A focal sale retains at most five comparison rows.

Match rule Same area + same rooms + size within ±10% + sale within 90 days

For every retained comparison, the analysis subtracts the comparison sale’s uplift from the focal sale’s uplift. Those directional differences are summarized by anonymous group. The stored analysis uses 1000 row-level bootstrap iterations; the public snapshot does not publish group confidence intervals or significance classifications.

3. Analysis Quality

The reviewed snapshot records enough information to assess the comparison structure, while also showing where interpretation must stop.

Reviewed comparison-design checks
Review check	Reviewed aggregate	Interpretation
Eligible cohort	7,397 sale rows	A defined historical cohort, not a market census.
Comparison volume	23,958 rows	Comparison rows can reuse the same underlying sale.
Timing balance	43.4 days apart on average	Matches are close in time under the stated rule.
Public grouping	30 anonymous groups	Only broad aggregate bands are released.
Resampling	1000 iterations	Row-level resampling does not fully model repeated-sale dependence.

Quality assessment

The matching rule makes the comparison more specific than a raw asking-to-final average. It does not make matched properties identical, and it does not remove differences that were not recorded.

4. Findings: Broker Effects

Two static views replace the live charts and ranked explorer: the full eligible-sale distribution and a deliberately broad anonymous-group summary.

Figure 1

Asking-to-final uplift across eligible sale rows

n = 7,397

Below asking 11.6%

859 sale rows

0% to below 5% 29.3%

2170 sale rows

5% to below 10% 24.9%

1842 sale rows

10% to below 15% 17.5%

1297 sale rows

15% or more 16.6%

1229 sale rows

Counts and shares of the filtered cohort in five uplift bands. “Below asking” is descriptive: asking price is not a measure of market value. Percentages may not sum to exactly 100 because of rounding.

Figure 2

Anonymous matched estimates occupy both reviewed zero bands

30 groups

Below 0 pp 43.3%

13 groups

0 pp or more 56.7%

17 groups

Exact values shown in Figure 2
Estimate band	Anonymous groups	Share
Below 0 pp	13	43.3%
0 pp or more	17	56.7%

Each estimate is the group’s average percentage-point difference from its matched comparison rows. The broad bands avoid publishing identities, rankings, group-level values, or inferential labels.

What the static evidence shows

Most eligible final prices were at or above asking. In the matched analysis, 13 anonymous group estimates were below zero and 17 were at or above zero. Variation remains visible without turning the groups into a ranking.

5. Data Explorer

The interactive paper’s broker and area explorers are replaced here by a small, fixed set of reviewed aggregates.

Mean uplift: 7.69%
Median uplift: 6.7%
Average match gap: 43.4 days
Anonymous groups: 30

Why this view is static

The page reads only a versioned publication snapshot. It does not query a live service or expose names, addresses, areas, prices, individual sale rows, or group rankings.

6. Area Effects

The source paper includes a named-area table. This static public port keeps that step visible but does not reproduce the table because area names, rankings, and area-level values are outside the reviewed aggregate snapshot.

Area result not published

No reviewed anonymous geographic aggregate is available at the public snapshot boundary. Accordingly, this note makes no claim about which areas had higher or lower asking-to-final patterns.

7. Conclusions

1

Asking is a reference point

The median eligible sale finished 6.7% above asking, while the cohort spans bands below asking through 15% or more above asking.

2

Comparable sales narrow the question

The match rule compares observed property and timing characteristics. Anonymous group estimates occur both below zero and at zero or more.

3

The result is descriptive

The evidence supports observed variation in this cohort, not a claim about intent, market value, or a causal broker effect.

How to read this public version

The section order mirrors the analytical paper—dataset, method, quality, findings, explorer, and conclusions—but every displayed result comes from the reviewed static snapshot. The former explorer is represented by safe aggregates rather than live or identifying records.

Limits

What cannot be concluded.

Asking price is a seller-side reference point, not observed market value. Uplift cannot establish intent, underpricing, or overpayment.
Matching does not observe condition, floor, view, renovation, micro-location, marketing, seller constraints, or every other property difference.
Reused directional comparison rows are bootstrapped as rows, so the intervals do not fully model dependence between repeated sales.
The collection came from a Stockholm-focused workflow and may include sales outside a narrowly defined Stockholm market. It is not a census.
The two anonymous group bands do not publish uncertainty intervals and must not be read as statistical classifications.
No names, rankings, addresses, areas, prices, or sale records are part of the public snapshot.

This is a descriptive public-data note, not a valuation, prediction, or recommendation.