A typical observation was about 22% away
That is the typical distance between one observed sale and the median for its exact card/variant group.
PokeSplain Research · September 20, 2026
ONE SOLD COMP
ISN'T THE MARKET
What 389,602 eligible Pokémon completed-sale observations reveal about judging a card's market from a single sale.
389,602eligible completed-sale observations analyzed*
52.75%more than 20% away from the median for their exact card/variant
21.61%typical distance from the median
23,012groups of the same exact card/variant with at least 3 observations
40.27%middle-half price spread relative to the group's median
*These are completed-sale records that passed the study's filters. We call them observations because we could not independently verify that every record represents a unique payment transaction.
The short version
One sale tells you what one copy sold for.
It does not necessarily tell you what every copy of that exact card is selling for.
PokeSplain looked at repeated completed-sale records for the same card and variant. The result was simple: Pokémon cards often do not sell at one neat, consistent price. A single Sold Comp can be real evidence and still be an incomplete picture.
More context beats one isolated sale.Look at multiple relevant Sold Comps, then check the card identity, language, condition, grade, timing, and spread.
Three numbers to remember
More than half were over 20% away
More than half of eligible observations landed more than 20% above or below their group's median.
The middle half covered a wide range
The middle half of observed prices stretched across a range equal to about 40% of the group's median.
In plain English
What does that actually mean?
Imagine a card with a median observed sale price of $100. Seeing one copy sell for $120 does not automatically mean the card suddenly has a $120 market price.
Likewise, seeing one copy sell for $80 does not necessarily mean the market collapsed.
Both sales are real evidence. Neither one, by itself, tells the whole story.
Illustrative example only$80 · $100 · $120
These numbers are a simple illustration, not a direct conversion of the study's aggregate percentages and not a claim about any specific card.
Why one Sold Comp can mislead
A completed sale has context.
The study did not prove that any one factor caused a particular price difference. But any single listing may reflect details that matter when you compare it with another copy:
ConditionTwo copies may not be equally clean, played, damaged, or described.
TimingBuyer interest and supply can change from one week to the next.
Seller and buyer circumstancesA quick sale and a patient sale can happen at different prices.
Grade and languageRaw, graded, English, and Japanese copies are different comparison groups.
Marketplace and auction dynamicsAudience, format, bidding, and visibility can vary by listing.
Shipping and outliersPrice treatment and unusual records can make a single number look more important than it is.
What the data showed
The pattern held across the study.
The core result was not tied to one arbitrary date range. The same broad pattern appeared in 30-, 60-, and 90-day views.
| Window | Eligible observations | Same-card groups with 3+ observations | Groups with 20+ observations |
|---|---|---|---|
| 30 days | 212,049 | 15,960 | 2,828 |
| 60 days | 337,537 | 21,134 | 6,374 |
| 90 days | 389,602 | 23,012 | 7,994 |
A closer look
More sales didn't automatically mean more consistent prices.
English, Japanese, raw, and graded groups were not identical.
Important context
English groups had a 43.45% median middle-half spread; Japanese groups had 33.37%. Raw groups had 42.21%; graded groups had 17.95%.
These comparisons can be shaped by card mix, price, marketplace coverage, condition, grade-company mix, and metadata quality.
Examples
Real observations can still be far apart.
These are a few high-observation examples from the study's raw English cohorts. They are illustrations, not a representative sample and not a claim about what any card is worth.
CardObserved recordsMedianObserved range
Pidgey · Obsidian Flames #208175$11.99$1.50–$240.00
Charizard ex · Obsidian Flames #223149$101.59$9.49–$200.00
Milotic ex · Surging Sparks #236120$30.00$15.50–$258.24
Dark Dragonite · Team Rocket #5/82116$110.99$8.50–$549.99
Ranges include the observed tails and should not be read as proof that every copy was perfectly comparable. Card labels were checked separately against public references; the aggregate study used production identity fields.
For collectors
More context beats one isolated sale.
1Check more than one relevant Sold Comp when you can.
2Confirm the exact card, set, number, variant, language, and finish.
3Compare condition and grade like with like.
4Pay attention to recency, sample size, and the spread—not just the center number.
For journalists and data readers
How we measured it
The main story is simple. The measurement underneath it is deliberately explicit.
Open the detailed methodology
Dataset and terminology
The production table contained 3,407,367 raw database rows at the audit point. That is a row count, not a count of sales. The 90-day date-bounded query returned 401,804 rows before the final analysis filters. The final extract contained 389,602 eligible completed-sale listing observations.
We use “observations” because provider listing identity does not independently prove that every record is a unique payment transaction.
Eligibility and grouping
Records had to be completed, dated within the selected window, positive-priced, USD, pricing-quality eligible, and attached to a non-empty normalized card identity. Unmatched NO_MATCH records were excluded. English and Japanese were kept separate. Raw and graded observations were kept separate; graded groups preserved grading company and exact grade identity, and required those fields to be present. Missing condition was grouped as unknown rather than silently treated as Near Mint or another known condition.
The primary result uses 23,012 groups of the same exact card/variant with at least three observations, representing 364,773 observations. Rows were de-duplicated on provider listing identity when available, then marketplace listing identity, then database row id. No duplicates were removed from the final extract under that rule.
Statistics
For each group, we calculated the median, interquartile range (the middle-half spread), median absolute deviation, coefficient of variation, high/low ratio, trimmed-mean comparison, and distance of each observation from the group median. Tukey-fence sensitivity testing checked whether outliers changed the central result. We repeated the analysis at 30, 60, and 90 days and at minimum thresholds of 3, 5, 10, and 20 observations.
Read this before drawing conclusions
What this study does not prove
- It does not prove that every record is a unique payment transaction.
- It does not tell you the “true value” of any individual card.
- It does not prove that condition, language, grading, marketplace, or timing caused a particular price difference.
- It does not cover every marketplace equally, and provider selection can shape the sample.
- Condition, edition, finish, shipping, taxes, timestamps, and grading details may not be perfectly comparable in every record.
- One observation still cannot establish a stable range; the study's spread claims require repeated observations.
About PokeSplain Research
Market evidence, explained clearly.
PokeSplain Research uses market evidence to investigate practical questions for Pokémon collectors. We publish the methodology and limitations because a useful number is only as trustworthy as the context around it.
Explore PokeSplainCiting this research
PokeSplain Research. “One Sold Comp Isn't the Market.” September 20, 2026.
https://pokesplain.com/research/one-soldcomp-isnt-the-market/
Media / research inquiries: daniel@pokesplain.com
Methodology is published on this page for review and reuse.