Koffing grins on its Japanese common with a toothy, cheerful face, a pale purple gas ball drifting against soft blue-white smoke swirls on a bright green card. It arrived in Clash of the Blue Sky in July 2004, months ahead of the English Deoxys set that carries the same Koffing, so early Japanese copies predate it. As a bulk common it is binder filler, though a charming one for Gen 1 poison fans. The modelled guide sits at $9.95 with no copies traded in 30 days; no PSA population is on file.

as of Oct 24, 2026 · UTC

history since Jul 2026

Koffing — 003 · 1DMkt$9.950.0%guide price · 7-day avgitem price

Vol — sold·0 prints·1 obssingle obs — no range

Live Listings

Pricevs MktQtyCond

TCGplayer0

eBay0

Fair value

Insufficient data

No fair value is published for this card — no confirmed print or completed auction in 30 days — an estimate built from live asks alone would be a restatement of the book.

Market safety

No book

Nothing is listed for sale, so there is no book for one buyer to move.

Flow

Thin tape

Insufficient data

Only 1 unit sold in 30 days — under the 8-unit floor for reading a pace.

Method

Daily units are the venue's own sold-quantity history, expanded off each bucket's real span, plus confirmed prints from the other side of the market. Prints come from the live sales tape at delivered prices. The floor is the cheapest delivered ask (item + required shipping) across venues — the same one FAIR VALUE prices against.

Bottom line

No call

21/100 across 2 of 11 inputs. Against: liquidity reads 0/100. That is under half the model, so no call is made on it.

Liquidity0

Supply

Nothing listed

0 units listed — 0.0d of stock at the 30-day pace of 0.0/day.

Listed0 unitsSold · 30d · 0.0/day1 units

Print scarcity

Set not covered

A set needs at least three cards carrying a price, a graded population and a pull rate at once before its print scale means anything. this set does not reach that, so there is nothing to rank this card against.

LightModerateHeavy

Method

Graded population is print run × how often anyone bothers grading it, and the second factor moves with VALUE — so population is divided by price^0.606, the elasticity fitted across 3,826 cards that carry a population, a price and a pull rate at once. That normalised count is turned into a SET pack scale using the card’s pull rate (every card in a set is an independent read on how many packs exist, pooled by median), then brought back down to this card by its own pull rate. The band is a tertile of the result across every card that clears the gates.

Nobody has ever published a Pokémon print run, and this does not recover one — it is an ORDER, not a count. It reads surviving graded supply across PSA, CGC and ACE, so ungraded and sealed stock are invisible to it. What it is checked against: the model puts Black & White at the bottom of the catalog and modern Scarlet & Violet at the top, which is where the smallest and largest print runs actually are.

Why It's Moving

What changedWhat changedThe observable that moved, stated before any reason for it — the fact the rest of this panel has to account for.: Modelled guide price moved +0.0% over the trailing 7 days and +0.0% over 30 days, now at $9.95.
Primary explanationPrimary explanationThe reading the evidence best supports. It is an explanation, never a forecast, and it is only as good as the Evidence row two lines down.: Primary read: Liquidity scored 0/100 (composite weight 38%). — the largest single contributor to the current ILLIQUID_UNKNOWN regime read.
AlternativeAlternativeA competing explanation the SAME evidence would also support. It is printed even when the primary looks strong: an explanation with no stated alternative is a story, not an analysis.: Alternative explanation: Data quality scored 17/100 (composite weight 24%).. With 0 distinct sellers/observations in the trailing window, either read is defensible — this is not a high-conviction single-cause move.
EvidenceEvidenceThe specific observations the explanation rests on. If this row is thin, treat the explanation as thin — it is the only thing holding it up.: Evidence: item-basis daily bars, absorption read ("Insufficient data: 0 inventory observation(s) in the trailing 30d window (need >= 4).").
ConfidenceConfidenceHow much weight to put on the primary explanation, given how much of the evidence actually resolved. It qualifies the row above; it never adds to it.: Confidence: 90% (floor-confidence 13/100), based on 237 days of observations.
What would disprove itWhat would disprove itThe observation that would break the primary explanation. It is what makes the reading falsifiable rather than unfalsifiable — and it is the thing to watch for next.: This read would be wrong if the price reverses over the next 7 days.

Highest Volume

The 12 most-traded cards by dollar volume over the last 30 days
#	Card	SetSetThe set this printing belongs to. Sets differ in print run and age, which is most of why two cards at the same price behave differently.	Trend 30DTrend 30DThe last 30 days of price, drawn from a zero-independent scale. The shape is the reading — the level is the Price column beside it.	30D30DPrice change over the last 30 calendar days, signed. A dash when the series has no price close enough to 30 days ago to measure from.	PricePriceThe card’s current market price as the wall quotes it (an item price, before shipping).

Provenance

EvidenceEvidenceWhat kind of fact this row is. A completed sale, a live ask and a calculated guide price are three different weights of evidence about the same number, and this column is the difference between them.	ObservedObservedThe day the fact was read — not the day the underlying event happened, and not today. An old date is an old reading.	AgeAgeHow long ago this row was read, as of the moment the page was computed.	ConfidenceConfidenceHow much weight this row carries where it feeds a model, from the resolver that matched it. Lower means a looser match or a thinner source, and a low-confidence row is reweighted rather than dropped.	ModelledModelledYes = a CALCULATED value (a guide price, an estimate); No = something actually observed (a sale, a listing, a census count). A modelled row is evidence about a model, not about the market.
CatalogCard identity: set, number, rarity, finish, release date. It dates from the printing itself, so an old reading here is expected and is not staleness.	Jul 1, 2004	22.3y	98%	No
Guide priceA calculated market guide — a model of where the product is trading. Never a confirmed sale, and always marked Modelled.	Oct 24, 2026	16h	90%	Yes

Raw · Japanese · Normal · 1st Edition · Near Mintpq-quant-2026.10.0

Highest Volume

The 12 most-traded cards by dollar volume over the last 30 days
#	Card	SetSetThe set this printing belongs to. Sets differ in print run and age, which is most of why two cards at the same price behave differently.	Trend 30DTrend 30DThe last 30 days of price, drawn from a zero-independent scale. The shape is the reading — the level is the Price column beside it.	30D30DPrice change over the last 30 calendar days, signed. A dash when the series has no price close enough to 30 days ago to measure from.	PricePriceThe card’s current market price as the wall quotes it (an item price, before shipping).

Relative strength48

2 of 11 inputs · 16% weight · -35% confidence

Not reading yet: value vs fair, price trend, money flow, book pressure, grading upside, manipulation risk, seller concentration, reprint risk, scarcity.

Liquidity9%0Relative strength7%48Value vs fair17%no readingPrice trend13%no readingMoney flow11%no readingBook pressure11%no readingManipulation risk9%no readingScarcity7%no readingGrading upside6%no readingSeller concentration5%no readingReprint risk5%no reading

The bottom line is a weighted mean of the signals that reported, renormalised so their weights sum to 1. Nothing defaults to a neutral 50: an input with no evidence is reweighted away and its absence is charged as a confidence penalty. Risk rows (manipulation, concentration, reprint) enter inverted. Data quality carries no weight at all — it decides whether the verdict may be spoken, because knowing a lot about a card is not a reason to buy it. Competitive meta sits OUTSIDE the weighted mean as a capped bonus added after it: a card nobody plays is never marked down for that, and a card the meta is built on can be marked up for it.

30d vol30d volume

Ranked by dollars traded over 30 days · refreshed hourly

The whole wall →

$8.3k

-33.6%

$5.1k

$4.7k

$4.2k

$3.9k

$3.8k

$3.6k