A $3,499.99 Pikachu Priced Off a Played Copy, and 13 More Like It

On September 19 we published The Frozen Catalog, a count of the cards whose recorded price had not changed in three weeks. Six four-figure examples went into the piece. One of them, Pikachu Star from Holon Phantoms, was printed at $1,899.99 because that is the number our daily price file had returned every single day. Five days after we published it the file moved that card to $900. Our own second reading, the one that carries a condition label, had said $900 since September 22 and $380 before that. We had the evidence to doubt the figure and did not look at it. This article is the check we should have run, run on everything, and the first thing it found is worse than a stale number.

Every price on this site starts as one of two recordings. The first is the daily price file, which gives one figure per marketplace product with no condition attached and no date of observation. The provider builds that figure by walking down a card's condition ladder whenever the top of the ladder has nothing recent in it, so the same column can describe a mint copy one week and a played copy the next. That behaviour is the reason the second recording exists. Our Near Mint series holds one Near Mint reading per product per day, Near Mint by construction, stamped with the day the provider observed it rather than the day we happened to write it down.

So this is not a labelled number against a worse number. It is a labelled number against an unlabelled one, and either can be wrong. Three cards later in this piece are cases where the labelled reading is the broken one, including a $1,252.62 Umbreon from Aquapolis whose Near Mint reading says $12.89, which sits below the $624.50 recorded for a damaged copy of the same card and therefore cannot be describing it. A disagreement between the two is a reason to look, not a verdict about which is right. Which Market Price is the piece on what each figure is made of, and How to Tell a Pokemon Card Price Is Stale on reading the date on one.

Every figure below is measured at one day, September 23, 2026, with a six day look back to September 18 and a five day look forward to September 28. 52,378 products carry a price on September 23. 52,072 of them carry one on all six days of the look back, and 35,588 of those 52,072, which is 68.3%, returned the identical price to the cent on every one of the six. That share is the reason the restriction exists rather than the finding: most of this catalogue does not move in six days, so an unchanged price on its own tells you almost nothing, and a useful check has to beat a base rate rather than a coin toss.

Three filters cut the 35,588 down. A price of two dollars or more leaves 15,131, because under two dollars a one cent difference is already a 3% gap and every comparison returns noise. Products with a single printing leave 5,838, because one product id with several printings carries several prices at once and a difference between two readings of it can be a real difference between two different cards. And a Near Mint reading actually observed on September 23 leaves 3,429. That last filter is the largest hole in the study: 58.7% of the frozen single printing cards above two dollars have one and 41.3% do not, and nothing here describes the ones that do not. No figure in this article can be restated as a share of the catalogue, and none is.

45 of the 35,588 frozen products held a four-figure price on all six days. Compare each of those 45 numbers against the condition quotes we separately hold for the same product, and 24 of them match the Near Mint quote, which is what should happen. 10 match the Lightly Played quote instead. 4 match the Damaged quote. 5 match no rung of their own ladder. 2 have no recorded ladder at all. That is 14 of 45 four-figure prices sitting on a played copy's number, and it is not a coincidence to the cent on a thousand dollar figure: it is the documented behaviour of the source, which falls down the ladder when the top of it is empty. A separate count on the same 45 is worth holding beside it. 19 of them have no daily Near Mint reading on any day we hold, so for those nineteen there is no second number to check the first one against, and that is unknown rather than agreement. The ladder quote and the daily reading are two different records here: the ladder is a current snapshot carrying the date each rung was last observed, some of them months old, and the daily reading is the series.

Read the last two columns together, because they are the uncomfortable part. Seven of these fourteen show the four-figure number on our card page right now with no condition ladder beside it, so a reader cannot see that it is a played copy's figure. The reason is a rule we wrote on purpose: the page states the headline as its Near Mint row rather than measuring Near Mint twice, and it drops any played quote priced at or above that headline, because a damaged copy cannot really be dearer than a mint one. On Latias ☆ the three quotes we hold are Damaged at $1,650.00, Heavily Played at $798.99 and Moderately Played at $2,300.00, with no Near Mint quote at all, so the whole ladder fails the coherence test and none of it renders. The page shows $1,650.00 under a Near Mint tab and shows nothing that would let you argue with it. That is our defect, not the provider's, and the table above is here because the check it describes is one a reader cannot currently run on this site.

Now the 3,429. 2,601 of them, 75.9%, match their September 23 Near Mint reading to the cent. 490 of those 2,601, which is 18.8%, were rewritten within five days. That is the number to beat: in this group, about 19 of every 100 frozen prices change inside a working week for no reason you can see in advance.

828 disagree with their Near Mint reading. Take the 335 that disagree by 2% or more, so the difference is money rather than rounding, and 269 of them, 80.3%, were rewritten within five days. 232 of those 269 revisions landed on the September 23 Near Mint number exactly, to the cent. The median revision arrived on the next recorded day: 223 of the 335 had moved by September 24, against 22 of the 2,601 matching cards. This is not a forecast of the market. It is a measurement of a queue. The labelled reading is the newer of our two numbers, and in four cases out of five the unlabelled one catches up to it inside a working week.

The whole table is printed rather than the 2% cut alone, so you can see the result does not depend on where the line is drawn. It also shows the thing an earlier draft of this study got backwards. The revision rate does not keep climbing with the gap. It peaks in the middle and falls away: 100.0% of the 2% to 5% group was rewritten inside five days against 41.5% of the group sitting 25% or more away. A wider gap is not a longer queue. Past about 10% it is usually a different problem, which is the next section.

The direction of the eventual change is called by the sign of the gap, and this is the part a seller can use. 92 of the 335 sit ABOVE their Near Mint reading. 68 were rewritten, and 64 of those 68 went down. 243 sit BELOW it. 201 were rewritten, and 0 of those 201 went down. The count difference between 92 and 243 is the market rising through this window and does not mean anything on its own. The direction does: among the 490 revisions in the matching group, 163 went down and 327 went up, so a revision picked at random from this file falls about a third of the time, and 64 of 68 is nothing like a third.

It holds at every price level we can test it at. Above ten dollars the matching group revises 19.2% of the time against 76.9% for the group sitting 2% or more away. Above twenty five dollars it is 18.7% against 74.3%. Above a hundred dollars, where the sample is 382 matching cards and 57 gapped ones, it is 21.2% against 63.2%. The effect narrows as the cards get dearer and does not disappear.

One marketplace product id can cover several printings of the same card, and each printing carries its own price. Our per-printing file records them separately, so this is checkable rather than arguable. Take every product with a Near Mint reading of twenty dollars or more on September 23 and a printing count we can read, which is 3,629 of them, and sort by how far the two numbers sit apart. Among the ones that agree to the cent, 11.7% have more than one printing. Among those 25% or more apart, 90.6% do.

On 30 of those 326 we can name which printing each number belongs to, because both figures equal a printing's own recorded price for September 23 and the printings are different ones. Those 30 are the mechanism made visible. The pairs they fall into are the four you would expect: 10 are a holo quote against the plain copy's, 9 are a reverse holo quote against the holo's, 6 are 1st Edition against Unlimited and 5 are 1st Edition Holofoil against Unlimited Holofoil. Twenty five of the 30 have a catalogue row and are named below. The other five are Base Set Shadowless products our card catalogue holds no row for, which is its own gap and not a price problem.

Nidoking from FireRed & LeafGreen is the clearest one to hold in your head. For September 23 we record $76.29 for the holo, $58.99 for the reverse holo and $28.78 for the plain copy. All three are that card's price. None of them is stale. If you saw $76.29 on one page and $28.78 on another and read the difference as a lagging number, you would buy the cheap one and receive a different card. Our card page does let you pick the printing, and picking it is the whole action here.

Across the whole file rather than the frozen part of it, 17,624 products carried both a daily figure and a Near Mint reading on September 23. 7,244 of them, 41.1%, agreed to the cent. Of the 10,380 that did not, 8,618, which is 83.0%, matched an EARLIER Near Mint reading for the same product, at a median age of three days. That is the queue, measured directly. 158 more matched a recorded Lightly Played, Moderately Played, Heavily Played or Damaged quote, which is the ladder fall back caught in the act. And 1,604, which is 15.5%, matched nothing we hold at any date and in any condition. We do not know what those are. They are printed as unexplained rather than folded into the nearest bucket, because a residual that gets quietly absorbed stays absorbed forever.

The Frozen Catalog named six four-figure prices that had not moved. Five of the six are still the identical number today. The sixth is the one this piece opened with. Two of the six have no Near Mint reading on any day we hold, so calling them steady was never a statement we had the evidence for. And on two of the three that agree, the agreement arrived because the Near Mint reading moved up to meet the daily figure rather than the other way round, which is the same mechanism pointing the other way.

Pikachu Star deserves the full sequence, because the tidy version of it would be a lie. On September 19, the day we published $1,899.99, our own Near Mint reading for that product said $380.00. It said $900.00 from September 22. The daily file moved to $900.00 on September 24 and has stayed there. So the figure we published was 111% above the Near Mint reading recorded for September 23, the check would have flagged it, and the labelled reading was not itself correct either: $380.00 was never the number anything settled on. Both of our readings were wrong on the day we published, in opposite directions. The fall from $1,899.99 to $900.00 is also already recorded in The Ask Is Lying to You as a movement in the list of four-figure asks, which is a different question about the same two numbers.

First, before you value a four-figure card off a number that has not moved, look at the played rungs on the same page. If the Near Mint figure is the same to the cent as the Lightly Played or Damaged figure, you are looking at a played copy's number wearing a mint label, and 14 of the 45 four-figure frozen prices in our own records are in that position. If no played rung is shown at all, treat the figure as unverified rather than as confirmed, because on seven of those fourteen the ladder is being withheld precisely because it contradicts the headline.

Second, if you are setting an ask, the direction is the useful part. A frozen price sitting above the labelled Near Mint reading is a markdown waiting to land: 64 of the 68 that were rewritten went down, against a base rate of about a third, and the median one moved on the next recorded day. A frozen price sitting below it will be revised up, and 0 of the 201 in that group that were rewritten went down. Pricing off the higher of two numbers when the lower one is the labelled and newer one is how you end up cutting twice.

Third, when two numbers for one card are 25% or more apart, stop reading the difference as a date. 326 of the 360 cards that far apart have more than one printing, against 191 of 1,632 where the numbers agree, and on 30 of them both figures are a real printing's real price. Check which printing you are being quoted before you decide one of the numbers is stale. That is also the group where the check is weakest as a predictor: only 41.5% of the widest gaps were rewritten within five days, so a wide gap is a question and not a signal.

None of this is tradeable and nothing here is a sale. The daily figure is neither a bid nor an ask you can hit, the Near Mint reading is a marketplace figure and not a transaction, and a change in either is a change in a published number rather than money anybody paid. The Ask Is Lying to You is the study of the distance between a listing and a sale, and nothing in this piece closes it.

It does not say how often a Near Mint reading is the wrong one. Three cards in this article show that it can be, and we have not counted the rate, so the check tells you to look rather than which number to believe. It does not describe the 41.3% of frozen cards above two dollars with no Near Mint reading at all, and absence there is unknown rather than agreement. It measures one day with a six day look back and a five day look forward, in a window where the market rose, so the direction results are sound and any count of how many cards went up rather than down is partly the window. It says nothing about cards under two dollars, nothing about sealed product, and nothing about graded copies, whose prices come from a separate record entirely. And it is not a claim that a frozen price is a wrong price: 2,601 of the 3,429 cards here agree with their labelled reading exactly, which is most of them.

SOURCES. price_history, our daily accretion of the TCGplayer CSV file, one figure per marketplace product per day with no condition column and no observation date; nm_history, one Near Mint reading per product per day stamped with the day the provider observed it; condition_quotes, a current snapshot of the per-condition ladder with the date each rung was last observed; and tcgcsv_prices, per-printing market figures per product per day, used only to count printings and to name which printing a figure belongs to. Card identity is resolved by collapsing catalogue rows that share one product id to a single row, preferring the row whose set name matches the price file's own spelling and otherwise the lower card id, because 1,215 product ids in this catalogue are claimed by two rows each. WINDOW. Every figure is measured at day 20719, September 23, 2026, with a six day look back to September 18 (day 20714) and a five day look forward to September 28 (day 20724). Nothing is measured past September 29. That window was chosen because it is the only six day stretch in our records that is clean on both sides: on September 1, September 3 and September 6 the file returned the identical price for 100% of products against the previous day, and on September 8 and September 25 for 99.78% and 99.96%, so those are files that did not land and a frozen count spanning them is inflated by construction. None of September 18 to 23 is such a day (84.2% to 87.2% identical against the previous day, against an ordinary rate near 85%), and all six carry more than 52,100 products. FROZEN means the identical price to the cent on all six of those days. REWRITTEN means a recorded price differing by at least one cent from the frozen figure on any recorded day from September 24 to September 28; a product with no row on a given day is not treated as unchanged on it, and September 24 holds 44,106 products rather than the full file, so the next-day figures are floors. FILTERS, in order: priced on all six days (52,072), identical on all six (35,588), two dollars or more (15,131), one printing (5,838), a Near Mint reading observed on September 23 (3,429). PRINTING COUNTS COME FROM TWO SOURCES AND THE STUDY SAYS WHICH IT USED WHERE. The single-printing filter that produces the 5,838 and the 3,429 uses the count recorded in the condition snapshot, which covers 21,478 products. The printing counts and the printing names in the third section use the per-printing file, which covers 31,502 products on September 23. On the 16,600 products both sources cover they agree on whether a product has one printing in 16,510 cases, 99.5%. Sets released inside the window are excluded: 30th Celebration released on September 16, and every set name on the price file containing that phrase was excluded, which is four spellings. Two of the four are the ones our release metadata dates; the other two are a Japanese expansion and a premium deck carrying the same phrase, excluded on the name because we hold no release date for them. No product from any of the four held one price across September 18 to 23, so the exclusion removes nothing here rather than being untested. Delta Reign has not released. FOUR-FIGURE SECTION. The 45 are every frozen product at $1,000 or more with no printing filter applied. A price 'matches' a rung when the two agree to the cent, tested against Near Mint first and then Lightly Played, Moderately Played, Heavily Played and Damaged, so a card counted against a played rung is one whose Near Mint quote differs. The ladder observation dates are printed because several are months old. 'Our page today' is the figure the live card market endpoint returned on September 29, 2026, read one card at a time, and 'no ladder shown' means the endpoint returned no per-condition prices. That happens for two reasons, both of them the site's own coherence rule: a ladder whose played rungs are priced at or above its Near Mint rung is dropped whole, and so is a ladder with no Near Mint rung at all. Four of the seven cases in the table above are the second kind. LIMITS ALREADY STATED IN THE BODY and repeated here: the Near Mint reading can itself be wrong and this study does not measure how often; the 41.3% of frozen cards with no Near Mint reading are unknown rather than in agreement; the 10% to 25% and 25%-or-more groups are 67 and 65 cards; the market rose across the window, which affects any up-versus-down count and not the sign results; and 15.5% of all disagreements on September 23 match nothing we hold and are printed as unexplained. A set name in a table with no link has no set page on this site, which is true of every ex-era set here except Power Keepers. Nothing in this article is financial advice.