Before You Buy Sealed, Count How Many Times the Price Changed

Every sealed gainers list in this hobby, ours included, ranks products by how much their price has changed in thirty days. Nobody asks the question underneath it, which is how many times the price got changed at all. We hold daily unopened prices for 1,516 sealed products over the month to 2026-09-29, and once you count the edits the list splits in two. 217 of those products were never re-priced once in the whole window, so their thirty day change is exactly zero by construction. Another 425 were re-priced once or twice, and 294 of those read up. Put together, 294 of the 840 products showing a gain got there on two edits or fewer, against 63 of the 312 showing a loss. Stale sealed prices drift upward, and the gainers list is where they collect.

The number we track is the provider's unopened price for a sealed product, which is the TCGplayer Market Price for that SKU. It is not an asking price, it is not a sale we watched happen, and it is not a graded figure. A price change in this piece means two consecutive recorded readings that differ at all, one cent included. That is a fact about the tag rather than about the market, and the distinction is the whole article: a product whose tag was rewritten twenty two times is telling you something, and a product whose tag was rewritten once early in the month and left alone since is telling you what somebody thought early in the month.

So this is not a claim that a still price means nothing traded. That argument belongs to The Frozen Tape, which measured 37.5% of week-apart card prices as identical to the cent, and to 14,466 Cards With a Price and No Sale, which found how many prices have no sale behind them at all. Both ran on singles and on their own populations. What is new here is the sealed shelf and the counting.

The medians do not fall in a straight line and we are not going to pretend they do. The steepest median gain in the table is the three-to-five group at +2.66%, not the one-to-two group. What does move in one direction all the way down the table is the balance: 69% up against 15% down where the tag was rewritten once or twice, and 46% up against 48% down where it was rewritten fifteen times or more. A price that gets tested constantly lands somewhere near flat and splits evenly on direction. Where the tag was rewritten once or twice it reads up 4.7 times as often as it reads down.

On 2026-09-15 we published Will the Pokémon Market Crash?, and one of its eight readings was that sealed had thawed: 833 of 2,007 products up more than 2% over thirty days against 258 down, after the sealed freeze of August counted six gainers in 651 products. The arithmetic was right and we are not disputing it. Rebuild that reading the same way it was built, on all 2,011 priced sealed products with a recorded change and no exclusions, and today it comes to 794 up more than 2% against 243 down. 266 of those 794, 33.5%, got there on two price changes or fewer. On this article's own stricter universe of 1,516 the same count is 614 up more than 2%, 203 of them on two changes or fewer, and 146 products reading up on exactly one edit. The thaw is real in the sense that the numbers on the tags are higher. It is not the same thing as a market that has started trading again, and the reading should have carried that.

Split the buy-decision band, products priced $100 to $550, into booster boxes and Elite Trainer Boxes on one side and everything else on the other. The two sides are not the same kind of measurement. Boxes and Elite Trainer Boxes get a new price a median of fourteen times in twenty nine readings, and only 5 of those 108 products were rewritten twice or fewer. Everything else in the band gets a new price a median of twice, and 225 of the 367 were rewritten twice or fewer. Then look at what each side says.

The side of the band that is priced continuously reads down, and 57 of its 62 decliners were rewritten six times or more, so those are not stale tags drifting either. On the other side, 97 of the 190 products reading up were rewritten twice or fewer. The same split holds on the full 1,516 without the price band, which matters because a band is an exclusion and a finding that only exists inside one is a finding about the band: boxes and Elite Trainer Boxes, 155 products, median -0.22% on a median of 9 changes, 52 up against 74 down; everything else, 1,361 products, median +1.24% on a median of 3 changes, with 207 of them never rewritten at all. The band sharpens the contrast rather than creating it.

The ordering is the point. Sort by liveliness rather than by return and the boundary lands in one place. The four formats whose prices are rewritten most often all read at or below +0.67%, and the nine formats below them all read at or above +0.88%, on a median of five edits a month or fewer. In those nine, the share of products never rewritten once runs from 3% of mini tins to 22% of the products our classifier does not give a format name to. The Convenience Tax published a six-format sealed split earlier in September on a different universe and at different figures, including Elite Trainer Boxes at -2.5% on 51 products where this run reads 0.00% on 68. Those are two different populations over two different windows, and the number to trust for the question asked here is the one computed from the table we now keep.

The first row is the largest single-edit gain among products priced $100 to $550. The tag went from $210.00 to $405.96 between two readings in mid September and has printed $405.96 on every one of the twelve readings since, which is a 93.3% thirty day gain resting on one edit. It is also a product whose recorded window carries fourteen readings rather than twenty nine, so the gaps are part of the picture and we are showing them rather than smoothing them. Drop the price band and two cheaper rows go further on the same single edit: a Scarlet & Violet Premium Checklane Blister with Machamp on it at +172.7% from $11.00 to $30.00 on six readings, and a 2017 World Championship deck at +113.9% from $25.00 to $53.48.

The two Paldean Fates tins are the reason this piece stops short of dismissing the gainers list. Both climbed on six and seven separate steps, and a buyer who skipped them because the list is full of stale tags would have skipped something real. 267 of the 840 gainers look like that. The test sorts them from the furniture; it does not condemn the whole list.

The three products this study turns on

Stormfront Booster PackOne edit, then twenty seven readings at the same number+16.4% / 1 change

151 Elite Trainer BoxTwenty two edits, walking down-17.0% / 22 changes

Paldean Fates Tin (Iron Treads ex)Six edits, walking up, the honest exception+43.3% / 6 changes

It does not tell you a sealed product is cheap or dear. Counting edits to a price tag says how live the quote is and nothing about whether the level is right. A box rewritten daily can be badly mispriced every single day.

It does not say sell. The boxes and Elite Trainer Boxes reading -0.93% in the buy band are one month of readings on 108 products, and our own guard against reading a thirty day move as a trend blocks that inference in both directions. It says wait and check the chart, not liquidate.

It does not prove that a still price means nothing sold. A run of identical readings means TCGplayer did not restate the market price for that product. Whether anything changed hands behind it is a question this column cannot see, and the two articles cited earlier answer a nearby version of it on singles rather than on sealed.

It cannot be compared with the August sealed freeze on equal terms. For the whole stretch 2026-07-30 to 2026-08-26 our record holds 159 readings across 93 products, against 1,251 products carrying a reading on 2026-08-31 alone, so there is no August panel to compare a September panel against. The six-gainers figure published then stays exactly as published, unverifiable, and any sentence pairing August with today inherits one endpoint we cannot check. That is the reason the window starts where it starts.

It is not a statement about Japanese sealed, about graded product, or about anything outside the TCGplayer sealed catalog we sync. And it is not a second opinion on price: the cross-check below reads the same TCGplayer market price through a different pipe, which tests our freshness and nothing else.

UNIVERSE. Every sealed_products row with price_usd above zero and a non-null change_30d, 2,011 rows on 2026-09-29. Excluded: 421 multi-unit lots by LOT_RE from src/server/set-products.ts, because a case of six boxes is not one product's price; 33 products whose recorded readings span under seven days; and 41 products belonging to ME: 30th Celebration, ME: 30th Celebration Classic Collection or ME06: Delta Reign, whose release dates fall on or after the window's first day or in the future, because a preorder tag settling to a retail price is not a market move. That leaves 1,516. No price floor and no top-N cut. Format grouping is classifyProduct() from the same file; groups under twenty products are printed as counts and not treated as findings. WINDOW. Readings dated 2026-08-31 to 2026-09-29 inclusive, median 29 readings over 28 days per product. The window floor is stated rather than discovered: the table holds 189 scattered readings from earlier in August across 113 products, left by the sync's first partial passes, and catalog-wide daily coverage starts 2026-08-31. Windows neither all start nor all end on the same day, because the sync runs in paged passes and the provider hands each page its own window: 922 series start 2026-08-31 and 574 start 2026-09-01, with 20 starting later; 853 end 2026-09-28, 537 end 2026-09-29, and 126 end on one of 21 earlier days, the earliest 2026-09-07. The first window day is itself partial, 1,251 products against 1,975 on 2026-09-01. SOURCE AND WHY IT IS NEW. The dated prices come from sealed_price_history, a table created for this study (src/server/db/sealed-history.ts, declared in schema.ts, written by the sync-sealed cron on every pass and backfilled once by scripts/sealed-history-backfill.mts: 55,489 points across 2,127 products, span 2026-08-01 to 2026-09-29). Before it existed these points lived only in api_cache rows keyed sealedhist:<tcgPlayerId> under a 35-day TTL that each pass replaced in full, so no figure computed from them survived a day. The write merges on (id, day) and has no delete path, so a refresh may add a day or correct a day and can never shorten the record. DEFINITIONS. A price change is two consecutive recorded readings that differ at all, one cent included; there is no threshold. Readings the last price has stood is the count of consecutive readings, ending at the most recent, carrying the final price. Medians take the upper-middle value on an even sample, the convention the-convenience-tax also states. COLUMN MEANING. price is the provider's unopened price, which is the TCGplayer Market Price for the SKU: not an ask, not a sale, not a graded figure. VERIFICATION. change_30d as the catalog stores it reproduces from our own recorded points on 1,515 of the 1,516 products. The single exception is the 2024 World Championship Deck: Fernando Cifuentes (Crushing Thorn), whose recorded readings stop on 2026-09-13 while its catalog price kept updating, and it is left in the universe at its recorded values rather than dropped. CROSS-CHECK. Against tcgcsv_prices on day 20725, sub_type Normal: 1,363 of the 1,516 products match by product id and 1,332 of those agree within 2%. Both sides are the TCGplayer market price arriving by different routes, so this measures whether our sync is current and says nothing about whether the price is correct. SCRIPT. scripts/sealed-requote-figures.mts prints every figure in this article in one run against production.