Special Delivery Charizard Value: a 14% Gem Rate

Special Delivery Charizard SWSH075/307 asks $208.31 on our daily tape and its PSA 10 sells for a median $1,311.93 across 98 verified sales, both read September 23 2026. The number that makes the card interesting is neither of those. PSA has recorded 32,567 submissions against this card and returned 4,546 tens, a lifetime gem rate of 13.96%, and over the 32 recorded days to September 22 2026 the line added 474 slabs of which twelve were tens. That is a marginal gem rate of 2.53%, and it is the 126th lowest of the 1,330 cards in our cohort where the measurement is possible.

This is one of 143 products in our 5,879-card cohort carrying a sold median at all seven recorded grades. The PSA 9 is the row that matters most on a card that gems one time in seven: 251 verified PSA 9 sales stand behind $249.75, more than twice the PSA 10 sample.

The PSA 10 fetches 5.25 times the PSA 9 and 2.29 times the CGC 10. Both sit near the catalog-wide readings in is a PSA 9 half a PSA 10 and is a CGC 10 worth a PSA 10. The CGC 9 sells for slightly less than the ungraded card asks. The same thing one grader over, a PSA 9 sold median below the raw ask, is a real feature of deep modern registries and is measured across the whole cohort in Alakazam ex from 151.

A gem rate means nothing without a population behind it, because a thin registry can read any number at all. Restrict to products whose registry line holds 10,000 slabs or more, is proved unshared and supports a trustworthy rate, and 522 remain, 298 of them English. Sorted ascending by gem rate, Special Delivery Charizard is 51st of the 522 and 45th of the 298 English. That puts it inside the hardest 10% of the 522 and the hardest 16% of the English 298. It is not the hardest, and the ten English cards at the very bottom of that ranking are printed below.

Every one of those ten is a Wizards of the Coast era card. That is the era effect measured at scale in hard to gem and the gem rate curve. What makes SWSH075 unusual is being a 2021 card in that neighbourhood at all. Whether the same difficulty holds across a whole vintage era is EX era PSA 10 gem rates; the Neo sets in the table above are which Neo set holds its value; and the Wizards promo family belongs to are Black Star Promos worth anything.

We hold five registry readings for this card, from August 21 to September 22 2026, a span of 32 recorded days. The line went from 32,093 submissions to 32,567, a gain of 474 slabs, about 14.81 a day and 1.48% of the standing population. Twelve of those 474 came back tens. The marginal rate of 2.53% is a fifth of the lifetime rate, and the PSA 9 count rose by 164 over the same window, which is where the submissions are actually landing.

The gap between 56.5% and 53.3% across 1,330 cards is the general drift. The gap between 13.96% and 2.53% on this one card is eleven and a half points on 474 slabs, and it is what a grader tightening on a specific card looks like from the outside. It is also the shape of the Van Gogh Pikachu case in PSA changed its mind about the felt hat, at a smaller scale and on a much lower base.

Against the wider catalog the registry is the 86th deepest of 20,805 products with an unshared reading. Read alongside every Charizard's PSA 10 premium, ranked, which ranked 82 Charizard printings on September 22 2026, this card's premium is unremarkable for the species: recomputing that ranking on our own day and rule gives 83 printings, and SWSH075 is 63rd of them. A hard card with a modest premium is the combination the hobby least expects, and the reason is the raw ask: at $208.31 the ungraded card already carries most of the price.

Unlike the other four cards in this series, Special Delivery Charizard has exactly one product in our catalog under that name, in either language. There is no second printing to mistake it for. The family it belongs to is three cards, all SWSH Black Star Promos and all sold through the same channel, and on September 23 2026 their asks run from $208.31 to $558.05.

The Charizard is the cheapest of the three raw, has the lowest PSA 10 median of the three, and has more than twice the verified sold tape of either. It is also the only one of the three carrying a registry figure we will print. We hold three readings each for the Pikachu and the Bidoof and every one of them fails our line-continuity rule: both series sat at a total of 1 and of 2 through September 14 2026 and then read 14,591 and 5,013 on September 21, which is a registry line being re-pointed rather than fourteen thousand slabs graded in a week. One reading on a new line is a value we watched replace another one, so both are withheld rather than printed as absent, and neither card can carry a gem rate or enter this study's cohort. The promo family as a whole has a working set screen at [SWSH Black Star Promos](/research/set-screen/swsh%20black%20star%20promos).

Our recorded PSA 10 median fell from $1,787.50 on 91 verified sales on August 28 2026 to $1,311.93 on 98 sales on September 23, a fall of 26.61% over 26 recorded days in which seven sales were added. The daily ask tape does not reach back that far: over its own 24 days from August 30 the raw ask fell 1.39%, from $211.25 to $208.31. The weekly per-card series runs 116 readings from January 27 2025 to September 22 2026, from $65.92 to $210.11, a rise of 218.7%, with its low of $56.51 on April 7 2025 and its high of $211.25 on August 30 2026. Its last reading is 0.5% below that high, and the day-20719 ask of $208.31 is 1.4% below it.

The shared-line detector flagged 883 of 24,385 population keys and did not flag this one. The grade-by-grade pooling ratio, recorded sales per graded copy against a declared bar of 1.0, passes with room: 98 verified PSA 10 sales against 4,546 recorded tens is 0.02, and 251 verified PSA 9 sales against 18,384 recorded nines is 0.01. Both absolutes are therefore printed. Across all 66 products this series links the staleness assertion ran on 258 graded rows at every grade and every grader and removed one, and the pooling ratio ran on the 114 surviving rows that sit at PSA 10 or PSA 9, the only two grades our registry records a count for. Nothing on that ground cleared 0.94. The 18,384 recorded PSA 9s against 4,546 tens is itself the story: four nines exist for every ten.

The raw price is an ASK and every graded price is a SOLD median. A 6.30x multiple divides a listing by a transaction and contains no fee, postage, wait or failure rate; at a 2.53% marginal gem rate the expected outcome of a submission is emphatically a 9, and is grading Pokemon cards worth it is where that arithmetic lives. The marginal rate rests on 474 slabs over 32 days, which is a real measurement but a short one, and a grader's queue can be lumpy. The registry is a count of submissions recorded by our provider, not a count of cards in existence.

The claim is that this card gems far below its cohort and that the slabs being added now gem far below its own history. A recompute of the 522-product deep-registry population placing this card outside the lowest 15% by gem rate would overturn the first half. A later registry window in which the marginal rate returns to the lifetime 13.96% would overturn the second, and that is the observation to watch. Related card studies: Mewtwo & Mew GX SM191, M Charizard EX from Evolutions, Magikarp & Wailord GX, Alakazam ex from 151. Handout and giveaway cards as a class: McDonald's Pokemon cards value.

Sources: the `cards` catalog, `price_history`, `graded_snapshots` and `population_snapshots` in production, read on day 20719, September 23 2026. The card is resolved by NAME AND NUMBER: Special Delivery Charizard, number SWSH075, set SWSH Black Star Promos, catalog id swshp-SWSH075, TCGplayer product id 282259. A folded same-name sweep across all 62,388 products returns exactly one row, so this card has no confusable printing. RAW is an ASK, the day-20719 `price_history` reading. GRADED figures are SOLD medians over verified completed sales with the sale count beside them, latest `graded_snapshots` row per grade across both keyspaces with the fresher row winning; the product-id keyspace is fresher at every grade here. REGISTRY is `population_snapshots`, latest reading per key, September 22 2026 for this card; the latest day differs per card across this series. Gem rates pass `gemRateIsTrustworthy()`. Line sharing is tested by `sharedRegistryKeys()` over all 24,385 population keys (883 flagged, 788 products affected, this card not among them) and by the grade-by-grade pooling ratio described below. A registry series is also cut by `publishableRegistry()`, which withholds a figure when the line's identity changed and the new line carries only one reading; that rule is what withholds the Special Delivery Pikachu and Bidoof above. Two assertions run over the graded rows this series prints, and they do not cover the same rows. STALENESS: on every product this module prints a graded figure for, and after the fresher-keyspace rule, any median dated before day 20695, the first day of the current daily tape, is dropped, because the never-refreshed catalog-id keyspace holds rows old enough to invert against their own neighbouring grade. That assertion runs on all 258 graded rows the 66 products this module links carry, at every grade and every grader, and removes exactly one, a CGC 10 dated day 20694 on a card whose graded figures are not printed here, leaving 257. The cohort below is built WITHOUT that gate: applied across the whole catalog it drops nine PSA 10 medians, leaving 8,684 products with a PSA 10 median on two or more sales and a cohort of 5,871, 4,183 English and 1,688 Japanese, and no rank printed in this series moves by more than seven places. POOLING RATIO: recorded sales at a grade divided by the registry count at that grade. This is NOT an impossibility test, and calling it one would be wrong: `graded_snapshots` counts SALES and one slab can sell more than once, so a ratio above 1 is evidence that a graded key is pooling another product, not a physical contradiction. The declared bar is 1.0. `population_snapshots` records a per-grade count at PSA 10 and PSA 9 only, so the ratio is computable on 114 of those 257 rows and on no BGS or CGC row at all; the highest of the 114 is 0.94, 29 PSA 10 sales against the 31 tens on the Wizards-era Eevee's line, and only three exceed 0.5. THE COHORT: 62,388 products after deduplicating 63,600 catalog rows on the TCGplayer product id; 5,879 carry an ask, a PSA 10 median on two or more verified sales, a trustworthy gem rate and an unshared registry line. THE DEEP-REGISTRY POPULATION is the 522 of those products, without the ask and comp gates, whose registry holds 10,000 or more slabs, 298 of them English; ranks are ascending by gem rate with ties broken on the product key. THE MARGINAL-RATE POPULATION is the 1,330 cohort members with two or more publishable registry readings, 40 or more slabs added between the first and last, and a nonzero gem count on the first; the marginal rate is tens added divided by slabs added over that span. Six further rows are dropped because their tens added fall outside zero to slabs added, which is a line contradicting itself inside its own publishable segment rather than a rate; without that rule the population is 1,336 and this card ranks 131st rather than 126th. The Charizard ranking recomputed here uses English priced products whose name contains Charizard and no ampersand and which carry a PSA 10 sold median, 83 products on day 20719; the published Charizard premium ranking used a canonical single-species matcher on day 20718 and reported 82, and both rules are stated so a recompute can pick either. Percentile is the share of the cohort strictly below this card's value; medians are upper-middle on even samples. The weekly per-card series is `ppthist:282259`, first reading January 27 2025, last September 22 2026. Not financial advice.