Cheapest Way to Own a Gold Star Pokemon Card

The cheapest English Gold Star you can buy on September 23, 2026 is Registeel ☆ from [Legend Maker](/research/set-screen/legend%20maker), asking $500.00, and that ask has not moved across any of the 25 days of our daily tape. That is the short answer. The long answer is that the population is 27 cards rather than the 22 a glyph search returns, that six of the 25 EX-era cards carry no ask at all today, and that only four of them carry a PSA 10 sold median backed by ten or more recorded sales. On a run of cards this famous, the thing we are shortest of is not price. It is evidence.

A Gold Star is marked with a star beside the name, and the obvious way to find them is to search the catalog for that character. Across the whole catalog that search returns 25 rows, and inside the ten sets this study is about it returns 22. It is wrong twice over. Two different characters are in use: of those 22, twenty carry U+2606 WHITE STAR (☆) and two carry U+2605 BLACK STAR (★), and a search for one of them silently drops the other. Fold both and you have 22. Then look at the residual, which is the step that matters: five more rows spell the word out and carry no star character at all. They are Groudon Star, Kyogre Star and Metagross Star, numbered 111, 112 and 113 in [Delta Species](/research/set-screen/delta%20species), and Mewtwo Star and Pikachu Star, numbered 103 and 104 in [Holon Phantoms](/research/set-screen/holon%20phantoms).

Twenty-seven is also the number the run is supposed to have: three each in Team Rocket Returns, [Deoxys](/research/set-screen/deoxys), [Unseen Forces](/research/set-screen/unseen%20forces), Delta Species, Legend Maker and Holon Phantoms, two in [Crystal Guardians](/research/set-screen/crystal%20guardians), two in [Dragon Frontiers](/research/set-screen/dragon%20frontiers), three in Power Keepers and two in POP Series 5. The five spelled rows are not a rounding difference. Pikachu Star carries the highest PSA 10 sold median in the whole run, and a glyph search reads it as an ordinary Rare.

Three cards carry a star character and are not EX-era Gold Stars, and they are excluded here by set and date rather than by symbol. Umbreon ☆ in Celebrations Classic Collection and its Celebrations sibling Umbreon ☆ 17A are 2021 reprints of the POP Series 5 card; the Classic Collection copy asks $89.93 and its registry line holds 36,426 entries against 841 on the original. Greninja ☆ is a Sword and Shield promo numbered SWSH144, asking $20.55. A Japanese reprint, Umbreon Star 012/025 from the 2021 promo pack, is excluded on the same rule and asks $263.82. None of the four was printed between 2004 and 2007, which is the window this study is about.

Nineteen of the 25 EX-era Gold Stars carry an ask on September 23, and they total $28,313.48 at a median of $1,333.56. The floor is Registeel ☆ at $500.00, and it is the same $500.00 on every one of the 25 daily readings we hold. Flareon ☆ at $600.00 is the same story, flat for 25 readings. The two cheapest Gold Stars in English have not changed price since our tape began.

What that flatness means is worth saying plainly, because it reads like stability and is closer to the opposite. An ask is one listing. Across 52,282 card lines we found 83.5% are priced by a single listing, and those are the numbers behind Five Cards in Six Trade on a Single Listing. A price that has not moved in 25 days on a card this scarce is most likely one seller who has not sold and has not repriced. It is what you would pay, not what anyone paid. The gap between the two is the subject of Pokémon Card Ask vs Sold Price.

Six of the 25 have no ask today. Three of them have never carried one on our tape: Mudkip ☆, Rayquaza ☆ and Mewtwo Star. Three had one and lost it: Treecko ☆ last read $2,400.00 on September 20 after a 22-day tape that ran between $900.00 and $2,400.00, Charizard ☆ δ last read $1,000.00 on September 19 after a 17-day tape that ran between $1,000.00 and $4,000.00, and Kyogre Star last read $1,602.99 on September 16 from four readings. A card with no ask is not a cheap card. It is a card nobody is offering, which on this run is the ordinary state of affairs. It is also specific to this run: those six are the only six cards without an ask in the entire 1,694-card EX era, a population worth $87,845.77 on the same day, which EX Era Pokémon Cards Value measures in full. Six of the twenty-five cheapest-route candidates cannot be priced today, eight of the twenty-seven once the two POP Series 5 cards are counted, and nothing else in their era has that problem. Their card pages are not blank, which is worth knowing before you click one: on September 23 Rayquaza ☆ quoted $2,500.99 and Mewtwo Star quoted $1,400.00 from a source our daily tape does not carry, and Mudkip ☆ quoted nothing at all.

Graded figures here are sold medians, and two rules run before any of them is used, because on a 27-card population the thin rows disagree wildly with each other. The first rule drops a median dated before August 30, 2026, when the current tape began: the graded feed rotates about a seventh of the catalog a day, so a row older than that is a reading from a different week of the market. It removes exactly one row here, Metagross Star's PSA 10 median of $5,500.00 on three sales, dated July 21, 2026. That row is also the clearest argument for the rule, because it sits below the same card's 32-sale PSA 9 median of $8,057.19 dated today, which no functioning market produces. The second rule is a minimum of ten recorded sales, and it removes seven more PSA 10 medians and one PSA 9: Pikachu Star at $99,374.99 on three sales, Jolteon ☆ at $44,444.85 on six, Regice ☆ at $27,569.00 on four, Vaporeon ☆ at $24,129.38 on seven, Registeel ☆ at $21,500.00 on five, Regirock ☆ at $16,000.00 on six, Celebi ☆ at $13,200.00 on eight, and the PSA 9 on Latios ☆ at $14,847.50 on four. All nine rows are printed in the table above with their samples and none of them ranks anything.

The median of those 18 PSA 9 medians is $6,659.50. The cheapest is Latias ☆ at $910.00 on eleven sales, and it is the only card in the run whose graded median sits BELOW its own raw ask: the slab sold for 55 cents on the dollar of what a raw copy is being asked. That is the one row in this study where the answer to the headline question changes shape. If you will take a PSA 9, the cheapest recorded route into a Gold Star is not the $500.00 raw Registeel, it is a $910.00 Latias in a slab, and the raw Registeel's own PSA 9 median is $4,153.50. Read that pair carefully: those are two different instruments, an ask and a sold median, and the Latias figure rests on eleven sales.

Three of the four are the [Unseen Forces](/research/set-screen/unseen%20forces) legendary birds, and the fourth is Flareon. That set has its own study in Unseen Forces Card Prices, and the era around it in EX Era Pokémon Cards Value. Of the 21 EX-era Gold Stars outside that table, fourteen carry no usable PSA 10 sold median, thirteen because the feed holds no row for them at all and Metagross Star because the stale rule withdrew its, and the other seven carry one on eight recorded sales or fewer, so none of the 21 has a PSA 10 price we would quote. The [Unseen Forces](/research/set-screen/unseen%20forces) trio is also the only place in the run where the graded tape is deep on both grades at once, 155 and 156 PSA 9 sales beside 46 and 20 PSA 10 sales, and it is the part of this study a competitor could actually check against their own sales feed.

Eleven of the 25 EX-era cards carry a usable PSA 10 sold median of any sample size, eight carry a PSA 9 and no PSA 10, and six carry neither. Those three buckets sum to 25. Absent is not zero: a card in the last bucket has not sold for nothing, it has not been recorded selling. Across the English catalog we found 14,466 of 21,416 cards carry a live price and no graded sale at all, which is the population Cards With a Price and No Sale measures. The Gold Stars sit inside that majority, not outside it, and they are among the most recognisable cards the game has printed.

The two POP Series 5 cards are excluded from every priced and graded figure above, and the reason is that the graded rows do not describe the 2007 card. Umbreon ★ carries a PSA 10 sold median of $363.00 on 121 recorded sales against a registry line holding 53 PSA 10s, which is 2.28 sales per graded copy. That alone would not condemn it, because slabs resell and a bar at 2.5 per copy would have published it, so the threshold is a judgement and the corroboration is what settles it. Its $363.00 sits $10.09 below the 2021 Celebrations Classic Collection Umbreon ☆, which reads $373.09 on 718 sales, and $21.50 below the Japanese promo-pack reprint at $384.50 on 43. Its own ask tape agrees that the row is not the 2007 card: thirteen readings between September 7 and September 19, $104.00 once and $71.35 for the other twelve days. Espeon ★ has no tape reading and no graded row at all, and when we loaded its card page on September 23 it quoted $1,900 from a source our daily tape does not carry. We publish the tape and name the gap rather than reconcile two instruments we did not measure together. The provider-collision problem itself is the subject of The Most Graded Cards in Pokemon.

Every English Gold Star has a Japanese original, and we hold all of them. Thirty-two catalog rows carry the Japanese Gold Star names, and they collapse to 27 distinct prints once the five sets we hold under both a romanised and a Japanese name are folded together: Kyogre Star, Registeel Star, Treecko Star, Entei Star and Suicune Star each appear twice. Twenty-seven Japanese prints for 27 English cards is the right shape. We price none of them. Not one of the 32 rows carries an ask on September 23, and not one carries a graded sold median at any grade. So the question this study answers cannot be answered on the Japanese side with our data, and a reader hoping the Japanese copy is the cheap route will have to look elsewhere. Japanese sets have no page on this site, so those set names appear here as plain text while the cards are linked normally. The Japanese catalog as a whole is measured in Japanese Promo Card Prices and Vintage Japanese Pokémon Card Prices.

Four limits, in the order they would bite. First, every raw figure is an ask on one day, and on most of these cards it is one listing that has not moved in 25 days; this study holds no evidence about what a raw Gold Star actually sold for. Second, our daily tape is 25 days long, so no trend, no peak and no annual figure is available for any of the 27, and none is published here. Third, the graded medians come from a provider's recorded sales, and the seven PSA 10 rows under ten sales are printed but not ranked because rows that thin disagree with each other by an order of magnitude on adjacent cards. Fourth, a PSA registry line can be shared by several products, so no slab count is printed as a slab count anywhere in this study. On the specific question of whether the Gold Stars are hard to gem, the era-level answer is in Hard to Gem and the EX-era set-level answer is in EX Era PSA 10 Gem Rates; we do not restate it here from 27 rows.

A single sold comp under $500.00 on any of the 27, at any grade, would overturn the headline. So would a PSA 10 sold median appearing on ten or more sales for a fifth card, which would move the count from four. And a sixth catalog row spelling a Gold Star some third way, which our residual check did not find, would move the population from 27. The residual query we ran is printed in the methodology so that anyone can run it and disagree.

Population. Every catalog row whose folded name carries U+2606 ☆, U+2605 ★, or the word `star` on a word boundary, over the whole 62,343-product catalog after deduplicating on `tcg_id` (1,212 rows collapsed). Folding normalises to NFKD, strips combining marks, maps curly apostrophes to straight, maps both star characters to a single token, and drops the gender signs. Inside the nine EX-era sets and POP Series 5 that query returns 22 rows carrying a star character, 20 of them U+2606 and 2 U+2605, and on the residual five more spelled rows in Delta Species and Holon Phantoms, 27 in all. Across the rest of the catalog it also returns modern and Japanese `Star` and `Prism Star` rows, excluded by set and date. The derivation was re-run over all 63,555 catalog rows rather than the 62,343 dedupe survivors, in case a survivor carried a different spelling from a row it replaced: it returns the same 27, none of the 27 shares a `tcg_id` with another row, and every card number in the run is occupied by a row the query catches. The EX-era English population is the 25 cards numbered 107 to 109 in Team Rocket Returns, 105 to 107 in Deoxys, 113 to 115 in Unseen Forces, 111 to 113 in Delta Species, 90 to 92 in Legend Maker, 102 to 104 in Holon Phantoms, 99 to 100 in Crystal Guardians, 100 to 101 in Dragon Frontiers and 100 to 102 in Power Keepers, plus POP Series 5 numbers 16 and 17, which are excluded from every figure for the collision documented above. Every card was resolved by name AND number against the catalog, not by an id we typed. Exclusions: `cel25-17A` and `cel25cc-CC015` (2021 Celebrations reprints), `swshp-SWSH144` (a Sword and Shield promo), `jat-594630` (a 2021 Japanese promo-pack reprint). Prices. `price_history` on day 20719, 2026-09-23, 52,378 priced products; these are ASKS, the daily dump's market figure for the canonical printing rather than the dearest one. The 25-day tape runs from day 20695, 2026-08-30. Where a card has no reading on day 20719 the last reading and its date are printed instead of a blank. Graded. `graded_snapshots`, latest day per card key and grade, read across the catalog-id and TCGplayer-id keyspaces and resolved by taking the fresher row; preferring the catalog id would have served Raikou ☆ at $4,653.02 from day 20675 while the current row reads $20,675.50. These are SOLD medians. Stale-row rule, applied across this whole batch: after the fresher-keyspace read, any median dated before day 20695 (2026-08-30, the start of the current daily tape) is dropped, because the graded feed rotates roughly a seventh of the catalog a day and an older row is a reading from a different week. It removed one row in this population: Metagross Star's PSA 10 median from day 20655. Collision test, run on every graded row used in this study: several products can share one PSA registry line, so a sold count at a grade is compared against the registry count at that grade whenever the registry row passes the trustworthiness check, and a row recording more than two sales per graded copy is withheld. One sale per copy is not a ceiling, because a slab can be sold more than once, so this is a declared threshold and not an impossibility; the stricter one-sale-per-copy bar is counted beside it below. All four gated PSA 10 rows and all eighteen gated PSA 9 rows in the EX-era population pass it; the only failure among the 27 is pop5-17 Umbreon ★, already excluded. Sensitivity: the stricter one-sale-per-copy bar withholds the same single row, so no figure in this study differs between the two thresholds. A minimum sample of ten recorded sales applies to every comparison and every ranking; the nine rows it removed are named in the text and printed in the tables with their samples. Registry. `population_snapshots`, latest day per key, used only for the collision test on Umbreon ★ and never printed as a slab count. Median convention: upper-middle on even samples, used for the $6,659.50 PSA 9 median and the $1,333.56 ask median. Japanese prints: the same folded query restricted to the fourteen Japanese sets that hold EX-era Gold Stars, 32 rows folding to 27 distinct prints, zero priced and zero graded.