issue: Cardmarket idProduct shared across unrelated cards — systematic report (commons inherit Shiny prices, promo blocks, ex2/ex4 mirror)

[marcelfmm-dotcom]

Hi! First of all, thanks for TCGdex — it powers the pricing layer of a Brazilian collection site and works great overall.

While validating prices across the full catalog (~23,400 cards), I found that many unrelated cards share a single Cardmarket idProduct, producing absurd prices. This seems to be the systematic version of what #906, #1939 and #1941 report for individual cards. I verified the shared ids directly through the API:

Card A	Card B	Shared idProduct	avg30
pl4-72 Ponyta (Common)	pl4-SH11 Ponyta (Shiny)	278869	€60.34
sm12-66 Pikachu (Common)	sm12-241 Pikachu (Secret)	398534	€137.12
xyp-XY101 Mewtwo	xyp-XY107 Mewtwo EX	554275	€4,675
xyp-XY125 Mewtwo EX	xyp-XY192 Mew	554275	€4,675
ex2-5 Flareon	ex4-5 Team Aqua's Sharpedo	275782	€52.23

The patterns I could isolate:

1. Commons inheriting the set's Shiny/Secret product. The Platinum-era SH shinies and SM-era secret rares share their product with a regular common of the same Pokémon: pl4-72/74/52 ≡ pl4-SH11/SH12/SH10, pl1-81/97/102 ≡ pl1-SH4/SH5/SH6, pl3-7/8/9 ≡ pl3-SH7/SH8/SH9, dp7-58/59/60/80/81 ≡ dp7-SH1/SH2/SH3, sm12-54/66/68/76/96 ≡ sm12-239/241/242/243/245, sm1-162≡168, sm1-163≡171. Result: bulk commons priced €25–€140.

2. One set mirrored onto another, number by number. ex2-N and ex4-N share products for 25+ numbers (2, 5–14, 16, 20, 24, 25, 31, 63, 72, 73, 94, 95, …) — completely different cards (see the Flareon/Sharpedo row above).

3. Promo sets: many cards → one product. Besides the xyp quartet above (all at €4,675!): xyp-XY203–XY210 (7 cards, one price), xyp-XY64≡XY66≡XY73, and in swshp a 10-card cluster (SWSH020/039/063/143/145/153/198/234/285/286) plus several 4–5 card blocks (SWSH139–142, SWSH155–158, SWSH159–162+223).

4. Single cards mapped to an unrelated product. me02-014 Moltres (Phantasmal Flames, plain Rare) carries idProduct: 858503 at avg30 €17.46 with no TCGPlayer block — that places a bulk Rare above most Special Illustration Rares of its own set (the rest of me02 looks perfectly sane). Cross-set identical-price pairs suggest the same class: me01-163≡neo4-33, me01-136≡swsh12tg-TG07, me03-099≡ecard3-65, me04-104≡col1-27, me04-105≡sv06-225, me04-098≡swsh5-182. (#1939's swsh1-140 Snorlax also showed up in my sweep.)

How I found these: daily full-catalog collection through the API, then (a) clustering cards with byte-identical (trend, avg30) pairs and (b) flagging cards ≥15× the median of their own set+rarity group with no TCGPlayer price to corroborate — then confirming the shared idProduct via the API for the examples above. I have a machine-generated list of ~243 suspects (most falling into the patterns above) and I'm happy to share it or help test a fix.

Thanks again for the project!

[marcelfmm-dotcom]

Follow-up on pattern 2 (the ex2/ex4 mirror) — it turns out to be both much bigger and much more mechanical than I described, and I think it reduces to a single wrong offset.

Cardmarket idProducts are assigned in contiguous per-set blocks. Deriving each block from idProduct − card number against the API today (2026-08-20):

set	cards	block implied by its offset
ex2	100	275778–275877
ex3	100	275878–275977
ex4	97	275778–275874 ← ex2's block
ex5	102	276075–276176
ex6	116	276177–276292

ex2 → ex3 → ex5 → ex6 chain without gaps. And the hole between ex3's last product (275977) and ex5's first (276075) is exactly 97 slots — precisely the size of ex4.

So ex4 didn't get scattered collisions. It got ex2's offset: 275777 + N where it should be 275977 + N.

Two things fall out of that:

• 91 of ex4's 97 cards carry 275777 + N and collide 1:1 with ex2 — numbers 1–89, 94 and 95. My original "25+ numbers" was a big undercount; it's effectively the whole set.

• 6 cards already have it right. ex4-90/91/92/93/96/97 carry 275977 + N (276067–276074) and price independently:

  card	id	avg30	same number in ex2	id	avg30
  ex4-90 Cradily ex	276067	€49.66	ex2-90 Claw Fossil	275867	€0.23
  ex4-91 Entei ex	276068	€81.26	ex2-91 Mysterious Fossil	275868	€0.33
  ex4-92 Raikou ex	276069	€104.81	ex2-92 Root Fossil	275869	€0.20

Those 6 are the useful part: the correct block is already present in the data, so the remaining 91 should just be 275977 + N.

Spot check as of right now:

ex4-5   Team Aqua's Sharpedo   id=275782   avg30 €52.23
ex2-5   Flareon                id=275782   avg30 €52.23   <- same product

ex4-89  Blaziken ex            id=275866   avg30 €0.58
ex2-89  Wally's Training       id=275866   avg30 €0.58    <- a Trainer pricing an ex

ex4-89 Blaziken ex being priced from a bulk Trainer is the clearest single symptom.

If it helps, I'm glad to run the same test across the whole catalog — "does this set's offset chain with its neighbours?" is cheap to check and would catch any other set that inherited a neighbour's offset the same way. Happy to open a separate issue for anything it turns up, or add it here.