Every price site will quote you a value for every card ever printed. Type in the most obscure promo you own and a number appears, rendered with the same confident typeface as the number for a card with ten thousand recorded sales. We built our desk on the opposite premise: the honest output of thin data is a refusal. When the desk wrote long-form views on the hundred deepest tapes we track, a handful of those hundred, cards that EARNED a spot by tape depth, still came back stamped no position. Here are seven of them, and what each one teaches about when a number should not exist.
The seven, and the specific hole in each file
What was missing, card by card
| Card | The loud number | Why we declined |
|---|---|---|
| Lillie, Ultra Prism | $1,842 PSA 10 (21 comps) | one grade row: no nine tape, no pop, no gem rate, no EV, no velocity |
| Rocket's Scizor ex, Team Rocket Returns | $2,832.61 PSA 10 (5 comps) | five comps, tape stale since July 21, round $600 ask |
| Mudkip, Nintendo promos | $3,650 PSA 9 vs $1,997.50 PSA 10 | inverted ladder: comp pool likely mixing three Mudkip promos |
| Jolteon VMAX, SWSH promos | 11.95x multiple, $900 ten | registry series is two identical snapshots four days apart |
| Raichu IR, Paldea Evolved | 8.23x vs a 4.22x peer median | two identical snapshots: no growth, no drift, no velocity |
| Gengar, SWSH promos | 26.11x multiple, $705 ten | no population, no gem rate, no EV: odds unrecorded |
| Roaring Moon ex, Prismatic Evolutions | 313 comps on the ten | deep tape, three flat snapshots, nothing anomalous to trade |
Start with the famous one. The full art Lillie is among the most recognizable trainer cards of its era, and our entire file on it is a single row: a $1,842 PSA 10 median on 21 comps against a $469.31 raw ask. No PSA 9 tape. No population. No gem rate, therefore no grading odds, therefore no EV. Any site quoting grading advice on this card is working from data we cannot see or from imagination. The desk's stub says exactly that, and adds the one caveat the lone figure carries: the 3.92x multiple divides a sold median by an ask, and per our ask study, that construction flatters the number.
Rocket's Scizor ex shows the staleness problem: its headline $2,832.61 ten stands on five comps, and the newest print is from July 21, more than five weeks old at authoring time. Five prints from midsummer is a sketch, not a market. Mudkip shows the contamination problem: a PSA 9 at $3,650 on 6 comps above a PSA 10 at $1,997.50 on 11 is not an arbitrage, it is a title-matched comp pool mixing the three separate Mudkip entries in that promo series, numbers 5, 10, and 18. How Many Sales Is a Price is the desk's standing rule on both; these two cards are why the rule exists.
The subtler refusals are the two-snapshot registries. Jolteon VMAX and the Paldea Evolved Raichu both carry loud structures, an 11.95x and an 8.23x multiple, hard gem gates, fat spreads. Both also carry population series that consist of exactly two identical snapshots a few days apart: no measurable growth, no recent gem window, no drift, no velocity on any grade. Whether those registries are minting or frozen decides whether the multiples are cheap or rich, and two identical photographs cannot answer that. Anyone showing you a trend line through two points is showing you a ruler, not a market. Gengar SWSH241 is the extreme case: a 26.11x multiple with NO registry data at all, so the odds behind the biggest number in the batch are simply unrecorded.
And then there is the strangest stub: Roaring Moon ex, which we declined to rate not because the tape is thin but because it is boringly clean. A genuinely deep tape, 313 comps on the $685.84 ten, 278 on the nine, prints a 4.09x multiple mildly above its 3.1x peer median and nothing else worth a position. A desk that must say something about every card would manufacture a thesis here. Ours says: tracked, held at no position, next card. Two near-misses round out the family: Flareon ex, where three Prismatic siblings printing nines under the ask earned a set-level rule but no card-level trade, and Volcanion EX, a 1,810-slab registry photographed exactly once, priced at 20.45x as if the pool were proven closed when it is merely unphotographed.
All figures are the recorded tapes behind the desk's house views, re-verified against the live report payloads on August 29, 2026: verified sold comps, PSA registry snapshots, and the disclosed comp gates (n>=5, 0.5x-45x band) every report enforces. Stub status reflects the data available at authoring; several of these files are one population feed away from a real note, and the notes say so. Not financial advice.