THE POKEMON CARD TERMINAL
Base 100 = January 1, 2021. A level of 145.6 means the graded market is up 45.6% since.
The index moves only on chained same-card price returns — tracking more cards never moves it; only prices do. Adding coverage changes what the index can see, never what it says.
No synthetic series, ever — a scope renders the moment its index derives.
Every estimate derives from real sales — never asking prices. Model outputs never feed back in as inputs.
Every estimate derives from real sales — eBay clears, auction-house hammers, verified captures. Model outputs never feed back into index estimation, so the index can't launder its own opinion into a price.
Every sale is screened against the card's own history before it's allowed to move anything.
A single bad print — the wrong variant, a mislabeled grade, a fat-fingered auction — can poison a card's price for months. Sales are screened against the card's own history before they're allowed to move anything. And trust is earned by agreement: a lone sale that the surrounding transactions contradict moves the price only partially — until the market corroborates it, at which point it counts in full.
Every estimate carries its tier — LIVE, MODELED, or LOW CONFIDENCE. Staleness is never hidden.
A price from six sales this week is not the same as a price from one sale last year. Every estimate carries its tier — LIVE, MODELED, or LOW CONFIDENCE — and staleness is never hidden behind a confident-looking number.
Dormant cards roll forward with their cohort, inside hard bounds — no flat-lines, no fantasy moves.
Dormant cards roll forward with their cohort — inside hard bounds. No flat-lining at a 2019 price, and no fantasizing a 40× move out of thin air. When the leash tightens, the estimate says so.
Auction-house results and editorial capture cover the cards machines can't see — reviewed sale by sale.
Six-figure vintage doesn't clear on eBay. Auction-house results and editorial capture cover the cards machines can't see — reviewed sale by sale before they enter the record.
Price history is fitted through every sale at once — with the market's aggregate move as context — not bounced print-to-print.
A price history should read the way an analyst would draw it: one best-fit line through every sale on record, informed by how the card's cohort moved in between — not a line that lurches to each print in isolation. Clusters of sales pin the line at their level; disagreeing prints split the difference. And like any serious index, history is restated as new transactions arrive — the past line sharpens as evidence accumulates, while today's price is computed independently and is never bent by the redraw.