How SlabIndex reads the market

What this is

SlabIndex indexes the trading card market the way a financial data provider indexes equities: a daily price for every tracked card and for every graded copy of it, aggregated into per-game indices, with a model on top that reads each asset's own price series and says what it thinks is happening.

It currently tracks 49,887 cards across 4 games, of which 33,543 carry a graded market.

• Pokémon23,142
• Pokémon (Japanese)20,992
• One Piece Card Game4,318
• Riftbound: League of Legends TCG1,435

Where the prices come from

Raw prices come from TCGplayer's published market data. Graded prices, certified populations and completed eBay sales come from a mix of providers, and which one answered is recorded on every figure — because they are not equally good evidence.

A price backed by completed sales is labelled confirmed. A provider quote consistent with the grades either side of it is labelled a quote. A figure that survives neither check is never published as a price at all. Price history carries the same distinction: daily completed sales, a provider's monthly market level, and our own daily readings of a quote are three different things, and a chart says which one it is drawing rather than blending them into a single line.

Why graded cards get their own model

Most trackers price the raw card. But a large share of the money in this market sits in slabs, and slab prices are driven by a different variable: the population report. A card with ten thousand PSA 9s and forty PSA 10s behaves nothing like one with an even distribution, even when the raw copies trade identically.

So both are modelled, separately. Every card with a graded market carries a full ladder — PSA, BGS and CGC, rung by rung — with the certified population at each, the completed sales behind each price, and that grade's own price series. Each rung gets its own signal read from its own series, because a PSA 10 can climb on a shrinking gem population while raw copies sit flat. On top of that sits an expected-value calculation: given this card's own population report, what does a submission return after grading fees and postage, and what are the odds it loses money?

How the rating works

One weighted model, five factors: trend (35%), momentum (20%), range position (20%), grading edge (15%) and liquidity (10%). It produces Strong Buy through Strong Sell, or Not rated — which is a real answer, not a polite way of saying Hold.

Two things make a call harder to reach rather than different. Thin evidence raises the bar: a card with few readings and few recorded sales must clear a score half again as high before it leaves Hold. Thin coverage raises it too — when only part of the model can be computed, the rating is re-weighted over what fired but is not allowed to claim full conviction on a fraction of itself. A reading resting on a single factor cannot reach Strong Buy at all.

How the record is kept

Each week the model publishes its strongest calls across every game it covers, on raw cards and graded slabs alike, at most one per set. A call names the exact asset it is on — a slab is entered and judged at that grade's price, a limited printing at its own — and is written to an append-only ledger with an entry price and date that are never revised, then tracked to resolution at the end of a fixed 30-day window.

Raw calls are made at Strong Buy. Slab calls are made at Buy or better against full factor coverage, because the graded market's scores sit structurally below the raw one and the same label would otherwise mean two different things. Which bar a call was made at travels with it, and the track record is tallied by game and by asset rather than blended into one number. Losses stay on the record beside the wins.

Nothing has resolved yet — the ledger is younger than one 30-day window, and the page says so rather than showing a placeholder.