How we score

The current fair value is always anchored to real, dated sales — weighted by how reliable, well-sourced, and recent they are. If there is no defensible comp, we publish no number. A dollar figure with no sale behind it is a guess dressed as data, and the whole point of the record is that we do not do that.

Each note is produced by the same repeatable steps, so two cards are scored the same way:

1. 1 · Identify

   Resolve the exact variant — this serialized prize, not the base retail print that shares its card number. Different objects, different markets.

2. 2 · Weigh the evidence

   Score each real sale by reliability, provenance, and recency. A dated eBay-sold comp counts; a hearsay asking price does not.

3. 3 · Anchor to market

   Set the current fair value from the weighted comps — never an invented number. A figure with no comp behind it is a guess dressed as data.

4. 4 · Read the cohort

   Regress value on print-run across the winner-card cohort. The residual says whether this card is rich, fair, or cheap versus its scarcity peers.

5. 5 · Forecast the band

   Project a 6–24 month range, deterministically widened by how thin and volatile the market is — plus the pre-committed triggers that would move it.

6. 6 · Put it on the record

   Freeze the note with a horizon. It resolves later against a real sale to a terminal verdict — and is never edited.

The fair value tells you what the card is worth today. The cohort read tells you whether that price is reasonable: we fit value against print-run across the serialized winner-card cohort, and measure how far this card sits above or below that scarcity line. A card can be a genuine grail and still read overvalued if its price leans on character demand rather than scarcity — that is a caution, not a sell signal.

Desirability is scored on two independent axes so competitive stats never drag down a pure collectible. Both are transparent weighted summaries on a 0–100 scale — they explain why a card is wanted; they are never added to, or confused with, the dollar fair value.

• Collector score — scarcity, character demand, artwork, and prize prestige. What makes it a chase piece.
• Competitive score — real tournament play and current meta relevance. For a serialized prize this is usually near zero, and that is correct.

The two combine into a profile label: Pure Collectible (high collector), Meta Staple (high competitive), Crown Jewel (both high), or Niche (neither). Thresholds: collector ≥ 65 and competitive ≥ 45.

Each axis score is a weighted average of the factors below, normalised to 0–10 and shown as 0–100. Weights are founder priors for model v0.1 — they will be recalibrated from observed error once real sales land against these forecasts.

Confidence is not a vibe — it is derived from how much real comp data backs the reference class. More corroborating sales, more confidence:

• ≥ 6 corroborating comps → High
• 3–5 comps → Medium
• < 3 comps → Low

Thin markets get honestly wide forward bands. A confident-looking narrow range on two sales would be the dishonest choice.

Status is deterministic, never an editorial judgement: it is computed from the frozen forecast, its pre-committed horizon, and real sales as they land. The horizon is set at publish and never extended — so we cannot quietly move the goalposts.

Independent analysis of physical collectibles. Numbers are model forecasts, not live market quotes, and not financial advice. We disclose held positions on every call and never trade against a published note.