Pokémon & collector appeal
Species, published fan rankings and separate editorial collector tiers help describe demand. Unassessed Pokémon are not treated as unpopular.
Collector analytics · RVM V1.1
What would comparable cards suggest this raw card is worth? Relative Value estimates a price and a range for the version you select, then compares that range with its market price.
The short version
Species, published fan rankings and separate editorial collector tiers help describe demand. Unassessed Pokémon are not treated as unpopular.
The model separates rarity, known alternate/full-art styles, collector form, foil treatment and edition. Switching treatment selects that version’s own estimate.
Audited specific-card pack odds, era, set, release age, released printing availability and artist are considered together. Missing odds stay missing; current pack prices never enter the estimate.
Held-out pricing errors determine the range. The target is about 80% coverage, so a precise-looking range is only useful if the model can support it.
A model you can follow
The model learns relationships on a logarithmic price scale. In plain English, it combines price adjustments for the card’s characteristics. It also allows important relationships to differ: a popular Pokémon’s premium can change with rarity, and a reverse-holo premium can change with era and set.
Estimate = a learned reference price × the card’s factor adjustments × a calibration adjustment
Expected range = the estimate’s independently measured lower and upper pricing errors
Relative Value label = the selected version’s market price compared with that range
Price factors show model associations, not causal premiums. Artist is compared within the same era, rarity and treatment, with other regression inputs controlled. Card category is context only. Limited-printing older commons and uncommons do not receive a price verdict or a displayed numerical range: a broad peer model can miss their species-specific scarcity. Catalog printing counts measure available card designs, not remaining physical supply.
Normal, reverse holo, holo, Poké Ball, Master Ball and editions are modeled using their own priced records. There is no universal 1.5× or 2× reverse-holo rule. An unmodeled version is shown as unavailable rather than borrowing another version’s estimate.
The supplied popularity research covers 1,025 species and distinguishes fan surveys from 214 editorial collector assessments. The collector tiers partly reflect market judgment; they are not an independent statistical measurement. Breakout predictions and the spreadsheet’s selected price comparisons are excluded.
The model uses applicable canonical specific-card odds, including reviewed historical and community estimates. Rarity-group odds are divided among the eligible cards under an equal-likelihood assumption. Parallel foil treatments do not automatically inherit another slot’s odds.
The selected card’s price, price-derived chase rank and current sealed-product prices are not predictors. Scarcity cannot by itself explain why collectors prefer one equally scarce card over another.
Reading the result
Prices can sit outside the range for good reasons. A low result is not a promise of a bargain, and a high result does not mean collectors are wrong. This estimates current raw-card pricing, rather than next month’s price or the value of a graded copy.
Too few comparable printings for a reliable range or price verdict. The market price remains available.
The selected version’s current market price is below its estimated peer range.
The selected version’s current market price falls inside its estimated peer range.
The selected version’s current market price is above its estimated peer range.
Uncertainty
The range targets about 80% coverage across held-out prices, not every individual card category. It is calibrated using pricing-error groups with at least 100 independent card/product observations. Narrower groups are used when supported; broader groups provide a fallback.
Narrower measured uncertainty and at least 50 matching training peers.
Moderate measured uncertainty and at least 20 matching training peers.
Wider uncertainty or limited comparable records.
Release evidence
The V0.3 comparison uses the same 18,864 preferred card versions at the current snapshot. All five published prediction folds keep linked card/product groups separate. A whole-set stress test checks unfamiliar sets, where error is higher. These are cross-sectional checks on one dated price snapshot, not evidence of future-price forecasting accuracy.
Price snapshot: 10/27/2026. Estimates are versioned and do not silently retrain when the daily market price changes. Previous releases remain available for rollback.
Why a card may differ
The details panel highlights relevant limitations, such as a release within the last 90 days, uncertain pull odds or few comparable records. Launch excitement may settle, but the model does not assume when or by how much.
Standout chases. A memorable composition, character pairing or collector following can command a premium beyond general popularity and rarity.
Market conditions. Playability, short-term supply, stale quotes and demand shifts can affect individual cards.
Condition and printing. Grading, condition, stamps and unusual print differences need their own comparison.
No invented certainty. Missing odds stay missing. A model miss does not establish that the market is mispriced.