1. Canonical identity
Every card is resolved to a single canonical identity — game, year, set, card, edition, language — and then to a graded variant for a specific grader and grade. Prices are only ever compared within the same graded variant.
Intelligence
Methodology
The engine is deterministic and explainable. The same inputs always produce the same scores, and every score can be traced back to the transactions and population records that produced it.
2. Cleaning
Raw transactions are converted to sterling, deduplicated across venues, and screened for outliers using a robust median-absolute-deviation test. Excluded sales stay visible in the report; they simply do not influence the valuation.
3. Fair value
Fair value is a weighted robust location estimate of cleaned comparables. Recent sales carry more weight through exponential time decay, verified sales carry more weight than unverified, and public auction results carry more weight than private treaty. Where comparables are thin or widely dispersed, the confidence interval widens and the data confidence score falls.
4. Component scores
Eleven components are calculated independently from observed data: value, scarcity, demand, liquidity, momentum, stability, grade quality, structural collectibility, long-term strength, supply risk and data quality. Each is bounded 0-100.
5. Two headline scores
Quality asks how good the asset is, independent of price. Opportunity asks how good it is at the price available today. A superb card at a poor price scores high on quality and low on opportunity — that distinction is the point of the model.
6. Maximum acquisition price
Fair value is reduced by a required discount that grows with illiquidity, price volatility and data uncertainty. The result is the highest price at which we would transact. It is a discipline rule, not a forecast.
Quality score weights
tcgs-v1.0.0- Scarcity
- 20%
- Demand
- 20%
- Liquidity
- 15%
- Structural
- 20%
- Grade Quality
- 10%
- Long Term Strength
- 10%
- Data Quality
- 5%
Opportunity score weights
tcgs-v1.0.0- Value
- 22%
- Liquidity
- 16%
- Scarcity
- 14%
- Demand
- 12%
- Momentum
- 10%
- Stability
- 10%
- Grade Quality
- 8%
- Structural
- 8%
What the engine deliberately does not do
- It does not predict future prices. Momentum measures what has already happened.
- It does not use machine learning in this version. Every number is a rule you can read here.
- It does not score a card it cannot evidence. Thin data produces an INSUFFICIENT DATA signal rather than a confident-looking guess.
Scores, fair values and maximum acquisition prices are analytical opinions produced from illustrative demonstration data. They are not valuations, appraisals, price predictions, financial advice or an offer of any investment.