Methodology
TCGBerg's analytical framework rests on three pillars, applied in sequence: (1) Fair Value, a confidence-scored estimate of each atom's current price, derived from a blend of four statistical methods over the trailing transaction window; (2) Market Cap, fair value multiplied by the graded population of that atom, giving the implied total slabbed value; (3) Indices, cap-weighted, confidence-dampened aggregations that produce a single benchmark level for a set or print run at a grade tier. Every figure on TCGBerg flows from these three layers. The atom, the (printing, grader, grade) tuple, is the irreducible unit of analysis: not the card, not the printing, but the specific quality-graded copy the market actually transacts. Two tools then act on the pillars: deal detection, which flags listings priced below a confident fair value and verifies them, and the sell-price optimizer, which prices a card to sell by your deadline. This page is the gateway to the detailed methodology for each.
Three pillars
Each layer compounds the one below. Atoms are valued; values are summed into cap; caps are weighted into indices.
Fair Value
Confidence-scored price estimate per atom. Four blended methods (sale-rank EWMA, trimmed median, recent-window robust, trend projection) with adaptive weights based on dispersion, sample density, and recency, over a winsorized 30-sale window. A 0-100 confidence score per atom captures uncertainty.
Market Cap
Fair value x graded population per atom. Aggregations are sums across atoms, rolled up by set, grade, or print run. The atom is the unit; market cap is the rollup. Populations come from the graders' published reports and pair with each day's fair value.
Indices
Cap-weighted index series with square-root confidence dampening and continuous-divisor accounting, so population growth and rebalances never move the level. Per-grade indices isolate one tier; composites span grades. Base 100 at inception.
Tools & signals
Two features built on the pillars: deal detection flags listings priced below a confident fair value and verifies them; the sell tool prices a card to sell by your deadline.
Deal Detection
Active listings priced below a confident, fresh fair value, then run through a verification cascade (grader, grade, printing, PSA certificate) before they reach the screener. A four-component score ranks them; auctions are framed as current bids, not lockable discounts.
Sell Price Optimizer
The price to list a graded card at to net the most by your deadline. A probability-of-sale model over price relative to fair value, hierarchical pooling for thin cards, and fee-aware optimization, with an honest confidence read.