How Our Prices Work
Five Numbers, One Rule

Sold Comps

What a card actually traded for — completed sales from PriceCharting (Pokémon) and SportsCardsPro (sports). This is the ranking basis for the Bangers board, the marking basis for every one of our set indices (the three Bowman indices since Sep 25, 2026), and the only basis we ever grade a call against. When a sold comp and anything else on this page disagree, the sold comp wins. If the candle is green, more people bought than sold that day — that's the only absolute truth in this market.

Live Asks

What sellers are asking on eBay right now, pulled live through our own API. Asks are wishes, not trades — so every ask on this site is labeled "asks" and never dressed up as a sold price. When we publish an ask mark it goes through the house trim: drop the two cheapest listings (bait and "read description" traps live there), take the median of the next ten, shipping included. No number publishes unless at least six clean listings survive with a spread under 4x — when a market fails that gate, the page says so instead of carrying a fake number.

Engine Marks

Every night an automated engine re-marks a fixed list of tracked cards — the Signal Board names and a few more — using a trimmed median of the cheapest verified asks. Engine marks power the Signal Board and flow into the Vault as daily chart points. They are ask-side and labeled that way; they can and do diverge from sold comps, and we show both rather than blending them.

Index Levels

An index, not a vault. Each index is a basket of one set’s cards, price-weighted like the Dow, so the expensive chase cards move it most. No caps. The level is the cost of one copy of every card in the basket, indexed to 100 at the set’s base month — a card’s weight is its price over that sum and moves with prices. Nothing is ever trimmed and there is no rebalance, because a set is not an ETF: you cannot hold 0.47 of a Charizard. (We ran a 25% single-card cap from Sep 15 to Oct 12, 2026 and retired it; every page now prints the top card’s share, the top-5 share and the effective number of holdings, and sets where one card is a quarter or more of the basket carry an “ex” line showing what the rest of the set did without it.) Membership changes quarterly at a reconstitution, through a divisor change that never moves the level. A new set gets a release window: for 90 days after release, a card that newly clears the sales screen enters at the next Monday or Thursday mark instead of waiting for the quarter (a new set’s chase cards list their sales late), logged as a divisor change that never moves the level; cards only leave at a reconstitution. 100 is the set’s release month, or for sets older than the sold history, the first month it reliably covers.

Each set index is a basket of marks divided by a divisor, set so that 100 is the set’s release month — so the level IS the cumulative return since release. For sets older than PriceCharting’s sold history (the 1999–2003 WOTC sets and Hidden Fates) 100 is the first month that history reliably covers the basket, and the page says so. Our live marks started in August and September 2026; the months before are reconstructed from PriceCharting’s monthly ungraded price history with the current basket, drawn dashed and labelled, never mixed up with a live mark (rebased Sep 30, 2026). The Bowman indices follow the same rule since Oct 1, 2026 (100 = the release month), and their charts shade the first 30 days after release, when most new Bowman cards print their high. Charts use a log scale, so a 10% move is the same height at 50 as at 200. When cards enter the basket the divisor adjusts (and the adjustment is logged) so composition changes never move the level; only prices do. Which cards are in a whole-set basket: every card that trades as an ungraded single (at least 6 clean sold comps in the trailing window, 4 to stay — 30 days on modern sets, 90 on the 1999–2003 WOTC sets, where raw singles sell a few times a month), plus, since Oct 6, 2026, every holo-tier card as a standing constituent: holos, Shinings, Crystals, secret and illustration rares by printed rarity. A set’s value sits in its holos, and raw holos sell slower than commons precisely because they cost more, so a pure liquidity screen was measuring the commons (Skyridge held 3 of its 38 holos). A holo with fewer sales than the screen is marked on the nearest window with clean sales (90, 180 or 365 days), flagged thin and dated in the table; a holo with no clean raw sale in a year is named on the page as tracked but unpriced. The six Pokémon chase indices are re-marked every Monday from sold comps; the whole-set indices (TH26, SV151, SSP24, MEG25, the fifteen classic sets and the nine WOTC 1st Edition and Shadowless sets) and the three Bowman indices (BOW26, BB26, BCB26) re-mark Monday and Thursday, also from sold comps. An index is a measurement, not a recommendation.

Proxy Feeds

The engine tracks a specific list of cards — not every card exists in it. If you Vault a card without its own feed (say, a Sapphire when the engine tracks the Chrome auto), the Vault can follow a related card's feed instead, marked with an orange PROXY badge naming exactly what differs. A proxy shows you a related market's direction; it is never presented as your card's own price. Your own logged prices always take precedence on your charts.

Which Number Do I Trust?

Solds beat asks. Dated beats undated. Your own logged purchase beats everything, because you know what you actually paid. And when one of OUR published numbers turns out wrong, it doesn't quietly disappear — it gets graded on the track record, misses included. That page is the reason you can trust this one.