How we price
The whole pipeline, start to finish: how a real completed sale becomes the number on a card page — and every point where we refuse to print one. This is the narrative version of the methodology; nothing here is aspirational, it is how the site actually works.
1. Only completed sales enter
We collect completed-sale data from eBay US and Japanese marketplaces (currently eBay US, Mercari JP, Yahoo! Auctions JP) — transactions that actually settled. Asking prices never become prices here: a card listed at $200 that never sells at $200 is worth whatever it last sold for. Where we show asks at all, they appear as a clearly separated "what sellers are asking" strip for context — useful on a card nobody has resold yet — never as the headline.
2. Every sale keeps its receipt
Each sale is stored with a link back to the source listing. That link survives all the way to the page: click any dot on a price chart or any row in a sales table and you land on the underlying receipt. If our median disagrees with what you see in the market, the evidence is one click away — that is the fastest way to catch us being wrong, and we want it used.
3. Japanese sales are converted at a dated rate
JP marketplace sales are priced in yen and converted at the European Central Bank reference rate (via the Frankfurter ECB feed) — currently 160.24 JPY/USD, dated 2026-07-31. Every USD figure derived from a yen sale rests on the native amount plus that dated rate — re-priceable, never a hidden constant.
4. The median, in the freshest window that has depth
A card's headline is the median of completed sales in the freshest time window with enough of them, per market and printing. Medians resist the one weird sale; the filtering (outlier fences, dedup, spike quarantine, retraction voiding) is spelled out in the methodology.
5. Confidence is sample depth, and it is never hidden
- HIGH — 10 or more completed sales. Shown as a plain price.
- MEDIUM — 4–9 completed sales. Shown as a plain price.
- LOW — 1–3 sales. Never shown as a bare number, anywhere — not in tables, search, share images or the API. It renders only as "~$X — rough number, only a few sales".
- NONE — no verified sales. The page says "no verified solds — tracking live" rather than inventing a figure.
That never-bare rule is enforced in code: the build fails rather than render a thin price as a plain number. Recency is a separate FRESH/STALE axis — a deeply-sampled card can still be stale, and you see both facts, never a blended score.
6. The snapshot has a freshness contract
The site rebuilds from a fresh data snapshot on a daily sweep. If a snapshot is ever older than 26 hours — one daily cycle plus slack — the build refuses it, and every page carries a stale banner rather than quietly serving yesterday's market as today's. The current snapshot's age and per-source freshness are public on the data status page.
7. When we get it wrong, we say so in public
Corrections to published numbers and methodology changes that moved a price are logged, dated, on the corrections page — including the calls we got wrong before we got them right. The honest audit of coverage and everything we refuse to print lives at how accurate are we?. Spotted a bad price? contact@palworldprices.com reaches a person.
Companion guides: the price guide · spotting bootlegs.