Corrections
The data-accuracy changelog. Every correction or methodology change that moved a published number, plus every near-miss we caught before it did — dated, newest first, including the calls we got wrong before we got them right. Entries are appended, never rewritten: undoing a call means adding another entry, not erasing one.
7 dated entries, hand-curated. If it's here, it changed a number you might have read — or came close enough that you deserve to hear about it. Site changes go in the changelog feed (see RSS feeds), the rules themselves in how we price, and what they cover in how accurate are we?
Audit note: a four-box lot, caught before it reached a published price
- This one is a near-miss, not a correction. No price you have read on this site was ever wrong because of it, and no published figure changed when we fixed it. We are publishing it anyway. A site that grades itself in public owes you the misses it caught as well as the ones it did not.
- What could have gone wrong: for part of the launch window, our record of US eBay sealed sales kept a tidied-up product name in place of the seller's own title. The check that sets multi-item lots aside reads titles to do its job, so it had nothing left to read — and a listing selling four booster boxes for $596 was recorded as one box at $596.
- How we found it: we went back and recovered the seller's real title for every sealed sale already on record — 273 of 332 from copies we had kept separately; the remaining 59 are marked unknown rather than guessed at — and re-ran the lot check across all of them. Exactly one sale turned out to be a genuine multi-item lot. Every other title that looked like a lot was a single unit describing its own contents: "Booster Box (12 Packs)" is one box, and "12x Booster Box Sealed Case" is one case.
- How close it came: that sale had not reached a published median yet. A lone price far from its neighbours is held out of the figures for seven days before it can count, and that week was nearly up — around 2026-08-08 it would have entered the 30-day and 90-day booster-box windows and pushed the published price range from a top of $288.06 up to $596. It is now set aside for good, with the seller's real title recorded as the reason.
- What changed underneath: setting a sale aside no longer means editing the sales record. A sale we recorded wrongly is never rewritten and never deleted — it is cancelled out by a dated entry that names it, gives the reason and carries the evidence, so the original stays visible and the reversal has to be written down too. Undoing a call means adding another entry, never erasing one. We also taught the lot check to read counts written as words — "FOUR (4) Booster Box", "Qty (2)" — which digit-only matching had missed. Neither change moves a number on the site today; both exist so this cannot happen twice.
Refusing to invent parallel printings we cannot attest
- A freshly captured listing titled "ETD02-012 SP Holo" would previously have created an "SP" parallel printing of a trial-deck card. No such printing exists — "SP" is a real rarity in the booster set, but that trial deck was only ever printed with its own TSP/TSR parallels.
- The card matcher now refuses any code-plus-rarity combination that is absent from the official card list. The sale is set aside as unresolved instead of being priced against a printing that does not exist — and instead of being silently folded into the standard printing, which would have polluted a real median.
US eBay completed-sale refresh; best-offer sales excluded
- Captured a fresh sweep of US eBay completed sales: 467 new single-card sales and 118 new sealed-product sales entered the record. US receipts are now current within 24 hours, and capture now runs daily through the launch window instead of in occasional batches.
- Methodology note, stated plainly: sales completed via "best offer" are excluded. eBay displays the original asking price on those, not the amount the buyer actually agreed to pay, so including them would overstate medians. Fewer samples, honest samples.
Eleven phantom "RR" printings removed
- What was wrong: when a seller writes the printed rarity next to the card code — "BP01-050 RR Double Rare" — our reader minted a separate "RR" parallel identity. No official Palworld card carries an RR, RRR, or SEC suffix; every one of those listings was the standard print, whose printed rarity simply is RR.
- What changed: eleven phantom card identities were deleted and 283 recorded sales were re-attributed to the printing their titles actually assert. Two of the affected cards exist only at a single non-standard rarity, and their rows were adjudicated individually rather than moved in bulk.
- Effect: sample counts and medians moved on the affected standard printings — deeper samples under honest identities.
Rarity-aware name matching
- What was wrong: a listing with no card code, like "Jormuntide Ignis SSP", matched by name to the standard printing — collapsing printings that differ by up to ~100× in value into a single price series.
- What changed: the matcher now reads printed-rarity words (SSP, SP, SR, OSR, TSP and their Japanese equivalents) and assigns the sale to that parallel printing when it officially exists, falling back to the standard printing only when no rarity is stated.
- Effect: high-rarity sales moved off standard-printing price series and onto the parallels they belong to; some thin-sample medians moved as a result.
Chillet soul card: a split we reversed within a day
- On 2026-07-31, Japanese listings describing a "summer" version of the Chillet soul card diverged in wording from the standard soul card, so we split it into its own identity and began pricing it separately.
- On 2026-08-01, the accumulated evidence said otherwise: same card. We reversed the split, merged the sales back into one series, and recomputed the median.
- This entry stays here on purpose. A price site that never admits a wrong call is asking to be trusted on faith; the correction record is the trust.
Yen conversion moved to the dated ECB reference rate
- What changed: every USD figure for a Japanese sale is now derived from the native yen amount and the European Central Bank reference rate, dated — the rate and its date travel with the data, so any converted price can be re-derived.
- What was wrong before: conversion used a fixed rate that was only as fresh as its last hand update and could drift from the actual market rate, skewing every JP-sourced dollar figure by the drift.