How we calculate values
Every number on this site is derived from observed Discord trade messages. No editor sets prices, no hand-tuned multipliers. The solve is pegged to Mythical Fruit Chest = 10,000 units, and every other item's value is worked out from its trade ratios to it. That 10,000 is the unit's reference point, not a fixed price: the later refinement steps, including the demand adjustment, move the chest's own listed value like any other item's.
Trades that players log on /log-trade (or with /logtrade in our Discord) are an extra input on top of the Discord ads. Each one is checked against screenshots of the in-game Past Trades list before it counts, and together they carry a small, capped weight: one player can move an item by only a few percent, logged trades can never outweigh the ads, and an item needs trades from more than one player before they count fully.
The pipeline
Step 1 — scrape
A bot reads every message from three opt-in Discord trading channels and saves timestamp, author id, and content. Around a million raw messages parsed to date.
Step 2 — parse and dedup
Emojis resolve to a 237-item dictionary; text matches against roughly 970 aliases (PCC, ICC, ASE, CQASE, PCHAK). Two dedup layers kill spam: a 60-second bump filter (same author reposting within a minute counts once) and a per-user-per-day cap (each author/message pair counts once per calendar day).
Step 3 — filter scams
Obvious lowball posts (tier letters two or more apart with no compensating mods, pure 1:1 cross-tier trades) are dropped before the math runs.
Step 4 — solve: chain plus bundle
Dijkstra ratio chains propagate values from the MFC anchor outward along the strongest trade paths. Multi-item bundle trades then refine: for each item, we take the weighted percentile of the implied values from bundles where it's alone on one side.
Step 5 — time decay and damping
Each trade carries two weights: a 12-hour half-life (short window, picks up fresh market shifts) and a 30-day half-life (stable baseline). Per item we pick whichever is more trustworthy right now. Iterative updates use 35% EMA blending with a 30% per-round step cap to prevent oscillation.
Step 6 — adjustments
Demand boost or discount of ±3–20% based on the wanted-versus-offered ratio. Scarcity-aware 85th percentile for Collectables with thin bundle evidence. A learned +Adds value per trade-scale tier lets us include mod-bearing trades that were previously discarded. 1:1 alignment and a chain-outlier cap catch items stuck at bad Dijkstra estimates.
Step 7 — confidence
Items with fewer than 200 trades get a low-confidence badge. Items with 200–999 trades are medium confidence. 1,000+ trades is high confidence. The bands are purely trade-count driven, with no subjective override.
How these values compare to the Watashi Ruby value list and the Trade Hub tier list
Two other GPO value references are widely used: the Watashi Ruby GPO value list, a community-maintained list published through its Discord server, and the GPO Trade Hub trading tier list, a hand-curated tier document. Both are opinion-based in the sense that people decide the numbers or tiers and update them when they choose to.
gpovalues.com works the other way round: every value here is solved from 125,990 logged Discord trades and re-solved every day, so a value only moves when players actually trade differently. That is why the two can disagree, especially in the days after an update when a new item's price is still settling, or for thin-traded items where our confidence badge is low. Neither source copies the other; if you trade by a community list, this site shows what the market has actually been doing, and the GPO trading tier list shows how the two approaches line up item by item.
What's intentionally not here
- No editor opinions. No "we think X is worth Y."
- No values copied from other lists. Our math stands or falls on the trade data alone (see how that differs from community lists below).
- No fake confidence. Thin-data items are flagged prominently.
- No paywalls, no premium values, no "pro tier" unlocks. Every value is free forever.
Known blind spots
- Exclusives (Marine Cap, Marine Cape, Cool Shades) trade mostly via DM or invite-only channels. Our public-channel parser systematically under-samples these; their values here are a floor, not a market-clearing price.
- Image-only trades (screenshots, videos) are not parsed. Text trades only.
- Very new items (under 30 days since release) have volatile values; expect ±30% swings.
- Mods beyond
+Addsand+Strong_Addsaren't priced.
Source code and transparency
If you think a value is wrong, tell us in Discord and we'll show you which trades produced it. We're working toward a community-submission path with evidence review.