How prices and rankings are calculated

Where the data comes from

We read what players actually have for sale in their own stalls. Nothing here is user-submitted, and none of it is an official price from the game's operator. Each scan is stored as its own snapshot rather than overwriting the last one, because the sequence of snapshots is what becomes price history.

Each server is a separate economy

Loki and Odin are held as completely separate datasets. The same item can be worth many times more on one than the other, and some names are only sold on one of them at all.

So prices from two servers are never averaged together — a median drawn from two different worlds describes neither. An item page shows what the other server is asking, but as its own figure, clearly kept apart.

The market has two kinds of stall

The first kind carries an expiry timer and holds at most three items. It is a leased stall, and it keeps selling after its owner logs out. The second kind has no expiry and holds up to twelve items. It is a live cart with a character standing behind it, and it disappears the moment that player leaves.

The difference matters when you act on a price. A live cart's price may be gone before you walk over. A leased stall's price stands until its timer runs out.

Listed prices are per unit

Every price is for one unit, not for the stack. An item sold in bulk has to be multiplied out before you know what you will actually pay. Wherever the quantity is greater than one, the stack total is shown underneath the unit price.

Why two items with the same name differ by tens of times

Anything with "Option" in its name carries randomly rolled options. Two items whose names match character for character are not the same product and should not cost the same.

Because of that, we never show an average price for a bare name, and any "below median" label always has that listing's options next to it. A price far under everyone else's usually means the roll was bad, not that the seller is being generous.

Only compare items at the same refine level

The median is calculated among items at the same refine level. Pooling a +7 with unrefined copies makes every unrefined listing look absurdly underpriced when it is priced normally.

The "worth a look" formula

A scan is a single snapshot with no price history to compare against, so interest is derived from comparisons made inside the snapshot itself, at these weights:

• 34% — how far below the median for its own name and refine level, multiplied by how much that median can be trusted given how many stalls there are to compare. Under four stalls, it counts for nothing.
• 22% — absolute price, on a log scale so the single dearest listing does not drown everything else.
• 14% — refine level
• 12% — whether cards are slotted
• 8% — number of options
• 7% — rarity, from how many stalls carry the name
• 3% — stall about to expire

The result is then capped at one listing per name. That cap is load-bearing, not a nicety: without it the top of the board fills with copies of whichever item is both common and widely priced, because every cheap copy scores well at once.

Limitations worth knowing

• This is a snapshot at the stated time, not a live feed. Prices move.
• It covers only the market zone that was scanned, not the whole server. You will not find the words "cheapest on the server" anywhere on this site.
• The item names and card counts the scan produces do not always match the raw data. We decode the raw data ourselves rather than trusting the name.
• The scan mislabels several item types — bows tagged as pet armor, for one. We reclassify from the equipment slot instead.
• Anything the scan could not resolve into a real name (an "Unknown" entry, or one full of question marks) is dropped entirely rather than shown.