Recorded boxes and packs
What a randomised purchase actually returns depends on three things a product page rarely puts together: the published odds, a value for each outcome that someone other than the seller has recorded, and what it costs to turn that outcome back into money.
Every record here is a dated observation of a seller’s own published table, not a summary of one. Odds tables are revised without announcement, and the version that was published last month cannot be read after it has been replaced — so each observation is kept rather than overwritten, and the difference between two of them is shown.
Nothing is recorded yet
This section holds no records. The schema, the arithmetic, the change history and the pages are built and tested; what is missing is readable published odds from a seller, captured on a date, together with prices for the items involved observed somewhere the seller does not control.
Both halves are required and neither is optional. A published odds table with no independent prices supports no expected value, because the only values available would be the seller’s own. Independent prices with no published odds support nothing at all, because there is no distribution to average over. Where either is missing, no figure is published rather than an estimated one.
The calculator does not wait for any of that. It works on whatever odds table you supply, and it applies the identical arithmetic: price a box or pack yourself.
What is in scope here, and what is not
This section covers randomised purchases made with money by adults, of goods with a resale market: mystery boxes, sealed collectible packs and breaks, and digital cases whose contents can be traded on. The resale market is what makes the arithmetic possible at all, because it is the only source of a value the seller did not choose.
Randomised purchases inside games marketed to children are deliberately out of scope. They are a real subject and a serious one, and they are not this site’s: their items usually have no resale market, so there is no independently observable value to compute with, and the questions they raise are about who is being sold to rather than about what a purchase returns.
How a figure gets published, or does not
An expected value is a probability-weighted average. Averaging over part of a distribution does not produce a slightly uncertain figure — it produces a confident figure for a different distribution. So where a seller’s published probabilities do not add up to certainty, or an outcome carries a probability but no independently recorded price, no expected value appears on the page. What appears instead is which piece is missing.
Where the figures can be published, three of them are, side by side: what the purchase returns at the seller’s own stated values, what it returns at prices observed elsewhere, and what is left after the item has been sold back or shipped. The gap between the first and the last is usually the largest number on the page.
The vocabulary these pages use is defined in the glossary, and how each published figure is commonly misread is covered in reading the numbers.