What the official material supports
The publisher-provided Steam listing describes raids as high-risk runs where extraction determines whether the player keeps the loot. That supports treating “found” and “extracted” as different scenario outcomes. It does not provide this site with a reusable probability table for a boss, loose item, container, key, room, map, route, or case return.
Define the event before entering a probability
“Saw the item,” “put it in inventory,” and “successfully extracted it” are different events. This tool asks for P(find) and P(extract | find). Under the scenario you entered, their product isP(find and extract). The same target definition, raid eligibility rule, and observation window must be used for both inputs.
Probability and time formulas
If the combined extracted-target chance per eligible raid is p, the chance of at least one extracted target within n independent raids is 1 - (1-p)^n. Expected raids for one are 1/p; expected raids for k targets are k/p. Expected time multiplies that result by your measured full-cycle minutes. These are model outputs, not a guarantee about the next raid.
Use a full cycle, not time spent at one loot spot
Start timing before queueing and stop when the next comparable raid can begin. Include matching, loading, route time, combat or avoidance, extraction or death, healing, stash work, and re-queueing. A faster room check can still be a slower repeatable cycle when setup and recovery are omitted.
Why the value result is deliberately narrow
Target gross contribution is the extracted-target chance multiplied by the value you entered. The tool then subtracts the same direct cost from every raid. It does not price the rest of the route, estimate the value of experience or tasks, infer kit-loss probability, or fetch a live market. Use it only to compare the stated target under consistent assumptions. The related evidence-first EFT route guide provides a blank observation schema.