What is a bounded sample?
DeckBeacon's catalog is a bounded sample of Path of Legends battles — not a global census and not a seven-day history of Clash Royale.
A bounded sample is the honest name for DeckBeacon's data. The pipeline watches a finite set of Path of Legends leaderboard players, reads the battle logs the official API returns for them, and ranks decks that appear often enough in that window. It does not watch every match in Clash Royale. It does not keep a rolling seven-day chart.
This sentence is load-bearing. Search results, social screenshots, and language models all want to upgrade a sample into "the meta." The methodology page and llms.txt exist so that upgrade fails.
What bounds it
- Who — the Path of Legends leaderboard the API exposes (1,000 seeds in a full ranked crawl), not mid-ladder, not trophy road
- Which games — ranked 1v1 modes on the allowlist, not 2v2, not a published Grand Challenge tab until that sample clears the same floors
- How many lists — 200 ranked decks, not every distinct eight-card combination in the logs
- How recent — whatever the battle logs actually contain for those players. Top-ladder logs can reach back weeks. That is an API property, not a seven-day product promise
A deck that is stomping mid-ladder and never appears in those logs is invisible here. That is a limit, not a bug.
Why we will not "un-bound" it with someone else's chart
RoyaleAPI and Stats Royale publish other windows. Scraping them would make the grid look complete and make every provenance sentence on this site false. We do not do that. Is DeckBeacon made by Supercell? — unofficial, still not a pirate of a competitor.
How to read a number inside a sample
Wilson Rating and win rate describe decisive games in the sample. Usage is share of appearances in the sample. Compare two decks from the same snapshot; do not paste a number from another site beside them and call it a controlled experiment.