Methodology
This page explains how BISPath builds its BIS guides and meta gear statistics. The goal is to make sure you understand the sampling method and limits of any data before trusting it at face value.
1. Data Sources
BISPath pulls data from two places. Game metadata (item info, class and spec definitions, stat keys, item set bonuses) comes from Blizzard Entertainment's official API. We use the profile-{region} namespace for dynamic data and static-{region} for static data. Character syncing is performed via the user's own OAuth login token and only touches characters on that account. Top-player gear statistics come from the WarcraftLogs GraphQL API.
2. Sampling Method
Meta gear data is collected by calling the first ranking page on WarcraftLogs for every raid boss in the current tier plus a representative Mythic+ encounter. WarcraftLogs returns 100 ranking entries per page. When the same character appears across multiple bosses, we deduplicate by the (character name, server, region) tuple and keep only the highest-ranked entry. From that deduplicated sample, we extract the actual gear, enchants, and gems each character was wearing and aggregate the distribution per slot. Stat priority distributions are computed by a separate analyzer over a subset of the same sample. We want to be explicit that the current Mythic+ sample is built from a single representative encounter — expanding to a multi-encounter Mythic+ sample is on the roadmap.
3. Collection Schedule
Data collection runs automatically via a GitHub Actions cron workflow every day at 00:03 KST (15:03 UTC the previous day). Each run refreshes raid data, Mythic+ data, and stat priority distributions, and stores the new sample size and last-collected timestamp per spec. As more logs accumulate over the week, the sample stabilizes — but right after a patch, the sample can be small, so caution is warranted.
4. Editorial Workflow
The narrative portion of BIS guides is written by humans, not auto-generated. The process is: (1) an automated fact sheet pulls the BIS item list, trinket tiers, stat distribution, and sample size; (2) an editor combines that sheet with their game knowledge to write the gear philosophy, stat priority rationale, exception cases, and FAQ for each spec; (3) the result is reviewed and committed with a lastReviewed date. Specs without editor notes show only the data table and are excluded from search engine indexing.
5. Known Limits
The limits of this approach are explicit. First, the WarcraftLogs top-ranking sample is biased toward already-geared top players, so it can diverge from the optimal choices for beginners or mid-tier players. Second, gear that's strong on a specific boss or content type (raid vs Mythic+) can be weak elsewhere, and that difference doesn't always show up in the aggregate. Third, right after a new patch or hotfix, the sample reflects the old meta until enough new logs accumulate. Fourth, collection is based primarily on the KR region, not a global average.
6. Changes and Corrections
When the methodology or collection code changes, this page is updated first. Factual errors get corrected as fast as we can verify them, and each data record's last-updated timestamp lets you see how fresh the underlying sample is. Send error reports or methodology suggestions to bispath.gg@gmail.com.