What is the GSA PATTERN library?
The PATTERN library is the set of numbered methodology rules that governs how Gold Silver Analytics computes every analysis in its coverage universe — each one stating a required procedure, the patterns it prohibits, and the conditions under which it stops applying.
Consistency across a large coverage universe is a process problem, not an analytical one. A single analyst covering a handful of companies can hold a method in their head; across roughly 300 companies re-derived on a monthly cycle, an unwritten method drifts, and once it drifts the peer comparisons built on top of it stop meaning anything. A peer-relative percentile is only informative if every company in the peer group was scored under the same framework in the same cycle.
So the method is written down as numbered rules, versioned, and loaded into every analysis run.
Each rule states three things: what procedure is required, what specific patterns are prohibited, and where the rule stops applying. The third is not decoration — a rule without a stated boundary gets applied by intuition, which is the condition it was written to replace.
The library also functions as a record. When a rule turns out to be incomplete it is amended and the amendment is dated, so the library carries the history of what the methodology has been wrong about alongside what it currently holds.
What the rules cover
- Cost derivation — how all-in sustaining cost behaves under price-variable royalties and streams, and how gold-equivalent reporting distorts margin analysis when the scenario price has run ahead of the reporting period.
- Structural classification — when a company with both a producing asset and a development asset should be treated as a hybrid rather than a developer, and what share of consolidated valuation the producing asset must carry before the classification changes.
- Source discipline — which sources are authoritative for which inputs, and which are never sufficient to confirm that something is absent.
- Completeness — the mandatory scans that must be explicitly resolved rather than skipped, including the balance-sheet and capital-structure scan where every item has to be answered either with terms and a source reference or with an explicit “scanned, not found”.
The prime directive
One rule outranks the rest. PATTERN-031 prohibits data locks: a rule may direct how to compute and where to source, and may never pre-supply a numerical answer the analysis would otherwise derive for itself. It is what makes monthly re-derivation meaningful rather than ceremonial, and it is enforced as a binary gate — any unresolved finding fails the cycle.
Related methodology
Ask your AI assistant
Prefer to ask an AI? GSA is now a connector your assistant can query directly — any covered miner's Re-Rate Score and factor breakdown, or a screen of the universe by metal, jurisdiction, and stage. Analytical classification, cited, with a link back.