Why versioning matters
Compliance decisions need to be reproducible. If a fund was screened on a given date under a given set of rules, the same inputs must produce the same outputs months or years later — for audit, for reporting, and for trust. Every compliance-bearing response identifies the exact methodology used, both in the resource and inmeta.methodology:
Version format
- methodology_id: the screening methodology,
aaoifi-ss21for an operating company;aaoifi-ss57,aaoifi-ss20orhalal-sh-etf-taxonomyfor a pooled vehicle decided on its structure (see Pooled vehicles) - year: calendar year of the implementation
- revision: incremental revision within the year
id and version are also returned as separate fields by the
methodologies endpoints and inside
every Evidence Packet.
What triggers a new version
Extraction improvements don’t bump the version: the screening logic hasn’t
changed, only the quality of the data fed into it. (Improved data can still
change a determination — the version tells you the rules were the same.)
Inspecting the methodology
The methodology, its thresholds, and its screen definitions are available directly:Every response uses the latest version of the methodology it names, and a
response names exactly one. An operating company is always
aaoifi-ss21; the
structural ids appear only on the pooled vehicles the compliance endpoint
serves as stocks. Pinned versions, if we add them, will be additive: your
existing requests will not change.Reproducibility
Given the same three inputs:- Methodology version —
meta.methodology(e.g.aaoifi-ss21@2026.1) - Filing set — identified by
filing.hash - Market-data snapshot —
as_of
explanation prose may vary, but the quantitative outputs are
deterministic. For a structural methodology the second input is the issuer
document set behind filing.hash, and the reproducible output is the checklist
in screens.structural: the same criteria, verdicts and quoted evidence.
