Repository navigation
docs: record the grading profiles and sweep limits - #732
Conversation
Write down the declared limits of both grading profiles in scripts/verify_corpus.py (strict, and trail / production for strategies whose strategy.exit uses a trail_* parameter), how the 90th-percentile gates are computed, that an excellent grade needs exact trade-count parity, how each probe's profile is selected, and how many of the 7,989 sweep probes each profile applies to. Documentation only; no code or test change. Co-Authored-By: Claude Opus 5.5 <[email protected]>
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The production profile is selected by a trail_points, trail_offset or trail_price assignment in the strategy source (outside comments), and the entry gate is graded at the 90th percentile too. Co-Authored-By: Claude Opus 5.5 <[email protected]>
![graphify-labs[bot]](https://avatars.githubusercontent.com/in/4356270?s=60&v=4){kind=small}
There was a problem hiding this comment.
Graphify reviewed this change.
No issues found by static checks — no coupling regressions or blocking issues in the code graph. That is not the same as safe to merge: see what was not checked.
Not checked: tests were not run; no formal proof of the changed code.
Graphify review — findings
Documents how scripts/verify_corpus.py grades parity, adding a "Grading profiles and sweep limits" section to docs/coverage.md with a strict vs. trail/production limits table. It spells out that excellent requires exact trade-count parity (the 1.0% count limit doesn't excuse missing or extra trades), that entry/exit/P&L gates are per-strategy p90 values strictly below the limit, and how p90 is interpolated. It also explains how the sweep picks a profile: detect_parity_profile selects production when a trail_points, trail_offset or trail_price assignment survives comment stripping, otherwise strict, with an optional parity_profile override in inputs.json for the standalone verifier. The doc states the sweep splits 341 trail/production probes from 7,648 strict ones out of 7,989, and the preceding paragraph now says the strict profile "declares limits" rather than "gates".
No blocking issues surfaced.
Analysis details — impact, health, verification
Impact & health
Graphify review
Impact — 42 functions depend on the 42 functions this change touches.
Health — grade A; no new coupling hotspots.
Verification — 42 functions in the blast radius were not formally verified this run (proofs are advisory here).
Gate & verification
graphify gate
PASS — objectively clean (no health regressions, tests not run — proofs not run this pass (advisory)). Grounded, not self-assessed.
Advisory (not blocking):
- verification_scope: 42 function(s) in the blast radius were not formally verified this run
Summary
Records, in
docs/coverage.md, the grading profiles and limits that the corpus sweep applies:strategy.exitsets atrail_*parameter), as a table;excellentgrade needs exact trade-count parity, and that the entry, exit and P&L gates compare each strategy's 90th-percentile delta strictly below its limit (with how the 90th percentile is computed);detect_parity_profile, plus theparity_profileoverride of the standalone verifier);Refs #664.
Documentation only: no code, test or behaviour change.
🤖 Generated with Claude Code