Repository navigation
Add selected opening-hour walk-forward research - #33
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 612cd59551
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if manifest["as_of"] != args.session_date: | ||
| raise SelectedOutcomesBlockedError( | ||
| "manifest.as_of must equal session_date" | ||
| ) |
There was a problem hiding this comment.
When a valid manifest has analysis_date == session_date, this check passes because normalization permits equality, so the screen can contain that session's closing data while the script compares it with the same session's opening-hour bars. That introduces direct look-ahead into the advertised point-in-time outcomes; require manifest["analysis_date"] < args.session_date, as the existing opening-hour collector does.
Useful? React with 👍 / 👎.
| rows = { | ||
| row.day: row | ||
| for row in qta1.read_csv(instrument["benchmark_csv"]) | ||
| } |
There was a problem hiding this comment.
If the benchmark CSV is updated or revised in place after the frozen manifest was created, this reads the changed file without comparing it to benchmark_csv_sha256; the resulting regime and pass/fail summaries therefore silently combine a frozen screen with untracked data. Validate the file against the manifest hash before reading it so the outcome remains reproducible and bound to its declared inputs.
Useful? React with 👍 / 👎.
| value = evaluation.normalized_screen(path) | ||
| method = value["method_version"] |
There was a problem hiding this comment.
For a valid qta-screen/v2 artifact whose screen_status is BLOCKED, screen_universe.py deliberately emits empty selected lists, but this path accepts the artifact and eventually reports the outcomes artifact as READY with zero candidates. Check screen_status == "READY" before consuming selections so failed screening is not converted into a successful research result.
Useful? React with 👍 / 👎.
1 participant
Summary
Validation