Use this plan to preserve the full experiment history while keeping main focused on the final
project narrative.
| Branch | Commit here |
|---|---|
archive-experiments |
Dated experiment logs, experimental configs, failed-gate research modules, and diagnostics |
main |
Final-facing README/docs, reusable pipeline code, tests, and concise current-status summaries |
First archive the full experimental state:
git switch -c archive-experiments
git add configs/experiments docs/experiments docs/research src/experiments src/data src/portfolio tests
git add README.md docs/README.md docs/PROJECT_STATUS.md docs/FINAL_VERSION.md docs/GITHUB_PUSH_PLAN.md docs/METHODOLOGY_EQUATIONS.md docs/research
git commit -m "Archive experimental research evidence"
git push -u origin archive-experimentsThen return to main and commit only the finalized project-facing pieces:
git switch main
git status --short
git add README.md docs/README.md docs/PROJECT_STATUS.md docs/FINAL_VERSION.md docs/GITHUB_PUSH_PLAN.md docs/METHODOLOGY_EQUATIONS.md
git add src/experiments/rank_model_diagnostics.py tests/test_rank_model_diagnostics.py
git add src/experiments/residual_rank_v2.py src/cli.py
git commit -m "Finalize ML rank selection documentation and diagnostics"
git push origin mainIf src/cli.py includes experimental commands from the archive branch, use git add -p src/cli.py
on main and stage only the final-facing rank-diagnostic command. Keep survival-filter and
smart-universe experiment commands on archive-experiments unless you intentionally want those
research utilities visible from main.
Before committing on main, check that generated files under data/ and reports/ are not
staged. They should remain local artifacts unless a small CSV or JSON is intentionally part of a
documented reproducibility record.
- If
archive-experimentsalready exists locally, usegit switch archive-experimentsinstead ofgit switch -c archive-experiments. - If Git says files would be overwritten when switching branches, stop and inspect
git status. - The repaired rank-40 evidence is promising development evidence, not a final alpha claim.