4.0.0: Split off finance/trading code to alphapy-finance

[mrconway]

Summary

BREAKING. Removes all trading, markets, and Alfi code from alphapy-pro. That code now lives in the private `alphapy-finance` repo (initial import landed in https://github.com/ScottfreeLLC/alphapy-finance/pull/1).

See tag `v3.1.1-monolith` for the pre-split state. See full `CHANGELOG.md` 4.0.0 entry.

Removed

• `alphapy/{mflow_main,system,backtest,metalabel}.py`, `alphapy/{portfolio,indicators,data_sources}/`
• Market-data functions in `alphapy/data.py` (10 helpers)
• Top-level `app/` (Alfi) and `agent/` (legacy trading agent)
• 13 trading project examples
• Trading tests, scripts, docs, configs
• Entry points: `mflow`, `scott`, `research`
• ~30 dependencies (alpaca-py, vectorbt, fastapi, langchain*, yfinance, polygon-api-client, etc.)

Kept

• 16 ML-core modules in `alphapy/`
• 3 example projects: `kaggle`, `pizza`, `time-series`
• 26 ML-core dependencies
• Single entry point: `alphapy`

Test plan

• `pytest tests/` → 18 passed, 1 skipped
• `alphapy --help` works
• `alphapy.version` → `4.0.0`
• `rg "yfinance|alpaca|polygon|vectorbt" alphapy/` → 0 hits
• alphapy-finance test suite → 396 passed, 4 xfailed (verified pre-merge)
• Smoke-run kaggle (Titanic) project end-to-end
• Tag v4.0.0 after merge to main

[claude]

Claude finished @mrconway's task —— View job

---

PR Review: 4.0.0 Split Trading/Finance Code

Reviewed this major architectural change that removes all trading/finance functionality to create a pure ML framework.

Review Tasks

• Read key modified files and assess changes
• Check for lingering trading/finance references
• Review dependency cleanup in pyproject.toml
• Verify test coverage adequacy
• Review configuration file updates
• Assess documentation completeness
• Provide comprehensive feedback

Overall Assessment ⚠️

This is a well-executed major architectural change that successfully removes trading/finance code, but has several critical issues that must be addressed before merge.

✅ Strengths

Excellent Dependency Cleanup: pyproject.toml is properly cleaned - removed ~30 trading dependencies (alpaca-py, vectorbt, fastapi, langchain, yfinance, polygon-api-client) while retaining 26 ML-core dependencies.

Clean Documentation Updates: README.md, CHANGELOG.md, and CLAUDE.md are well-updated with clear migration guidance and architectural changes.

Proper Version Management: Version bumped to 4.0.0 with appropriate BREAKING CHANGE designation.

Good Test Structure: Remaining test suite is minimal but focused (6 test files vs previous hundreds).

🚨 Critical Issues

1. Broken Code in alphapy_main.py - ❌ BLOCKING

/home/runner/work/alphapy-pro/alphapy-pro/alphapy/alphapy_main.py:208-216

full_path = SSEP.join([alphapy_root, 'config', 'systems.yml'])
with open(full_path, 'r') as ymlfile:
    trading_systems = yaml.safe_load(ymlfile)
logger.info("Getting Trading Systems")
specs['systems'] = trading_systems

Problem: Code tries to load systems.yml file that was removed in the split. This will cause runtime failures.
Fix Required: Remove this entire section or make it conditional.

2. Stale Trading References - ⚠️ MEDIUM

Multiple files contain outdated references:

• CONTRIBUTING.md:26 - Still mentions mflow_main.py
• config/alphapy.yml.template - Contains mflow: section that should be removed
• config/variables.yml:95 - wdev: 'Portfolio Weight Deviation' variable (trading-specific)
• alphapy/variables.py:468 - Comment about "assign global frame for trading"
• scripts/, docs/ - Multiple files with trading references that should be cleaned up

3. Configuration Inconsistencies - ⚠️ MEDIUM

• config/alphapy.yml still contains hardcoded paths (/Users/markconway/Projects/)
• Missing proper template configuration for ML-only workflow

📊 Security & Performance

• ✅ Security: No security issues identified
• ✅ Performance: Dependency reduction should improve load times
• ✅ No sensitive data: Clean removal of API keys/credentials

🧪 Test Coverage

Adequate for scope: 6 remaining test files focused on core ML functionality. However:

• ⚠️ Missing integration test for the incomplete smoke test mentioned in PR description
• ⚠️ Test fixtures in conftest.py still contain trading-related mocks (Alpaca, Polygon) that should be removed

📝 Documentation Quality

Generally excellent, but needs cleanup:

• ✅ Clear migration guidance in CHANGELOG.md
• ✅ Updated README.md with new architecture
• ⚠️ Many doc files in docs/ still reference mflow command and trading functionality

Recommendations

Must Fix Before Merge

1. Remove systems.yml loading code in alphapy_main.py:208-216
2. Clean remaining trading references in key config files
3. Update config templates to reflect ML-only architecture

Should Fix Soon

1. Remove trading-related test fixtures in conftest.py
2. Clean up documentation references to removed functionality
3. Complete the smoke test mentioned in PR test plan

Migration Strategy

The migration guidance is excellent - downstream repos using { path = "../alphapy-finance", editable = true } is the right approach.

Verdict

Conditional Approval - This is excellent architectural work that properly separates concerns, but the broken code in alphapy_main.py is a blocking issue that will cause runtime failures. Fix the critical issues and this will be a solid 4.0.0 release.