v0.2.0 - Public Validation and Contributor Workflow

v0.2.0 - Public Validation and Contributor Workflow

v0.2.0 turns the first public research baseline into a more reviewable
open-source workflow. The release adds public validation reports, a Baostock
data path, stronger contributor guidance, and clearer limitations around
public-data results.

Highlights

• Added Baostock public-data validation support and an A-share validation report.
• Added synthetic validation as a Windows CPU reproducibility baseline.
• Documented yfinance HTTP 429 rate-limit behavior and a smoke-test workflow.
• Added an English factor handbook for the 213-factor research surface.
• Added Chinese README summaries for simplified and traditional Chinese readers.
• Hardened neutralization and Baostock handling with additional tests.
• Added contributor pairing guidance for validation and benchmark reports.
• Added SECURITY.md for vulnerability reporting.

Validation Position

This is still a research and education toolkit, not a trading-signal release.
The public reports are useful because they make assumptions, costs, data-source
limits, and baseline behavior easier to inspect.

The current public results do not claim deployable alpha. They mainly show:

• how the validation harness behaves across synthetic and public data;
• how turnover and transaction costs affect interpretation;
• why simple baselines are important controls;
• where contributors can add more reproducible reports.

See docs/validation_digest_20260720.md for
the release validation summary.

Useful Entry Points

• Start here: docs/start_here.md
• Public-data validation: docs/public_data_validation.md
• Validation digest: docs/validation_digest_20260720.md
• Benchmark board: docs/benchmark_board.md
• Reality check: docs/reality_check.md
• Pairing issue: #22

Compatibility Notes

• Package version is now 0.2.0.
• The project remains Python 3.9+ and PyTorch based.
• The paper remains the research snapshot; main continues to evolve as a
  living open-source toolkit.