v0.2.1 - Validation Entrypoints and Outreach Follow-through

v0.2.1 - Validation Entrypoints and Outreach Follow-through

v0.2.1 is a patch release for the public validation funnel. It fixes a Colab
entrypoint, records the current visibility surface, and points new contributors
toward reproducible benchmark and public-data reports.

This is not a new alpha claim and not a production-trading release. It is a
small maintenance release after the v0.2.0 public validation workflow.

Highlights

• Fixed the Baostock Colab demo bootstrap so fresh Colab runtimes clone
  https://github.com/initial-d/ml-quant-trading.git.
• Added the Colab Baostock demo to the tracked live entry points.
• Added a follow-up validation digest for the current reproduction surface.
• Recorded the verified Hugging Face paper entry and the current public
  discovery channels.
• Consolidated outreach follow-up into Discussions #13 so benchmark and
  reproduction notes have one landing place.

Validation Position

The useful contribution in this release is operational clarity:

• public-data failures should be reported as blockers;
• synthetic runs remain plumbing checks, not signal evidence;
• Colab and Baostock are intended to make the A-share route easier to try;
• external discussions are used to collect validation patterns, not to claim
  that the project has deployable alpha.

See docs/validation_digest_20260727.md
for the follow-up validation and visibility summary.

Useful Entry Points

• Start here: docs/start_here.md
• Public-data validation: docs/public_data_validation.md
• Validation digest: docs/validation_digest_20260727.md
• Colab Baostock demo: https://colab.research.google.com/github/initial-d/ml-quant-trading/blob/v0.2.1/demo_baostock.ipynb
• Benchmark discussion: #13
• Pairing issue: #22

Compatibility Notes

• Package version is now 0.2.1.
• Python and PyTorch requirements are unchanged from v0.2.0.
• Existing validation commands and report formats remain compatible.