research: validate baseball and soccer context with guarded data pilots

[choigod1023]

Summary

• Add a bounded MLB pitch/SQLite pilot with boxscore completeness checks and unknown workload handling.
• Join observed K League xG and T-30 odds with timestamp provenance, freshness and latest cancellation guards.
• Run league-independent chronological score-form controls through the actual winner-market recommendation policy, with matched-count diagnostics.
• Export Markdown, compact results and a private immutable SQLite archive. Production promotion stays disabled.

Results

• 10 independent league controls: alpha zero in 9; EPL correction worsened outer evaluation. No production weight change justified.
• K League actual xG probe: 20 eligible games, market 8/20 versus xG 9/20; insufficient for learned-model or uplift validation.
• MLB: chronological 30 of 186 scheduled games, 8,768 pitches. Acquisition quality only, not accuracy lift.
• Explicit limits: reused historical sample, unknown archived odds timing, winner-market subset only, no all-league xG completion.

Validation

• Python: 797 tests and 58 subtests passed.
• Frontend: 229 tests passed.
• Isolated production build passed; no shared processes restarted.
• Research SQLite integrity and all seven saved artifact hashes verified.

No deployment, merge, production DB access or recurring job changes.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: 69a6c194-9096-4221-a938-d6dcd96ab58d

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}