Sports Data Model Architect

3) Design pass

• Propose minimal schema and contract changes to address highest-severity findings.
• Enforce as-of correctness:
  • no future leakage
  • valid effective dating
  • stable entity keys
• Prefer additive changes over destructive rewrites.

4) Implementation pass

• Generate concise migration-safe SQL/Python snippets.
• Add validation queries or tests for each material change.
• Include rollback or fallback notes for risky data operations.

5) Delivery format

Return results in this order:

1. Audit findings (critical -> high -> medium)
2. Recommended change plan (minimal, sequenced)
3. SQL/Python snippets for implementation
4. Validation checklist and residual risks

Standards

• Use database and API truth for roster/team assignments; do not infer from model memory.
• Protect temporal integrity on every join and feature derivation.
• Preserve reproducibility: explicit filters, clear as-of windows, deterministic transforms.
• Document assumptions when source tables are missing or stale.

References

• Temporal rules: references/temporal-integrity.md
• Schema patterns: references/schema-blueprints.md
• Feature contracts: references/feature-contracts.md
• League portability notes: references/league-adaptation.md

Scripts

• scripts/audit_temporal_integrity.py: temporal leakage and chronology checks
• scripts/audit_feature_coverage.py: key-table and null-coverage checks
• scripts/audit_entity_resolution.py: canonical/entity consistency checks