fix(prediction): bind execution reviews to token side

[proerror77]

Change contract

Every RepricingIcReport and FillabilityReviewReport metric row is computed from exactly one ReviewSide sample set, with per-side denominators and per-side top_n, so pooled Up/Down evidence cannot hide opposite token behavior.

Out of scope

• FullDepthExecutionMatrixReport and slippage math, which are already side-keyed
• liquidity gates, AutoFactor/AlphaSearch, settlement verdicts, or governed keep/discard feedback
• 15m/1h or ETH missions, full-update ticks, collector/ECS/runtime/deployment changes
• pooled DataHealthReport, which remains ingest metadata rather than an evaluator metric

Dependency or merge order

None. This is based directly on main e2d741a; PR #145 is already merged.

Focused validation

• Counterexample: Up IC is positive while Down IC is negative; both survive as separate rows instead of cancelling.
• Counterexample: Up is fillable while Down is unfillable; decisions and coverage use each side's own denominator.
• Counterexample: top_n=1 still emits real Up and Down data rows.
• rustfmt --edition 2021 --check passed for factors_v2.rs.
• Both mission JSON files pass jq; git diff --check passed.
• Two independent code/test reviews reported no findings.
• The 36-source policy identity and 12 checked-in dependency inputs were independently recomputed. Both mission templates are pinned to sha256:12579dd6aaf0e34e26faa25f72ffe30afc88bbb08cb494321560f9ec4b28b494.
• Cargo compilation and tests are intentionally delegated to GitHub cloud CI.

Rollout or rollback impact

The repricing and fillability text/CSV report schemas add a side column and side-specific sections. Repository-local consumers have no stale struct constructors; external report parsers must accept the new schema. No live trading or deployment authority changes. Rollback is a revert of this PR, including the two mechanical mission policy-hash pins.

Summary by CodeRabbit

• New Features

  • Repricing analysis now reports results separately for upward and downward market sides.
  • Fillability reports now provide side-specific statistics and clearly labeled sections.
  • CSV exports include the market side for easier filtering and comparison.

• Configuration

  • Updated search policy snapshots for Bitcoin and Solana five-minute market research configurations.

• Tests

  • Added and updated coverage for side-specific repricing and fillability reporting.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 4ee8daef-5c5f-4094-a6f6-c1a2db45d565

📥 Commits

Reviewing files that changed from the base of the PR and between e2d741a and 956ff2a.

📒 Files selected for processing (3)

• rust_hft/prediction-markets/config/research_missions/polymarket-btc-5m.example.json
• rust_hft/prediction-markets/config/research_missions/polymarket-sol-5m.example.json
• rust_hft/prediction-markets/crates/ploy-research/src/factors_v2.rs

---

📝 Walkthrough

Walkthrough

Repricing IC and fillability reports now calculate and render results separately for up and down sides. Their public rows include ReviewSide, with tests covering side-specific output. BTC and SOL mission templates reference updated search policy snapshots.

Changes

Side-aware research reporting

Layer / File(s)	Summary
Side-aware repricing IC rust_hft/prediction-markets/crates/ploy-research/src/factors_v2.rs	Repricing IC rows and scoring are generated per side and target, formatted into side-specific sections, and covered by updated tests.
Side-aware fillability reporting rust_hft/prediction-markets/crates/ploy-research/src/factors_v2.rs	Fillability rows carry side at the top level, use side-specific denominators, remove side as a bucket dimension, and emit separate CSV sections with side-aware tests.

Mission snapshot updates

Layer / File(s)	Summary
Update mission snapshot identifiers rust_hft/prediction-markets/config/research_missions/polymarket-*-5m.example.json	BTC and SOL example missions reference new search_policy_snapshot_id SHA-256 values.

Estimated code review effort: 4 (Complex) | ~45 minutes

Possibly related PRs

• proerror77/monday#76: Updates the same Polymarket mission template snapshot identifiers.
• proerror77/monday#145: Updates the same mission template snapshot fields while changing a different research pipeline.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title matches the main change: making review metrics side-aware.
Description check	✅ Passed	The description covers the required sections and is detailed enough, with only minor heading/section-template differences.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch codex/polymarket-side-bound-execution-evaluators

---

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.}