Integration: Race Outlook prediction stack#270
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: d8a7b05839
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| for (let r = 0; r < MAX_FUTURE_ROUNDS; r++) { | ||
| const outcomes = stats.map((s) => sampleRoundOutcome(s, rng)); |
There was a problem hiding this comment.
When the mechanics model is active, each player's blitzed flag is sampled independently and the simulation immediately scores that vector. For games where each player has a low or mid blitz rate, a large share of simulated rounds can have no blitzPileRemaining === 0, even though validateGameRules rejects such score sets and real rounds only end once someone blitzes. Those invalid rounds still advance scores and can skew win and next-round probabilities; resample or condition the outcomes so at least one player blitzes.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Pull request overview
This integration PR stacks the full “Race Outlook” prediction feature set: a richer Monte Carlo race forecast, optional permission-gated historical player profiles, a mechanics-aware sampling layer for Dutch Blitz, and the corresponding UI updates/tests to exercise the end-to-end experience.
Changes:
- Added a permission-gated server query to compute per-player historical prediction profiles from prior finished games in the same circle.
- Expanded the probability model into a coherent “race forecast” (win odds + next-round threats + finish ranges + unresolved outcomes), with optional history blending and mechanics-aware sampling.
- Updated scoring UI to pass history/mechanics inputs through and render the new Race Outlook card, plus added unit/component tests.
Reviewed changes
Copilot reviewed 11 out of 11 changed files in this pull request and generated 3 comments.
Show a summary per file
| File | Description |
|---|---|
| src/server/queries/predictionProfiles.ts | New server-only query + profile builder for permission-gated historical samples. |
| src/server/queries/index.ts | Re-export prediction profile query/builder and types. |
| src/server/queries.ts | Re-export getPredictionProfilesForGame from the server query barrel. |
| src/server/tests/predictionProfiles.test.ts | Unit tests for gating, scoping, caps, and profile math. |
| src/lib/scoring/probability.ts | New Race Outlook forecast model: unresolved outcomes, history blending, mechanics sampling, and helpers. |
| src/lib/tests/probability.test.ts | Model tests for determinism, clamping, percent allocation, history/mechanics behavior. |
| src/components/scoring/ScoringShell.tsx | Thread predictionProfiles into the scoring shell and down to the between-rounds view. |
| src/components/scoring/graphs/WinProbabilityCard.tsx | Replace projected finish with Race Outlook UI backed by calcRaceForecast. |
| src/components/scoring/BetweenRoundsView.tsx | Derive per-round mechanics samples and pass them + profiles into the forecast card. |
| src/components/tests/scoring/WinProbabilityCard.test.tsx | Component test coverage for Race Outlook rendering and copy. |
| src/app/games/[id]/page.tsx | Fetch and pass prediction profiles for circle members only. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| export function allocateOutcomePercents( | ||
| counts: [string, number][], | ||
| total: number | ||
| ): Record<string, number> { | ||
| const exact = counts.map(([id, count]) => ({ | ||
| id, | ||
| value: total === 0 ? 0 : (count / total) * 100, | ||
| })); |
| // If every player has non-positive mean, no one is progressing | ||
| if (stats.every((s) => s.mean <= 0)) { | ||
| return Object.fromEntries(players.map((p) => [p.id, 0])); | ||
| return { | ||
| players: Object.fromEntries( | ||
| players.map((p) => [ |
| const predictionProfiles = isCircleMember | ||
| ? await getPredictionProfilesForGame(params.id) | ||
| : {}; |
Summary
This integration PR contains the full Race Outlook prediction stack for end-to-end testing:
Use this branch/PR to test the full experience together. The smaller PRs remain useful as review slices.
Verification already run on the stacked tip
npm run lintnpm testRESEND_API_KEY=re_dummy npx next buildNote
I created this from the top of the stack (
mwickett/dutch-blitz-mechanics-forecast) so it contains every commit in the series.