nascarprob

NASCAR Cup Series race simulator: win and finishing odds, head-to-heads, parlays priced inside the simulation, DFS projections with an exact lineup optimizer, race-day weather with a written briefing, best plays, live running order and an honest accuracy record. Same stack and design kit as golfprob, nflprob and hrprob: Vercel serverless functions, optional Supabase Postgres, one app.html.

Deploy

1. Import the repo in Vercel (framework preset: Other). No build step.
2. Optional but recommended: set DATABASE_URL (or POSTGRES_URL, which Vercel's Supabase integration sets) to a Postgres connection string. In Supabase: Project Settings → Database → Connection string → URI, transaction pooler (port 6543). Without it the model runs on the last 40 races fetched live.
3. Backfill history (with a database): open /api/sync?season=2022, repeat until remaining is 0, then 2023, 2024, 2025 and the current season.
4. Open /api/health first: it fetches every NASCAR feed and shows what came back. The feed shapes were written from memory, so this is where a renamed field shows up.

Routes

Route	What it does
/	The app
/api/race?race=	Race payload: field, ratings, odds, finish distributions, weather, briefing
/api/races?season=	Schedule for the race picker
/api/plays (POST {book})	Best plays; pasted prices switch sections to value ranking
/api/parlay (POST {legs})	Prices a custom ticket inside the simulation
/api/dfs (POST {csv})	DFS projections joined to a salary CSV
/api/live	Live lap, flag and running order
/api/backtest?races=30	Leak-free backtest
/api/grades	Graded pre-race snapshots
/api/sync?season=	History backfill (weekly cron grades too)
/api/status, /api/health	What history is loaded; feed checks (?race= audits one race)

Tests

npm test runs everything against a synthetic world (tests/fixtures.js): fake drivers, fake races, the real handlers. See docs/BUILD-READOUT.md for how the model works.