F1 race winner prediction using historical and current data via FastF1. A weighted 12-feature scoring model with softmax outputs win probabilities for all drivers before each race. Features a detailed Circuits directory mapping track footprints, overtake rates, and safety car likelihood, along with Driver and Team head-to-head performance comparison tools. Historical data covers 1980–2026.
| Layer | Technology | Host |
|---|---|---|
| Frontend | Astro SSR + Tailwind | Cloudflare Pages |
| API | Hono + Drizzle ORM | Cloudflare Workers |
| Database | Neon PostgreSQL | Neon |
| Data Engine | Python + FastF1 | Render |
f1-prediction/
├── apps/
│ ├── web/ # Astro SSR (output: 'server', Cloudflare adapter)
│ └── api/ # Hono on Cloudflare Workers (NestJS-style modules)
│ ├── drizzle/migrations/ # Generated SQL migrations
│ └── src/
│ ├── db/schema/ # Drizzle table definitions (source of truth)
│ └── modules/ # races, drivers, teams, predictions, seasons
├── data-engine/ # Python ETL polling web service on Render
│ └── src/jobs/ # sync, ingest, compute jobs
└── docs/ # Architecture, API reference, schema, pipeline, deployment
See CODEMAP.md for the full file-level reference and DECISIONS.md for architectural rationale.
bun run install:all
bun run dev # runs api + web dev servers concurrentlycd apps/api
bun install
bun run dev # wrangler dev on :8787cd apps/web
bun install
bun run dev # Astro dev server on :4321cd data-engine
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
cp .env.example .env # fill in DATABASE_URLcd apps/api
bunx drizzle-kit push # apply schema to Neoncd data-engine
# First-time setup for a season
python src/main.py --job sync_schedule --year 2025
python src/main.py --job sync_season --year 2025 --round 1
# Weekly pipeline
python src/main.py --job ingest_qualifying --year 2025 --round 14
python src/main.py --job compute_features --race_id 42
python src/main.py --job compute_predictions --race_id 42
python src/main.py --job ingest_race --year 2025 --round 14
python src/main.py --job compute_season_stats --year 2025Historical backfill (2018+ FastF1, pre-2018 Ergast):
cd data-engine
python scripts/run_backfill.py 2000 2025| Variable | Service | How to set |
|---|---|---|
DATABASE_URL |
API (Worker) | Cloudflare Workers dashboard → Variables and Secrets → Secret |
DATABASE_URL |
Data Engine | Render dashboard → Environment Variables |
PUBLIC_API_URL |
Frontend | Cloudflare Pages dashboard → Environment Variables |
12 features, softmax with temperature T=0.3:
| Feature | Weight |
|---|---|
| Car Performance | 20% |
| Long Run Pace | 15% |
| Tyre Degradation | 8% |
| Reliability | 8% |
| Qualifying Delta | 8% |
| Driver Rating | 8% |
| Win Rate | 8% |
| Luck Factor | 7% |
| Circuit-Adj. Starting Position | 7% |
| Sector Strength | 6% |
| Circuit-Adj. Position Gain | 3% |
| Weather Impact | 2% |
See docs/prediction-model.md for full details.
- API: push to GitHub → Cloudflare Workers auto-deploys
- Frontend: push to GitHub → Cloudflare Pages auto-deploys
- Data Engine: Render Web Service (Hourly Polling)