Polymarket learner track
A two-hour path from public live data to a ClickHouse Cloud market pulse dashboard.
macOS terminal: Run workshop commands in Terminal using zsh or bash.
Work the modules in order. Each page states its input, exact commands, expected result, and completion check.
Progress
- Setup — prepare macOS or WSL 2, create the Cloud service, and load the environment.
- Discover markets — inspect active markets, outcomes, condition IDs, and token IDs.
- Model the data — create typed tables and the one-minute midpoint materialized view.
- Stream live data — run the collector and verify live or degraded-but-current ingestion.
- Real-time aggregates — read quote-midpoint OHLC from an incremental aggregate.
- Investigate movement — query movers, spread, freshness, and volume velocity.
- Cloud dashboard — save the queries and publish the market pulse dashboard.
- AI market analyst — make a ClickHouse Agent detect a move, investigate it, and referee its answer with SQL.
- Wrap up — prove completion, stop the collector, and understand the Kafka/ClickPipes scale-up path.
Guardrails
- Public market data only. No wallet, trading account, API secret, order, or financial advice.
- ClickHouse Cloud is the only database. Docker runs one stateless collector.
- Live mode is preferred. Fixture mode is an explicit fallback for a blocked or quiet feed.
- Never commit
.env.polymarket.
Start with Module 00.