Under the Hood
How Polytrend processes market data
From on-chain feeds to the signals you see—learn how data flows through ingestion, caching, and analysis.
Data sources
- Polymarket Gamma API for live market metadata and order books.
- Polymarket Data API for historical trades and whale detection.
- Goldsky subgraphs for resilient on-chain backfill when APIs lag.
Processing steps
- Ingestion workers fetch delta updates every ~60 seconds with circuit breakers guarding rate limits.
- Redis caches hot data (markets, odds, whale feeds) for low-latency UI responses.
- DuckDB/Parquet snapshots store historical series used for analytics, backtests, and AI prompts.
Signal enrichment
Feature builders annotate markets with momentum scores, whale confidence, and AI-ready narratives so downstream components ingest a unified schema.
Want to go deeper?
The autonomous trading agent inherits the same pipeline, layering risk controls and execution queues on top.
Read about the agent architecture