polyledger

PolyLedger pulls Polymarket's whole trade history into one file you can query with SQL. It hits the CLOB API for market and token metadata, streams OrderFilled events off Polygon, joins them, and writes it all into a single DuckDB database. Markets, tokens, raw fills, and a joined trades view, exportable to Parquet.

The design choice worth noting is Envio HyperSync instead of a normal RPC node. No Alchemy or Infura key, no rate limit babysitting, and a free HyperSync token covers it. Python 3.11 or newer, pip install from a clone. It checkpoints as it goes, so an interrupted index resumes instead of restarting, and it indexes both the V1 and V2 exchange contracts so history does not stop at the contract migration.

Anyone doing prediction market analytics, backtests, or research gets a working dataset in an afternoon instead of writing an indexer. MIT, no paid tier, nothing to host.

The catch is that a free tier on somebody else's API is still a dependency. If HyperSync changes its limits or pricing, your pipeline changes with it, and there is no fallback path to a plain RPC node. Budget disk space too. Every fill on an active venue is not a small file.