On-chain history, decoded and ready to query.

Every team that does serious on-chain research eventually builds the same thing: an indexer, a decoder, a backfill job and a warehouse to hold it all. datastore exists so you can skip that project entirely. You buy the files, not API access.

Datasets arrive as partitioned Parquet with typed schemas, SHA-256 checksums, manifests and lineage docs. Load them into DuckDB, Polars or your warehouse and run whatever queries you want, as often as you want.

Visit datastore.sh

52

datasets in the catalogue

51

Solana protocols decoded

0

API quotas, ever

What it does

Files you own

Download the Parquet files once and keep them. Your queries run on your compute, with no rate limits, no metering and no bill that grows with your curiosity.

Decoded, not raw

Instructions, events and state tables come pre-decoded with typed schemas and documentation, so analysis starts at the query, not at the byte layout.

A deep Solana catalogue

51 protocols including Pump.fun, Jupiter, Raydium, Orca, Meteora, Kamino, Drift, Marinade, Phoenix, Metaplex and the core SPL programs.

Hyperliquid market history

Perps market data for Hyperliquid, one of the catalogue's most requested datasets, in the same portable format.

Verifiable lineage

Every delivery ships with checksums, version information and manifests, so you can prove exactly what data a result was computed from.

Flexible delivery

Query locally with DuckDB or Polars, load into your warehouse, or have files delivered straight to your cloud storage.

Who it’s for

Quant researchers and backtesting teams
Data engineers replacing in-house indexers
Analysts and academic researchers
Teams building AI and ML models on on-chain data

How delivery works

1Pick datasets and a coverage window, from a recent slice to full history
2Files are cut, checksummed and versioned, with schemas and manifests included
3Download or receive them in your cloud bucket, then query with your own tools

The team behind datastore is the team you hire.

CLR3 designs, builds and operates datastore end to end. When we take on consulting and product work, it is this same team, with the same habits: measured latency, monitored systems and documentation people can actually use.

Work with us