Pons

Three Parquet tables. pons_token_creations is one row per token launched, with its name, symbol, decimals, initial supply, deployer and graduation threshold; pons_trades is one row per bonding-curve buy or sell; pons_launches_graduations is one row per lifecycle event, TokenLaunched, CurveCompleted, PoolGraduated, GraduationTokensPermanentlyLocked and LaunchSwept. 124 columns between them.

About 27.9 million trades across roughly 483,000 tokens, and about 602,000 launch events, as of 24 September 2026. Figures are approximate until the delivery manifest is generated; that manifest lists every file with its row count and sha256.

To 3 August 2026, the day the Pons factory went live. Launches, graduations and token creations run from that date, including the 16,080 launches recorded before the public ABI existed. Bonding-curve trades are decoded from 14 August in the standard index; when an order reaches further back we decode the curve logs for those days as it is filled, so you receive the whole window you bought. Pons is younger than the chain it runs on, so complete history here is a matter of weeks rather than years.

Join pons_launches_graduations to itself on the token address in Arguments, taking the TokenLaunched timestamp and the PoolGraduated timestamp. About 9,400 tokens in 602,000 have both, a rate of 1.56%, so most launches never reach the threshold.

pons_token_creations carries the contract address, name, symbol, decimals, initial supply, the wallet that deployed it and the graduation threshold set at launch. Names and symbols repeat heavily on a launchpad, in one 50-block sample, two of eight tokens were both called Super Inu with the symbol SI, so join on the Token address rather than the symbol.

Yes. Every trade carries Block_Number, Transaction_Index, Log_Index and Call_CallPath, so buys landing in a token's own launch block, or bundled inside a single transaction, are identifiable without an RPC.

All three tables are decoded from Robinhood Chain and filtered: trades to Trade_Dex_ProtocolName = pons_v2, token creations to the contract-creating call for each launched token, and lifecycle events to the Pons factory 0x7ed598bcef8bd9edd8c97a195c6d13f40801ec7e. The schemas were diffed field by field against the chain-wide dex_trades and events samples, and every sample file was read back with pyarrow before publishing.

Amounts are decimal strings, so casting to float loses precision on large uint256 values. Token symbols come from the contract and repeat across tokens, so join on the contract address. Failed and reverted transactions are kept, which is deliberate but means a naive count of launches is higher than the count of successful ones.

No OHLCV and no per-second candles. No pool reserve tracking after a token graduates onto Uniswap, since the chain-wide pool-event table is not populated for this chain. CurveCompleted is emitted by each token's curve rather than the factory, so it appears under the curve's address.

One line in DuckDB: SELECT * FROM read_parquet('pons_trades/*.parquet') LIMIT 10. Pandas works too: pandas.read_parquet(path). All three sample files open without credentials.

Refreshed daily with T+1 latency, so a purchase made today includes everything to yesterday. Delivered as flat Parquet under a stable layout, one prefix per table, with a JSON manifest listing every file and its sha256. We prepare the files within 7 business days of purchase and email you the location. You choose at checkout where to collect them: a Requester Pays S3 bucket, downloaded with your own AWS account, or a Cloudflare R2 bucket with a read-only API key and no egress fees. Files stay available for 7 days from delivery.