Arbitrum Transfers

One Parquet table with 24 columns: every native ETH and token transfer on Arbitrum, one row each, with sender, receiver, amount, token metadata, the transaction it settled in and the fee. Internal transfers of ETH made by contract calls are included.

About 4.9 billion rows and 378 GiB across 50 monthly partitions. The last 24 months hold about 3.0 billion of them and the last 12 months about 1.2 billion. Figures are approximate until the delivery manifest is generated.

To the start of indexed history: from 31 August 2022, the Nitro upgrade, at block 22,207,820. Arbitrum One launched a year earlier; that first year, Arbitrum Classic, is not included. The Full window runs from there to yesterday.

Arbitrum has no price history of its own, so ETH, WETH, USDC, the bridged USDC.e, USDT and DAI are valued on Ethereum's series: bridged ETH is ETH, and using the real series keeps a de-peg visible. Other tokens, including WBTC, LINK and ARB, carry 0. In the linked sample file 73% of rows are priced; across all three sample files it is 81%. A value of 0 means either a $0 transfer or no price; the column cannot tell them apart.

Transfer_Amount is decimal adjusted. Multiply it by 10 to the power of Transfer_Currency_Decimals for the raw on-chain integer. We ship the adjusted value because decimals differ per token and the raw figure is wrong without them.

Amounts are decimal strings, so casting to float loses precision on large values. Symbols and names repeat, and lookalike tokens copy them, so join on the contract address. Failed transactions are kept, so filter TransactionStatus_Success for settled flow. Zero-value transfers are common: they are address-poisoning attempts, not real movement. Transaction_Type includes Arbitrum's own system types, numbered 100 to 106, alongside the standard 0, 2 and 4.

Filter Transfer_Currency_SmartContract to the token's contract, lower case with 0x. For native ETH alone, filter Transfer_Currency_Native to true; its contract column holds the string 0x.

One line in DuckDB: SELECT * FROM read_parquet('transfers/*.parquet') LIMIT 10. Pandas works too: pandas.read_parquet(path). The sample file opens without credentials.

Refreshed daily with T+1 latency, so a purchase made today includes everything to yesterday. Delivered as Parquet named by block range, 50 blocks per file and sorted ascending, under a stable S3 layout with a JSON manifest listing every file and its sha256. Signed HTTPS links are emailed once the files are prepared.