Token OHLCV Hourly

One Parquet table, token_ohlcv_1h, with 20 columns: the hour, the token and its chain, USD open, high, low and close, USD and token volume, and four supply figures including market cap. About 617,000 rows for 1,000 tokens over 30 days.

About 617,000 rows and roughly 47 MB as Parquet, estimated from a one-day sample at 77 bytes per row. The export for the 30 days to 8 October 2026 returned 616,536 rows; exact figures arrive with the delivery manifest.

Tokens are ranked by their median market cap over the 30 days, counting only bars where the circulating supply is known, with a floor of $100,000 traded in the window. The list is fixed when your order is built and shipped inside the files, so you can see which tokens you got.

30 days. The price index keeps 30 days of bars, so that is the longest window this dataset can hold. Each order is built once, when you buy, and covers the 30 days up to the latest full UTC day. It is a fixed set of files and does not update afterwards. For prices that keep updating, use the Bitquery Trading API.

Bars come from the Bitquery price index, the same series the Trading API serves. It prices each token in USD from its DEX trades and filters out MEV trades. Each row is one hourly bar from that index. We ran the export for two windows before release. Both returned every day in the window, and no bar had a high below its low.

Tokens outside the top 1,000, tokens with no known circulating supply, and tokens that traded under $100,000 in the window. Centralized exchange trades are not in the source. Hours with no trade have no row; nothing is forward filled, which is why the file holds about 617,000 rows and not the 720,000 a full grid would.

Market cap belongs to the currency, so every copy of a coin carries the same figure: WETH on Base shows the market cap of ETH. Never sum it across rows. Circulating supply is known for about 4,800 tokens; for the rest the market cap column equals the fully diluted value. Symbols and names repeat, so join on Token_Id. Prices are 32-bit floats in the source, good to about seven digits. Supply figures can be 0 on a bar where the reference data had not arrived.

Filter Token_Id to the token, for example bid:eth:0xc02aaa39b223fe8d0a0e5c4f27ead9083c756cc2 for WETH on Ethereum, and order by Interval_Time_Start.

One line in DuckDB: SELECT * FROM read_parquet('token_ohlcv_1h/*.parquet') LIMIT 10. Pandas works too: pandas.read_parquet(path).

Each order is built once, when you buy, and covers the 30 days up to the latest full UTC day. It is a fixed set of files and does not update afterwards. For prices that keep updating, use the Bitquery Trading API. Delivered as Parquet with ZSTD compression, one file per day, under a stable layout 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.