Dataset Downloads: File Formats, CSV and Parquet
Details
Docs navigation
- Getting started: Quickstart, Authentication, Rate limits, Errors and retries
- Guides: Symbol formats, Timestamps and time zones, Pagination, Split and dividend adjustments, WebSocket streaming, Dataset downloads (you are here), Excel and Google Sheets, Changelog
- API reference: Stock bars, Stock trades and quotes, Snapshots and movers, Tickers, exchanges, calendar, Splits and dividends, Fundamentals and SEC filings, Options chain and Greeks, Options EOD, bars, OI, Forex, Crypto, Crypto derivatives, Futures bars and continuous, Settlements and COT, Indices, Economic data, Technical indicators
- SDKs: Python, JavaScript
Some jobs are better served by files than by thousands of API calls: a ten-year backtest across 500 stocks, a research database, a one-off study. You buy the dataset once, download it, and keep it. This page covers the formats we ship, how a market data Parquet download is laid out, how to verify what you received and how to load it.
Early access. Paid datasets are not on sale yet. The free samples on /sample-data are real and downloadable today, and they use the same column names as the paid files will.
Coverage
What you get
| Format | When to pick it | Notes |
|---|---|---|
| CSV (gzip) | Excel, quick looks, any language | One header row, comma separated, UTF-8, .csv.gz |
| Parquet | pandas, Polars, DuckDB, Spark | Columnar, typed, much smaller on disk, fast to filter |
| ZIP (free samples) | Trying our format | Contains CSV, JSON and a README.txt |
Columns match the API field names exactly. A file of 1-minute stock bars has symbol, timestamp, interval, open, high, low, close, volume and the rest, so code written against the API works on the file.
Datasets
How to download and use a dataset
1. Try a free sample
Pick any file on /sample-data. Each ZIP holds the same data as CSV and JSON. Here are the first three rows of the Bitcoin 1-minute bars sample:
symbol,interval,timestamp,open,high,low,close,volume,quote_volume,vwap,trade_count,taker_buy_volume
BTC-USDT,1m,2026-10-31T15:40:00Z,77598.76,77598.76,77598.75,77598.76,6.02156,467265.57,77598.76,479,4.31787
BTC-USDT,1m,2026-10-31T15:41:00Z,77598.75,77628.53,77598.75,77628.53,8.09918,628567.28,77608.76,1773,5.61679
Source: Binance
2. Verify the file
Every download lists a SHA-256 checksum. For the sample above, btc-usdt-1m-bars.zip should hash to 9145815ca7df6e6b7c45d3aad0e312353d9253d42ceee01b587e2c59d838a4af.
sha256sum btc-usdt-1m-bars.zip
3. Understand the layout of paid files
Large datasets are split so you can load only what you need. Paid Parquet files are partitioned by symbol and year:
stocks-bars-1m/
symbol=AAPL/year=2024/part-0.parquet
symbol=AAPL/year=2025/part-0.parquet
symbol=MSFT/year=2024/part-0.parquet
manifest.json
manifest.json lists every file with its row count and checksum.
4. Load it
Example (Python):
import pandas as pd
btc = pd.read_csv("btc-usdt-1m-bars.csv", parse_dates=["timestamp"]) # one CSV from a free sample
bars = pd.read_parquet( # partitioned Parquet, only two symbols
"stocks-bars-1m/",
filters=[("symbol", "in", ["AAPL", "MSFT"])],
)
DuckDB can query the folder without loading it into memory:
SELECT symbol, date_trunc('day', timestamp) AS day, max(high) AS high
FROM read_parquet('stocks-bars-1m/*/*/*.parquet', hive_partitioning = true)
GROUP BY 1, 2;
Pricing
Suggested dataset prices
| Dataset | Price (suggested, early access) |
|---|---|
| US stocks, daily bars, full universe | $149 |
| US stocks, 1-minute bars | $5 per symbol, or $999 full universe |
| US stocks, trades | $15 per symbol-year |
| US stocks, NBBO quotes | $25 per symbol-year |
| Options EOD with Greeks and IV | from $19 per underlying |
| FX daily / 1-minute bars / ticks | $19 per pair / $29 per pair / $3 per pair-month |
| Crypto bars / trades | $5 per pair-year / $3 per pair-month |
| Futures 1-minute bars / continuous / tick | $12 per root per year / $29 per root / $49 per root per year |
| Index levels | $19 per index |
Full list on /datasets.
Watch out
Notes and gotchas
- Timestamps in files are UTC strings in CSV and timezone-aware UTC timestamps in Parquet. See timestamps and time zones.
- Stock bar files are split-adjusted as of the day you download them and include raw prices. A later split will not change your copy. See adjustments.
- Excel opens files up to 1,048,576 rows. A year of 1-minute bars for one stock fits; trades for a busy stock do not. Use Parquet or split by month.
- Free sample licences differ by source. Binance samples carry a note to check Binance terms before commercial display; government sources are public domain. Each
README.txtsays which. - Files are yours to keep for internal use. Redistribution needs an Enterprise licence.
Keep reading
Next steps
Details
Was this page useful?
[Yes] [No]
Want a dataset cut we do not list? Ask on the custom data form.