Databento Python Client

0.75.0 · active · verified Wed Apr 15

The official Python client library for Databento, providing fast, lightweight access to both live and historical market data from multiple venues and asset classes. It supports various schemas like MBO, MBP, OHLCV, and trades, and features an efficient Databento Binary Encoding (DBN) for data storage. The library is actively maintained with frequent releases, currently at version 0.75.0, offering consistent message schemas across live and historical data.

Warnings

breaking `ts_out` field on record types is now permanent, and `__dict__` attribute has been removed from all Python record classes.
Fix: Access `ts_out` directly as an attribute instead of `__dict__`. `ts_out` returns an `int` (`UNDEF_TIMESTAMP` when not set).
gotcha A memory leak of approximately 64 bytes per record object was present in Python bindings due to a `pyo3` 0.28 regression.
Fix: This was fixed in v0.74.1 by downgrading `pyo3` to 0.27.2. Users on affected versions should upgrade to >=0.74.1 or ensure their `pyo3` dependency is compatible.
breaking The `DBNRecord` union type was moved from `databento.common.types` to `databento_dbn.DBNRecord`.
Fix: Update import statements from `from databento.common.types import DBNRecord` to `from databento_dbn import DBNRecord`.
breaking Multiple API and client library changes affected symbology types (stype) and field renaming. `product_id` was renamed to `instrument_id`. The `smart` stype was split into `parent` and `continuous`. The `native` stype was renamed to `raw_symbol`. The `product_id` stype was renamed to `instrument_id`.
Fix: Update parameter names and stype values in API calls according to the new naming conventions. For example, use `instrument_id` instead of `product_id`.
breaking The Databento Binary Encoding (DBN) format changed from `dbz` to `dbn`, and the `timeseries.stream` endpoint was renamed to `timeseries.get_range`.
Fix: Update encoding parameter to `dbn` and use `client.timeseries.get_range()` instead of `client.timeseries.stream()`.
deprecated The `mode` parameter in `metadata.get_cost` has been deprecated and will be removed in a future release.
Fix: Refactor code to avoid using the `mode` parameter in `metadata.get_cost`.
breaking Support for Python 3.9 was removed due to its end-of-life status.
Fix: Upgrade your Python environment to Python 3.10 or newer.

Install

pip install -U databento Install latest stable version

Imports

databento
```
import databento as db
```
DBNRecord
```
from databento_dbn import DBNRecord
```
As of v0.69.0, DBNRecord was moved from `databento.common.types` to `databento_dbn.DBNRecord`.

Quickstart

This quickstart initializes a Databento Historical client using an API key (preferably from an environment variable) and requests 100 trade records for E-mini S&P 500 futures for a specific hour on CME Globex, then converts the result to a pandas DataFrame and prints the head.

import databento as db
import os

api_key = os.environ.get('DATABENTO_API_KEY', 'YOUR_API_KEY')
if api_key == 'YOUR_API_KEY':
    print("Warning: DATABENTO_API_KEY environment variable not set. Using placeholder.")

client = db.Historical(api_key)

try:
    data = client.timeseries.get_range(
        dataset="GLBX.MDP3",
        schema="trades",
        symbols=["ES.FUT"],
        stype_in="parent",
        start="2023-01-01T00:00",
        end="2023-01-01T01:00",
        limit=100
    )
    df = data.to_df()
    print(df.head())
except Exception as e:
    print(f"An error occurred: {e}")

view raw JSON →