OpenBB

Overview

OpenBB (Open Data Platform, repository OpenBB-finance/OpenBB, Apache-2.0) turns many financial data sources into one typed Python client (from openbb import obb), a FastAPI REST server (openbb-api) and an MCP server (openbb-mcp). Results are OBBject objects with .to_dataframe(), .to_dict() and .results. Checked against openbb 5.0.0 / openbb-core 2.0.1 (2026).

Version 5 is a modular rewrite: openbb-core ships no commands, and what you can call depends on which extension packages are installed. Older tutorials (4.x) rely on providers such as yfinance, fmp and polygon that are not part of 5.0; commands like obb.equity.fundamental.metrics and obb.equity.discovery.losers no longer exist. Always list what you have before writing code (see "Discover what is installed").

Instructions

Installation

python3 -m venv .venv && source .venv/bin/activate
pip install "openbb[routers]"     # client + provider extensions + equity/economy/crypto/... routers

Plain pip install openbb installs the providers (cboe, nasdaq, sec, fred, oecd, ecb, imf, tmx, ...), obb.news, charting, the API and MCP servers, but not the equity, economy, crypto, currency, technical routers; those come with the [routers] extra or individual packages (pip install openbb-equity openbb-technical). After adding or removing an extension, openbb-build regenerates the static package (it also rebuilds on import).

Discover what is installed

from openbb import obb

print([n for n in dir(obb) if not n.startswith("_")])           # namespaces
print([n for n in dir(obb.equity.fundamental) if not n.startswith("_")])
import inspect
print(inspect.signature(obb.equity.price.historical))
print(obb.equity.price.historical.__annotations__["provider"])   # providers allowed for this command

Each command accepts provider="..."; without it, the first available provider in the command's priority list is used. In this install equity.price.historical accepts cboe, nasdaq, tmx; crypto.price.historical only nasdaq; currency.price.historical ecb and tmx; news.company nasdaq and tmx.

Equity data

from openbb import obb

prices = obb.equity.price.historical("AAPL", start_date="2026-01-01", provider="cboe").to_dataframe()
quote = obb.equity.price.quote("AAPL", provider="cboe").to_dataframe()

income = obb.equity.fundamental.income("AAPL", provider="sec", limit=4).to_dataframe()   # SEC filings
ratios = obb.equity.fundamental.ratios("AAPL", limit=4).to_dataframe()                   # nasdaq
hits = obb.equity.search("apple").to_dataframe()
universe = obb.equity.screener(provider="nasdaq").to_dataframe()   # about 7,000 rows: filter in pandas

Technical indicators take a DataFrame with a date index and close: obb.technical.sma(data=prices, length=20), obb.technical.rsi(data=prices, length=14), obb.technical.macd(data=prices) (columns macd, signal, histogram).

Macro, currency, news

cpi = obb.economy.cpi(country="united_states", provider="oecd").to_dataframe()
gdp = obb.economy.gdp.nominal(country="united_states").to_dataframe()
eurusd = obb.currency.price.historical("EURUSD").to_dataframe()
news = obb.news.company("AAPL", limit=5, provider="nasdaq").to_dataframe()

# FRED needs a free key
fedfunds = obb.economy.fred_series("FEDFUNDS", provider="fred").to_dataframe()

Credentials

Providers that need a key read <PROVIDER>_<CREDENTIAL> environment variables (they take precedence), then ~/.openbb_platform/user_settings.json, or you set them per session:

export FRED_API_KEY="your FRED key from fredaccount.stlouisfed.org"
export BLS_API_KEY="your BLS key"

import os
obb.user.credentials.fred_api_key = os.environ["FRED_API_KEY"]

A missing key raises OpenBBError: Missing credential 'fred_api_key'.

REST API and MCP server for agents

openbb-api                                   # http://127.0.0.1:6900, OpenAPI docs at /docs
curl "http://127.0.0.1:6900/api/v1/equity/price/historical?symbol=AAPL&provider=cboe&start_date=2026-09-28"

openbb-mcp --transport stdio                 # default transport is streamable-http on 127.0.0.1:8001

The MCP server exposes the installed commands as tools. Limit them with --allowed-categories and keep the server on 127.0.0.1. For OpenBB Workspace, add http://127.0.0.1:6900 as a custom backend in the Apps tab.

Providers (what each gives, key needed)

The openbb-yfinance extension exists on PyPI only as a 2.0.0 release candidate (July 2026); install it with pip install --pre openbb-yfinance only if you accept that.

Examples

Example 1: Stock snapshot for an AI agent

User request: "Give me a one-call function that summarizes AAPL: last close, 52-week range, latest revenue and net income, and three headlines."

from openbb import obb

def snapshot(ticker: str) -> dict:
    px = obb.equity.price.historical(ticker, start_date="2025-10-01", provider="cboe").to_dataframe()
    inc = obb.equity.fundamental.income(ticker, provider="sec", limit=1).to_dataframe()
    news = obb.news.company(ticker, limit=3, provider="nasdaq").to_dataframe()
    return {
        "ticker": ticker,
        "last_close": float(px["close"].iloc[-1]),
        "range_52w": (float(px["low"].tail(252).min()), float(px["high"].tail(252).max())),
        "revenue": inc["total_revenue"].iloc[0],
        "headlines": news["title"].tolist(),
    }

print(snapshot("AAPL"))

The agent checks column names with inc.columns first, since they differ between providers (total_revenue for sec, revenue for nasdaq). The result is a dict with the last close, a (low, high) tuple and three titles.

Example 2: Inflation versus rates without paid keys

User request: "Plot US inflation against the policy rate since 2020 using free data."

The agent calls obb.economy.cpi(country="united_states", provider="oecd", start_date="2020-01-01") and obb.economy.interest_rates(country="united_states", provider="oecd"), joins the two DataFrames on date, and plots both series with matplotlib. If the user has a FRED key, it swaps in provider="fred" for daily-fresh series.

Guidelines

• Do not copy 4.x snippets: check dir(obb.<namespace>) and the provider list of a command before using it.
• Pass provider= explicitly in agent code so results do not change when you install another extension.
• Free providers have delays, rate limits and gaps (crypto.price.historical("BTC-USD") returned no rows in testing); handle EmptyDataError and OpenBBError.
• Market data is not investment advice; cite the provider and date in anything shown to users.
• Never hard-code API keys; use environment variables or ~/.openbb_platform/.env.
• Run openbb-api and openbb-mcp on 127.0.0.1 only.
• Not a fit for tick-level or low-latency trading data.

Resources

• Documentation
• Repository and README
• Agents for OpenBB