FinzData provides point-in-time US equity data for Qlib and RD-Agent. The bundle is survivorship-bias-free, point-in-time, and includes every listed and delisted US stock that traded on NYSE and Nasdaq since January 2000.
Free US data commonly used with Qlib suffers from survivorship bias and inflates backtests. FinzData includes delisted names with the suffix -DELISTED, split- and dividend-adjusted prices with $factor, and a true point-in-time calendar. This gives realistic research and production signals.
- Get a free API key at https://finzdata.com/signin
pip install pyqlib requestspython quickstart.py
Set the environment variable FINZDATA_API_KEY before running.
Initialize Qlib with the FinzData provider:
import qlib
qlib.init(provider_uri="~/.qlib/qlib_data/finzdata_us", region="us")Example feature access:
from qlib.data import D
df = D.features(["AAPL"], ["$close"], start_time="2024-01-01")Instruments are in instruments/all.txt and include -DELISTED names.
Point RD-Agent's Qlib scenario at the same data. Set the Qlib data path RD-Agent uses to ~/.qlib/qlib_data/finzdata_us with region us. RD-Agent will then run research and deployment using the FinzData point-in-time universe.
Besides all (every listed and delisted name), the bundle ships three liquid universes rebuilt every year with no look-ahead: membership for year Y is the top 500, 1,000 or 3,000 stocks by dollar volume in year Y-1, and names that later delisted stay in for the years they qualified.
D.instruments("us_top1000") # also us_top500, us_top3000- Fields: open, high, low, close, volume, factor, change
- Calendar: NYSE/Nasdaq trading calendar
- Instruments: all.txt including
-DELISTEDnames - Features are split- and dividend-adjusted and point-in-time
Every one of the 16,000+ US stocks that has traded on the NYSE and Nasdaq in the last 26+ years, delisted included, point in time. Around them: 139 futures markets with every contract since 2007, the VIX and Cboe volatility indexes with the VIX futures curve, options-market gamma (GEX) for single stocks, ETFs and the major indexes, 5,100+ ETFs, and 512 crypto pairs with Bitcoin since 2013.
Macro exactly as it was known: 22,700+ series including CPI, payrolls, wages, GDP and the Treasury yield curve, every vintage kept. Plus COT positioning since 1986, 79 FX pairs, 123 indexes, and the real economy as it happens: daily ship traffic through 28 chokepoints and 2,000+ ports, flights by airport and airline, toll and congestion traffic, and electricity use and power demand by US state and for 87 countries.
Bars at every resolution you backtest on: tick, 1-minute, 5-minute, 15-minute, 30-minute, hourly and daily, with tick-level trades on the Institutional plan. In pandas with one line:
pip install finzdata
This kit ships the US stock universe in Qlib format today; more of the catalogue is coming to Qlib format.
Free accounts receive the last 12 months. Full history since January 2000 with every delisted name is available at https://finzdata.com/shop.
The launch slides (1, 2, 3, 4, 5) and the PowerPoint deck (qlib_launch.pptx) are free to share with credit to FinzData.
Data is licensed per FinzData terms https://finzdata.com/terms. Kit code is MIT.
Contact info@finzdata.com
The FinzData team


