Ecosystem · Free data All ecosystem services yfinance on GitHub ↗

yfinance — free OHLC, no key

The pragmatic Python library for "just give me prices for an Indian ticker". Excellent for what it does.

Decision matrix

What yfinance is great at

• Zero friction. No account, no key, no quota signup. Single pip install; works in 30 seconds.
• Wide coverage. Any ticker Yahoo Finance lists; for India use the .NS (NSE) or .BO (BSE) suffix.
• Bulk download API. yf.download(['TCS.NS','INFY.NS'], start='2020-01-01') returns a tidy DataFrame.
• Yahoo's adjusted close. Built-in Adj Close column handles splits/dividends per Yahoo's methodology — not always perfect, but a sensible default.
• What to be aware of. Yahoo can change response shapes, rate-limit, or break suffixes without warning — yfinance is a community lib reverse-engineering an internal Yahoo endpoint. For production, treat it as best-effort.

Where Genka complements yfinance

• Primary-source XBRL fundamentals. /companies/{symbol}/ratios, /balance-sheets, /cash-flow-statements — sourced directly from NSE filings, not scraped from Yahoo's summary page.
• Concall corpus. ~23,000 transcripts; BM25 + grounded Q&A. Yahoo doesn't have these.
• MF analytics. 4,700 funds, SEBI taxonomy, 20y NAV, screener, backtester — Yahoo's MF coverage is thin for Indian schemes.
• Auditable corp actions. /prices/{symbol}/corp-actions returns each split/bonus/dividend with a numeric factor. Long-window returns ship in two flavours (raw and _adj) so you see when adjustment matters.
• Production stability. Versioned API contract (genka-Version header) and a published deprecation policy. Code that pinned 2026-04-25-caracal still works in three years.

Compose them

yfinance for the quick prototype, Genka for the production dependency:

# Prototype: yfinance for fast price exploration
import yfinance as yf
df = yf.download("TCS.NS", start="2020-01-01")

# Production / agent loop: Genka for stability + depth
import requests
HDR = {"X-API-Key": os.environ["GENKA_API_KEY"]}
prices = requests.get(
  "https://genka.dev/v1/prices",
  headers=HDR,
  params={"symbols": "TCS", "from": "2020-01-01"}
).json()

# And get the indicator panel + fundamentals Yahoo doesn't compute:
panel = requests.get("https://genka.dev/v1/prices/TCS/latest", headers=HDR).json()
ratios = requests.get("https://genka.dev/v1/companies/TCS/ratios", headers=HDR).json()

Relevant Genka endpoints

• GET /v1/prices?symbols=A,B&from=1995-01-02 — 30y NSE bhavcopy as JSON
• GET /v1/prices/{symbol}/latest — pre-computed indicator panel (CA-adjusted)
• GET /v1/prices/{symbol}/corp-actions — auditable corp-action stream
• GET /v1/companies/{symbol}/ratios — quarterly fundamentals from XBRL
• POST /v1/concalls/answer — grounded Q&A across 23k transcripts

Full catalog: /llms.txt

Working with yfinance — gotchas

• Yahoo can rate-limit hard. yfinance reverse-engineers an internal Yahoo endpoint. During heavy use you may get 429s or empty payloads; back off, retry with jitter, or batch via yf.download(['TCS.NS', 'INFY.NS', ...]).
• .NS for NSE, .BO for BSE. Use RELIANCE.NS for NSE, 500325.BO for BSE (numeric BSE script code). Some illiquid tickers exist on only one suffix; if NSE returns nothing, try BSE.
• Adj Close vs Close. Adj Close applies Yahoo's split + dividend adjustment; Close is raw. For backtests use Adj Close; for matching exchange records use Close.
• Production reliability is best-effort. Yahoo can change the underlying API shape without notice and yfinance breaks until a maintainer ships a fix. For production loops, mirror what you need into your own store.