Best Market Data API for Trading
The two questions that decide which data provider fits your use case are: what granularity do you need (tick, minute, EOD), and is your historical universe survivorship-bias-free? Most free APIs fail the second test — fine for learning, insufficient for rigorous equity backtests. This page compares 9 providers on the axes that matter. No investment advice, no affiliate links — neutral comparisons for researchers choosing their stack.
Choosing your stack — Module 2 hub cluster:
- Backtesting Frameworks
- Broker APIs
- Market Data Providers ← you are here
- Trading Platforms
- Event-Driven vs Vectorized Backtesting (concept)
Provider Comparison Table
Prices marked [approx] were not publicly accessible without signup; re-verify at the provider's pricing page before committing. Star counts refer to GitHub repositories where applicable.
| Provider | Assets | Free Tier? | Granularity | Real-Time? | Bias-Free? | Pricing (~2026) |
|---|---|---|---|---|---|---|
| Polygon.io → Massive.com | Stocks, options, indices, FX, crypto | Yes (5 req/min, delayed) | Tick, minute, EOD | Paid only | No (delisted in history) | ~$29–$199+/mo per asset class [approx] |
| Databento | Futures, options, equities (70+ venues) | $125 credits (6-mo) | Tick / MBO (L3) → OHLCV | Both (RT + 15+yr history) | Raw venue feeds | Pay-per-GB or flat subscription [approx] |
| Tiingo | Stocks/ETFs/funds, crypto, FX, fundamentals, news | Yes (~50/hr, 500 symbols/mo) | EOD + intraday (IEX) | IEX intraday | Strong daily history | ~$10/mo Power plan |
| Alpha Vantage | Stocks, FX, crypto, commodities, fundamentals | Yes (25 req/day, 5/min) | Intraday → monthly | Paid | No | ~$50–$250/mo [approx] |
| Finnhub | Stocks (US+intl), FX, crypto, alt-data | Yes (60/min, real-time US quotes) | Tick (WS), candles, EOD | Free RT US quotes | No | ~$50–$100/mo [approx] |
| yfinance (unofficial) | Whatever Yahoo exposes | Free (no key) | Daily; limited intraday | Delayed | No | $0 (ToS / reliability risk) |
| Norgate Data | US/AUS/CAN equities, ~100 futures, FX | Trial only | EOD only | EOD updates | Yes — explicitly bias-free | Tiered [approx] |
| EODHD | Stocks, ETFs, bonds, options, FX, crypto, macro (60+ exchanges) | Yes (20/day, EOD, ~1yr) | EOD + intraday (paid) | Some RT on higher tiers | No | $0 → ~$99.99/mo |
| IEX Cloud | (historical US equities) | DISCONTINUED Aug 31, 2024 | — | — | — | Migrate away |
Six Points of Nuance the Table Cannot Capture
1. Polygon.io is now Massive.com
Polygon.io rebranded to Massive.com on October 30, 2025. The old api.polygon.io endpoint continues to work in parallel, but pricing, plans, and documentation now live at massive.com. Plans are split by asset class (equities, options, FX, crypto) rather than a single unified subscription. The underlying data, team, and API structure are unchanged. If you have existing code using the polygon-api-client Python SDK, it should continue working — but check for a Massive SDK update as the transition matures.
2. IEX Cloud shut down August 31, 2024
IEX Cloud discontinued its API service on August 31, 2024. All API keys were deactivated on that date. Any existing code using the IEX Cloud client libraries produces errors and will not recover without a migration. IEX itself pointed to Databento as a primary migration path. Tiingo, which offers IEX-grade intraday data via its own endpoint, is a cost-effective alternative for stock/ETF work. Polygon/Massive is the best alternative for real-time US equity data with an easy free tier.
3. Norgate is the only explicitly survivorship-bias-free source in this list
Survivorship bias is one of the most underappreciated sources of phantom alpha in equity backtesting. A universe that excludes delisted stocks overweights survivors, making trend-following and momentum strategies look better than they perform live. Norgate Data is the only provider in this review that markets itself as explicitly survivorship-bias-free: its historical universe includes companies that went bankrupt, were acquired, or were delisted, with correct continuous-contract adjustments for futures. The tradeoff is significant: Norgate is EOD only, with no intraday granularity. Pair these two facts when assessing fit for your strategy frequency.
4. yfinance is an unofficial scraper with known risks
yfinance is the most widely used market data library in Python tutorials and academic research. It is also the most misrepresented: it is an unofficial scraper of Yahoo Finance's internal API, not a sanctioned data provider. Yahoo's terms of service explicitly prohibit commercial use. The library breaks periodically when Yahoo changes its internal schema (rate limiting, endpoint changes), often without warning. For individual learning and prototyping on paper strategies, the risk is acceptable. For research that will inform real capital, the reliability and ToS risks are material.
5. Databento is the closest to raw exchange feed at retail pricing
Most retail data providers clean, normalize, and deliver OHLCV bars. Databento is different: it delivers raw venue feed data in its original schemas (including Market-By-Order level 3 data) from 70+ venues, with 15+ years of history for many instruments. This makes it the best choice for microstructure research, futures roll studies, and strategies that need tick-level fill-model accuracy. The pay-per-GB pricing model means small experiments are inexpensive; large backtests cost more. It is not beginner-oriented, but it is the most academically rigorous retail data source available.
6. Most "free" APIs are not survivorship-bias-free
This is the most important caveat for any researcher running equity backtests on free data. yfinance, Alpha Vantage, Finnhub, Tiingo, and EODHD (on free tiers) all cover primarily surviving symbols. If you backtest a cross-sectional momentum strategy on a universe of S&P 500 survivors from 2010 to today, you have implicitly given your strategy knowledge of which companies survived for 15 years — information that was unavailable in 2010. For strategies tested on individual instruments, the bias is lower. For strategies that rank or weight a universe of stocks, it is severe.
How to Choose / Best For Whom
- Free learning and prototyping: yfinance (daily prices, no key, huge ecosystem) — with the ToS and reliability caveats documented above.
- Tick-level or futures microstructure research: Databento. Closest to exchange-native data at retail pricing.
- Survivorship-bias-free equity research (EOD strategies): Norgate Data. The only explicitly bias-free source in this list.
- Production intraday with real-time, broadest US coverage: Polygon/Massive (paid tier). api.polygon.io still works post-rebrand.
- Low-budget fundamentals + EOD + international: Tiingo (~$10/mo Power) or EODHD ($0–$100/mo).
- Free real-time US equity quotes + alternative data: Finnhub (60/min free, includes sentiment and earnings data).
- Migrating from IEX Cloud: Tiingo (most similar free tier, IEX intraday via its own endpoint), Databento (tick-level, US equities + 70+ venues), or Polygon/Massive (broad coverage, easy REST).
What this page does not cover: This hub compares retail and prosumer market data APIs. It does not cover institutional data terminals (Bloomberg Terminal, Refinitiv/LSEG), broker-bundled historical data APIs (IBKR Historical Data, TD Ameritrade thinkorswim), proprietary tick databases, or alternative data vendors (satellite, credit card, news sentiment at institutional scale). For the backtesting engines that consume this data, see Backtesting Frameworks. For the concept of survivorship bias in backtesting, see Event-Driven vs Vectorized Backtesting.
Frequently Asked Questions
- Is IEX Cloud still available?
- No. IEX Cloud shut down on August 31, 2024. Existing API keys stopped working on that date. For migration, the three most commonly cited alternatives are Polygon.io (now Massive.com) for US equities and crypto intraday and real-time data, Databento for tick-level and futures data, and Tiingo for cost-effective EOD and fundamentals. IEX's own migration guide pointed to Databento as a transition path.
- What is the best free market data API for backtesting?
- For free experimentation, yfinance is the most widely used — it wraps Yahoo Finance with no API key required, covers daily prices for US equities and many international markets, and has broad library support. Its limitations: unofficial (Yahoo's terms of service do not permit commercial use), breaks frequently (rate limiting, schema changes), and history does not include delisted symbols. For a free tier with an official key and better reliability, Alpha Vantage (25 requests per day free tier) or Finnhub (60 requests per minute, free real-time US quotes) are the most common alternatives.
- Which market data providers are survivorship-bias-free?
- Of the nine providers reviewed here, only Norgate Data is explicitly marketed and built as survivorship-bias-free, including delisted symbols in its historical universe. It covers US, Australian, and Canadian equities plus approximately 100 futures contracts and FX. The tradeoff: Norgate is EOD only — no intraday granularity. Polygon.io (Massive.com) includes delisted symbols in its history files, which provides partial bias protection for equity backtests, though it is not marketed as a survivorship-bias-free product. Most other free and mid-tier providers cover only surviving symbols.
- What happened to Polygon.io?
- Polygon.io rebranded to Massive.com on October 30, 2025. The old api.polygon.io endpoint continues to work in parallel during the transition. Pricing and plans now live at massive.com and are split by asset class. The data product, team, and underlying API are the same; only the brand changed. Code using the polygon-api-client SDK should continue working; check the Massive.com migration guide for any updated client packages.
- Is yfinance safe to use for backtesting?
- yfinance is safe for learning and prototyping. Its limitations for rigorous backtesting are: (1) it is an unofficial Yahoo Finance scraper — Yahoo's terms of service do not permit commercial use, and the library breaks when Yahoo changes its internal API; (2) price history is adjusted retroactively, creating look-ahead bias if not handled carefully; (3) it does not include delisted symbols, creating survivorship bias in equity universe selection. For research that will inform real capital allocation, the data quality risks are significant enough to warrant a provider with an official API and clear terms of service.
Drill market data concepts — survivorship bias, look-ahead bias, point-in-time pricing, and data provider trade-offs — with AlgoDrill's spaced-repetition flashcards.
Start Flashcards → View Reading List →