Market Data API Benchmark
We asked each provider for the same data and recorded what came back. Every number below is a measurement, not a figure from a pricing page — including the ones that are unflattering to a provider we pay for.
Results
| Provider | History returned | Rows | Coverage | Split-adjusted | Days behind |
|---|---|---|---|---|---|
| Financial Modeling Prep | 1993-01-29 → 2026-08-19 (33.6y) | 8446 | 2/64 blocked | yes (-8% worst move) | 0 |
| Yahoo Finance (yfinance) | 1993-01-29 → 2026-08-19 (33.6y) | 8446 | 6/6 | yes (-8.01% worst move) | 0 |
Coverage, symbol by symbol
Entitlement gaps do not announce themselves. A plan can serve equities happily and refuse every ETF — which is invisible until the request that needs one fails.
| Symbol | Asset class | Financial Modeling Prep | Yahoo Finance (yfinance) |
|---|---|---|---|
| AAPL | US large-cap equity | served | served |
| PG | US large-cap equity (entitlement canary) | plan does not cover | served |
| SPY | index ETF | served | served |
| QQQ | index ETF | plan does not cover | served |
| XLE | sector ETF | plan does not cover | served |
| GLD | commodity ETF | plan does not cover | served |
Method
Each provider is probed through the same adapter interface the production pipeline uses. Figures are what the requests actually returned, not what the vendor documents.
History requests SPY from 1990-01-01 and records the earliest bar that actually arrives. This distinction matters more than it sounds: one provider here returns about twenty years to an open-ended request and thirty-three to date-windowed ones, because its row cap applies per request rather than per symbol. Catalogue depth and delivered depth are not the same number.
Split-adjustment is checked against a known corporate action — 4:1 split 2020-08-31. An unadjusted series turns that month into roughly a −75% move, which would be published as a seasonal effect by anything computing monthly returns downstream.
No ranking. There is no score and no winner, because the right provider depends on which of these dimensions binds for you. A free source with full coverage may still be unusable if you need terms of service you can rely on.
Generated 2026-08-20T00:20:39+00:00 · re-measured on the same nightly schedule as the rest of the site · reproducible with python -m pipeline.bench
Disclosure
QuantEngines uses Financial Modeling Prep for its own published statistics, and the entitlement gaps shown above are on the tier we actually pay for. Where this page carries an affiliate link in future it will be marked as one, and it will not change a measurement — see the affiliate disclosure. Yahoo Finance is included because it is measurable and widely used, not as a recommendation: it is an unofficial endpoint with no terms of service for programmatic use and no availability guarantee.