How far back does historical intraday stock data go?

There is no single number, because every provider digitized a different era first. What you get is a ladder of starting dates, and the rung matters more than the marketing. The table below is the whole answer in one view, and the sections after it unpack each rung:

Two bookends frame everything else. Massive publishes index data only from 2023, while Tiingo's corporate-action-adjusted end-of-day prices start in 1962 - a 41-year gap between the newest and oldest start dates in this pool. Of the self-serve sources, only Alpha Vantage's minute archive (January 2000) and Massive's US stock tape (2003) predate the mid-2000s, which is why strategies needing pre-2003 session detail end up negotiating with LSEG rather than signing up for an integration key.

Ranked by documented start date at the finest resolution each sells:

The deepest self-serve intraday archive belongs to Alpha Vantage. Its TIME_SERIES_INTRADAY function accepts interval values of 1min, 5min, 15min, 30min and 60min, and an optional month parameter takes a YYYY-MM value with documented history back to January 2000.

LSEG's Tick History - PCAP product is documented as a cloud-based, 20+ petabyte repository of ultra-high-quality global market data captured at the data-centre level, inside a platform totaling 80+ petabytes of tick history reaching back to 1996. That capture method is the point: the tape preserves what the venue emitted, not a vendor's later reconstruction of it.

Depth is asset-specific even inside one vendor. Massive carries US stocks from 2003 - 20+ years of history on Advanced plans - but currencies only from 2009, options from 2014, futures from 2017 and indices from 2023.

The stock tape is the scale story: 350B+ data points of tick-level trades and quotes from all US exchanges, dark pools and OTC markets, each event carrying nanosecond timestamps on both the SIP and participant clocks, recorded across 32,345+ tickers. Options add 722B+ data points over 1.67M+ contracts since 2014.

For Alpha Vantage the loop is month-driven: issue one TIME_SERIES_INTRADAY call per YYYY-MM slice, walking backwards from the current month toward January 2000, then stitch the responses into a single continuous series locally. Rebuilding the full span means roughly 320 monthly requests, which is why the 75-1,200 requests-per-minute premium limits exist.

Massive inverts the model into bulk delivery. Daily flat files arrive as downloadable S3 objects for every asset class, spanning day and minute aggregates as well as raw trades and quotes, and an on-request SQL interface queries the same store directly - no cursor pagination involved. Client libraries pass the key through a MASSIVE_API_KEY environment variable across Python, JavaScript, Go and Java.

Two sibling clusters continue the depth question from opposite ends: tick by tick stock data covers nanosecond-grain sourcing, and market data flat files s3 covers the bulk-loading patterns that make a 20-year pull practical. Every figure on this page traces to cataloged source records as of August 2026.