TickerTick

What it is

TickerTick is a free HTTP news API covering companies listed on US exchanges. One endpoint, GET https://api.tickertick.com/feed, takes a query string and returns stories newest first as JSON. There is no key, no account and no signup: a plain unauthenticated GET returned HTTP 200 on 14 September 2026.

The query language is the reason to choose it over a generic headline feed. Five term types — z:aapl for stories about a ticker, tt:aapl for a broader set, s:cnbc for a source domain, E:elon_musk for an entity in the title and T:sec for a story type — combine with (and ...), (or ...) and (diff a b). So (diff (or tt:meta tt:goog) s:reddit) is Meta and Alphabet news with Reddit removed, expressed once and filtered server-side rather than in your own code. Nine story types are defined — curated, earning, market, sec, sec_fin, trade, ugc, analysis and industry — and the curated ones resolve to domain lists published in the repository, from 80 top news sources to 32 user-generated-content platforms.

Pricing

Free, with nothing for sale and no pricing page to read. The terms are three bullets in the project's README: the API is free for commercial and non-commercial use, every endpoint is limited to 10 requests per minute from one IP, and you are invited to file an issue when stories look wrong. The README puts the block precisely — more than five requests in any 30-second window blocks the IP for 30 seconds — but it did not fire under test: on 14 September 2026 three bursts totalling 32 uncached requests, twelve of them inside five seconds, all returned HTTP 200. Treat the published number as a budget you are asked to keep to, not one the service enforces.

Data & coverage

News only. There are no quotes, no bars and no fundamentals here, so anything that needs a price needs a second provider. Coverage is US listings, including ETFs, plus a set of private companies addressed by a leading dot (.ftx, .ro). Each story carries an id, title, url, source site, a timestamp in epoch milliseconds, the tickers it was matched to, usually a description and sometimes a similar_stories list of other ids in the same response — enough to collapse the same wire story reprinted five times.

Depth per request is the thing to plan around. n is documented as 1 to 200, but on 14 September 2026 both n=200 and n=201 returned 150 stories, so 150 is the practical cap. For a heavily covered name that is a narrow window: 150 stories for tt:aapl spanned nine hours of the same morning. For a quiet one it is months — 98 stories for z:cato reached back to 27 May 2025. The last parameter takes a story id and pages backwards from there.

Bulk history is published as GitHub releases instead of through the API. The most recent, dated 26 May 2025, is a 4.9 GB bzip2 file of roughly 16 million stories, one JSON object per line; its predecessor from November 2023 held about 8 million. Nothing has been released since.

Integrations

Nothing official. The vendor publishes no SDK in any language; the pytickertick package on PyPI is linked from the README but belongs to a different publisher, pennacap, and has a single release uploaded in July 2023. The repository also ships a SKILL.md for Claude Code. There is no MCP server, no webhook and no push of any kind — you poll.

Limitations

• The service is closed, whatever the repository suggests. The GitHub repo is MIT-licensed and holds nine files: a README, a licence, two Claude Code instruction files, four lists of source domains and a one-page how-to. None of the service runs from it, and the licence covers the documentation rather than the data or the endpoint.
• There are no terms of service, no privacy policy, no SLA and no status page. The entire agreement is three README bullets from one maintainer, so nothing is promised and there is no contract to point at when a production feed stops. Weigh that before it sits under anything that matters.
• /tickers caps at 20 results whatever n says. Both n=9999 and n=100 returned 20 on 14 September 2026, even though the README links the n=9999 URL as its evidence for covering around 10,000 tickers. Ticker search works; enumerating the universe through it does not.
• tt: is broad enough to be noisy. In a 150-story tt:aapl response on 14 September 2026, reddit.com supplied 12 stories — second only to macrumors.com and ahead of every general news site. Use z:, or subtract with (diff ... T:ugc).
• US listings only, and no non-US markets at all.
• Last commit on the documentation repository: 4 June 2026.

Alternatives

Finnhub puts a year of news on its free key alongside quotes and fundamentals, so one provider covers both halves. Alpha Vantage attaches sentiment scores, which TickerTick does not compute. EODHD is the paid option when news has to arrive with non-US coverage.