Pulse Markets
Pulse Markets docsList event-reaction coverage
Which instruments have enough measured reaction history to be worth acting on, as ONE aggregate rather than one call per instrument. Per instrument: reactionCount (the N), firstEventDate, lastEventDate and the observationWindowDays they span. Both are published because neither alone is the claim - six reactions inside one month is six measurements of a single market regime, not a history - so a caller can require a count AND a span via minReactions and minWindowDays. The filters are echoed in assumptions, so any returned set is reproducible. Only instruments with a current symbol row are returned; an instrument we cannot name is not a usable answer. Screening a universe with the per-instrument event-reactions endpoint costs one request per candidate, which is exactly what this endpoint exists to remove. eventClasses takes a comma-separated SET (`earnings,fomc,cpi`) because measurable history is a union: an ETF or a crypto never reports earnings but can have years of measured macro reactions, and asking one class at a time made that history invisible.
Which instruments have enough measured reaction history to be worth acting on, as ONE aggregate rather than one call per instrument. Per instrument: reactionCount (the N), firstEventDate, lastEventDate and the observationWindowDays they span. Both are published because neither alone is the claim - six reactions inside one month is six measurements of a single market regime, not a history - so a caller can require a count AND a span via minReactions and minWindowDays. The filters are echoed in assumptions, so any returned set is reproducible. Only instruments with a current symbol row are returned; an instrument we cannot name is not a usable answer. Screening a universe with the per-instrument event-reactions endpoint costs one request per candidate, which is exactly what this endpoint exists to remove. eventClasses takes a comma-separated SET (earnings,fomc,cpi) because measurable history is a union: an ETF or a crypto never reports earnings but can have years of measured macro reactions, and asking one class at a time made that history invisible.
Authorization
ApiKeyAuth events:readPresent your API key as a bearer token: Authorization: Bearer pk_live_....
Two environments are issued per account. A pk_test_ sandbox key returns the SAME real historical data as a live key and does NOT consume your LIVE allowance, so learning the API is free. It has its own separate windows and can be rate limited on them. A pk_live_ key reads against your allowance.
Keys carry explicit scopes (instruments:read, prices:read, actions:read, fx:read, earnings:read, options:read, news:read, sentiment:read, events:read, analytics:read, correlations:read, technicals:read, macro:read) and are enforced fail-closed, so a leaked key is bounded to what it was issued for. Revocation is immediate.
In: header
Scope: events:read
Query Parameters
Comma-separated event classes to count, e.g. earnings,fomc,cpi. Omit to count every class together. A list naming only classes with no rows yields an empty result, never an invented one. A present-but-empty value is rejected rather than widened to "all": a caller passing a computed list that came out empty wants to hear about it.
1 <= lengthOnly instruments with at least this many measured reactions. Defaults to 1, bounded 1-1000.
1 <= value <= 10001Only instruments whose earliest and latest measured event are at least this many calendar days apart. Sample size alone is gameable by a cluster: twenty reactions inside one month is twenty measurements of a single market regime, not a history. Omit to apply no span requirement at all, which is a different thing from a floor of zero and is reported back as null. Bounded 1-36500.
1 <= value <= 36500Maximum instruments returned, best-covered first. Defaults to 1000, capped at 5000.
1 <= value <= 50001000Response Body
application/json
application/json
application/json
application/json
application/json
application/json
application/json
curl -X GET "https://example.com/v1/instruments/event-reaction-coverage"{ "data": { "instruments": [ { "instrumentId": "string", "symbol": "string", "reactionCount": -9007199254740991, "firstEventDate": "string", "lastEventDate": "string", "observationWindowDays": -9007199254740991 } ], "assumptions": { "eventClasses": [ "string" ], "minReactions": -9007199254740991, "minWindowDays": -9007199254740991, "limit": -9007199254740991 } }, "meta": { "asOf": "string", "source": "string", "stale": true }}{ "error": { "code": "invalid_request", "message": "string", "docsUrl": "http://example.com", "details": { "property1": [ "string" ], "property2": [ "string" ] } }}{ "error": { "code": "invalid_request", "message": "string", "docsUrl": "http://example.com", "details": { "property1": [ "string" ], "property2": [ "string" ] } }}{ "error": { "code": "invalid_request", "message": "string", "docsUrl": "http://example.com", "details": { "property1": [ "string" ], "property2": [ "string" ] } }}{ "error": { "code": "invalid_request", "message": "string", "docsUrl": "http://example.com", "details": { "property1": [ "string" ], "property2": [ "string" ] } }}{ "error": { "code": "invalid_request", "message": "string", "docsUrl": "http://example.com", "details": { "property1": [ "string" ], "property2": [ "string" ] } }}{ "error": { "code": "invalid_request", "message": "string", "docsUrl": "http://example.com", "details": { "property1": [ "string" ], "property2": [ "string" ] } }}List economic events GET
The macro calendar, newest scheduled release first. Filter by scheduled-date range, eventClass, country and importance. actual, forecast and previous are the provider strings (e.g. "3.2%") returned VERBATIM - never parsed into a number, because the figure is often not a clean number and a parse presented as authoritative would fabricate a value the provider never gave. eventClass and importance are closed enums drawn from the ingest classifier, so a value outside the documented taxonomy is a 400 rather than a silent empty result.
Get an FX rate GET
The rate for a currency pair on a date. Returns 404 for a date not yet ingested rather than interpolating one - a fabricated rate is worse than an honest absence.