Release speed
How quickly official releases were available on FXMacroData after the official source published them, measured release by release. Every recorded release counts, whatever the result.
All recent releases
Newest first. All times are UTC. For 9 series with more than 30 releases in the window, only the newest 30 releases of each are included.
How this is measured
Every release has three times: when it was scheduled, when the official source actually published it, and when the figure was on FXMacroData. The "available within" figure counts from the actual publication, not from the schedule.
Scheduled is the time on the official release calendar. Releases often do not land on the scheduled second, so this page measures from publication, not from the schedule.
Publication is bracketed by two of FXMacroData's reads of the official source: the last read that still showed the previous period, and the first read that showed the new one. Publishers rarely issue an exact publication timestamp, so these are FXMacroData's observations and not a time supplied by the publisher. The Source published column shows the first read that showed the new figure, with both reads beneath it and how far that was from the scheduled time.
On FXMacroData is the moment the figure was stored and readable through the API, to the millisecond.
Available within runs from the earlier of the two reads to that moment, so it is the longest the gap could have been.
Within one second means that longest-possible gap was one second or less. Only this proven figure counts: a release is never counted as within one second on an estimate.
Not measurable means the earlier read was not recorded, so there is no proven figure. Those releases are shown with their times, are counted separately, and are left out of every share and median on this page.
By source results need at least three measured releases before a source is given a result. They describe FXMacroData's measured time for that source's releases and nothing else.