How we measure speed
Every story carries the time its publisher gave it. We record the moment CryptoMarkets first saw it. The gap between the two is the latency on each card and in the figures below.
Last 24 hours
| Stories | Median | p90 | Measured |
|---|---|---|---|
| Exchange announcements | 113ms | 220ms | 9 |
| All news | 400ms | 853ms | 76 |
From the bundled snapshot. Live figures load from our server and update every 10 minutes. A figure needs at least 5 measured stories; below that it reads “measuring…”.
What the numbers mean
- Publisher timestamp: the publication time in the source’s own feed, API or channel post.
- First seen: the moment our fetch of that source returned the story, on our server’s clock.
- Latency: first seen by CryptoMarkets minus the publisher timestamp, in milliseconds. When a publisher’s clock runs ahead of ours, the story counts as 0 ms.
- Median and p90: half of the measured stories were first seen within the median, and nine in ten within the p90. Both are taken over the stories first seen in the last 24 hours (nearest-rank method). “Exchange announcements” counts only notices from exchanges; “All news” counts every measured story.
What it includes, and what it leaves out
The figure includes any delay on the publisher’s side (a story that reaches its own feed late counts as late) and our own processing until the story is on the site. It leaves out the time your browser takes to load the page.
How we watch sources
We read exchange announcements straight from each exchange’s official API and channels, and keep live connections open to them and to newsroom feeds, so a story reaches us the moment it is published, not at the next scheduled check. Exchange notices get a green badge on their card.
Why we publish this
Speed claims should be checkable. These are our measured times, as they are, for anyone to compare. Specialist trading terminals can be faster for single exchanges; CryptoMarkets covers every region in one place.