Every order Trading Floor copies is timed, per order, and shown to the user who placed it in their own Copy Log. These are the live aggregates from real production copies — not a benchmark, not a lab, the actual floor.
The headline is the whole copy: from the leader's fill to the follower's order, the time every Copy Log row shows, every copy in the window counted. The two engine figures are the copier's own share of that delay: fill detection, risk checks, order construction and submission. The broker's network round-trip is measured separately on every order: it is part of the headline and excluded from the engine figures — it's the broker's number, not ours, and it varies by broker and session. Both figures appear on each Copy Log row, so nothing is hidden: your log shows the total, and the tooltip attributes the wire.
This is the Copy Log every user sees — each row is one mirrored order with its measured latency (example rows, formatted exactly as in-app):
Example rows. Your own log shows your own numbers — including the slow ones.
Because in copy trading, latency is the product. A copier that adds 500ms to your entry hands every follower account a worse fill than the leader got — silently, on every trade. We think the only honest way to sell speed is to measure it on every order, show it to the customer, and publish the aggregate where anyone can check it.
Ask any copier for this page.
Each bar is one hour of real production copies. Height tracks that hour's 95th-percentile latency band; amber marks a slow hour — in our attribution those are almost always broker gateway time, and each user's Copy Log names the broker's share on every such fill. Hover a bar for the numbers.