Financial Technology
Turning live market data into decisions measured in seconds.
Meridian Markets Demo — fictional client
An original composition authored for this site, not a screenshot. Brief §108: every fixture asset is generated here so it is unambiguously licensable.
The challenge
The challenge wasn't more data. It was knowing what deserved attention.
Meridian’s desk already received every feed it could buy. What it lacked was a way to tell, in the second an instrument moved, whether that movement mattered to the position someone was holding — without asking them to watch nine panels at once.
The approach
We modelled the desk’s actual attention budget first: how many things a trader can hold, how long a glance lasts, what warrants an interruption.
That produced an architecture where the alert engine — not the interface — decides significance, and the workspace only renders what survived that filter. Application state is authoritative on the server, so a reconnect restores a session rather than reloading a page.
System architecture
Where data becomes action.
Market Data. Normalises several vendor feeds into one internal instrument model. Vendors disagree about identifiers, so reconciliation happens here rather than in every consumer downstream.
Outcomes
What changed.
- 68% less time navigating between monitoring workflows
- <250ms target UI update path, feed to render
- 14 integrated real-time data views
These figures are invented fixture data. They exist to exercise the metric component and measure nothing — no engagement produced them.
Traders stopped navigating between monitoring contexts and started working inside one.
The measurable change was not raw speed; it was how much of the session was spent reading rather than searching.