Real-Time Trading Platform

Real-time financial UIs fail in the browser long before they fail on the server. Naive state updates melt React under tick load, and charts built on DOM nodes redraw too slowly. A single slow consumer can back up an entire data stream.

• •Thousands of ticks per second overwhelm per-update React state changes
• •DOM-based charting cannot sustain redraw rates during volatile markets
• •Slow or backgrounded clients must not build unbounded buffers on the server
• •Order entry needs optimistic UI with strict server-side validation and reconciliation
• •Risk limits have to be evaluated before orders leave the client and again at the gateway

We separated the data plane from the render plane: streams are batched and conflated per animation frame, charts render on canvas, and every order passes a two-stage risk check.

• Batched and conflated tick updates per animation frame instead of per message
• Built charting on canvas with custom indicator plug-ins and viewport-based data windowing
• Multiplexed subscriptions over one WebSocket with per-channel backpressure and drop policies
• Implemented optimistic order tickets reconciled against authoritative gateway responses
• Modeled risk guards (position, notional, and rate limits) as declarative rules checked client- and server-side