Real-time portfolio analytics platform
Northbridge Financial's legacy analytics dashboard could not keep up with the data volume and speed their analysts needed during market hours. We rebuilt it as a real-time streaming platform with sub-second response times.
5x
throughput increase
200ms
avg response time
99.9%
uptime
The Challenge
What Northbridge Financial was facing
The legacy dashboard was a monolithic .NET application backed by stored procedures that took 15-30 seconds to return results during peak market hours.
Analysts resorted to exporting data to Excel for ad-hoc analysis because the dashboard could not handle complex queries or custom time ranges without timing out.
The system supported only 15 concurrent users before performance degraded. The firm had 60+ analysts who needed simultaneous access.
Deploying updates required a full maintenance window because the monolith had no separation between the API, UI, and data layers.
Our Approach
How we solved it
We decomposed the monolith into a React frontend backed by a Node.js API layer and a dedicated data service. Market data streams in via WebSocket connections, so dashboards update in real time without polling.
We replaced the stored-procedure-heavy database layer with a combination of TimescaleDB for time-series data and Redis for hot caches, cutting average query time from 15 seconds to under 200 milliseconds.
The frontend was built with a component library that analysts can use to compose custom dashboards, save layouts, and share views with colleagues - replacing the rigid, one-size-fits-all legacy UI.
We implemented a blue-green deployment pipeline so updates ship without maintenance windows, and the API layer scales horizontally to handle 200+ concurrent users.
Key Features
What we delivered
Real-Time Market Data Streaming
WebSocket-based data feeds push market updates to analyst dashboards in real time, eliminating the polling lag that made the old system feel slow.
Custom Dashboard Builder
Analysts compose their own dashboards from a library of widgets: charts, tables, watchlists, heat maps, and alert panels - saved and shareable across the team.
Sub-200ms Query Engine
TimescaleDB for time-series analytics and Redis caching deliver query results in under 200ms, even for complex multi-asset, multi-year analyses.
Horizontal Scaling
The API layer auto-scales to handle 200+ concurrent users during market open without performance degradation.
Automated Alerting
Analysts set threshold-based and pattern-based alerts on any metric. Notifications fire via in-app, email, and Slack within seconds of trigger conditions.
Audit and Compliance Logging
Every data access, export, and configuration change is logged for regulatory compliance with SOC 2 and internal audit requirements.
Our Process
How we got there
Legacy System Audit
We mapped every stored procedure, data source, and analyst workflow in the existing system to ensure nothing was lost in the rebuild.
Architecture Design
We designed the decoupled architecture: React frontend, Node.js API, TimescaleDB, Redis, and WebSocket streaming layer - validated with the engineering team before writing code.
Incremental Build
We built the platform in two-week sprints, delivering a working dashboard after sprint two and adding features iteratively based on analyst feedback.
Data Migration
We migrated five years of historical market data into TimescaleDB with validation checks to ensure data integrity across every asset class and time granularity.
Parallel Run and Cutover
Both systems ran in parallel for three weeks. Analysts compared outputs, reported discrepancies, and validated the new platform before the legacy system was decommissioned.
Results
Measurable outcomes
5x Throughput Increase
The platform handles 5x the data volume of the legacy system while maintaining sub-second response times during peak market hours.
200ms Average Response Time
Dashboard queries that previously took 15-30 seconds now return in under 200 milliseconds, transforming the analyst workflow.
99.9% Uptime
The platform maintained 99.9% uptime over its first year, including zero unplanned outages during market hours.
Technology Stack
What we used
Frontend
Backend
Data Layer
Infrastructure
DevOps
Business Impact
The bigger picture
The new platform enabled Northbridge to onboard 40 additional analysts without infrastructure upgrades, directly supporting the firm's expansion into two new asset classes. Analyst productivity increased measurably: the average time to generate a client portfolio report dropped from 2 hours to 20 minutes, and the firm eliminated its dependency on Excel-based workarounds entirely.
“Our analysts went from waiting 30 seconds for a dashboard to load to having real-time data at their fingertips. The team will not go back.”
Have a similar challenge?
Tell us what you're working on. We'll come back with a clear plan and honest technical guidance.