About
Work
Blog
Projects

Nexus — Backtesting Platform

End-to-end backtesting platform for financial traders that I architected and shipped as sole developer.Next.js · Node.js · PostgreSQL · Docker
1
engineer, end to end
6
screens designed and shipped
Nexus — Backtesting Platform
June 1, 2024

Overview

I architected and built the Nexus backtesting platform from the ground up, owning UI/UX design, frontend implementation, and backend engineering. The platform provides traders with a reliable environment to test strategies against historical market data.

Key Contributions

  • Designed the overall system architecture, including data ingestion, storage, and compute for backtesting large datasets.
  • Implemented a performant front-end UI to configure, run, and visualize backtests.
  • Built backend services to execute strategies, persist results, and provide analytics.
  • Deployed the platform and maintained CI/CD for ongoing releases.

Technologies

  • Next.js, React
  • Node.js
  • PostgreSQL (or MongoDB depending on dataset needs)
  • Containerized deployment (Docker)

Outcome

Delivered a robust and feature-rich application that enabled traders to iterate quickly on strategies with accurate historical simulations.

Problem

Traders lacked a reliable environment to validate strategies at scale. Existing tools were either too slow for large datasets or required extensive manual configuration.

Solution

I designed a system optimized for large-scale backtesting: efficient data ingestion and a compute layer that executed user strategies in isolated, reproducible runs. The UI made it simple to configure experiments and compare results.

Architecture

Text
Data Ingest -> Partitioned Time-Series Storage -> Compute Cluster (worker pool) -> Results Store -> UI

Key features

  • Fast data ingestion and normalization pipelines
  • Strategy sandboxing and deterministic execution
  • Rich visualizations (equity curves, drawdown, heatmaps)
  • Compare multiple runs side-by-side and export results

Screenshots / Media

Nexus screenshot 1
Nexus screenshot 2
Nexus screenshot 3
Nexus screenshot 4
Nexus screenshot 5
Nexus screenshot 6

Metrics

  • Handled backtests over datasets with millions of rows with acceptable latency
  • Reduced experiment turnaround time from hours to minutes for many strategy types

Lessons learned

  • Determinism matters: ensuring runs are reproducible simplified debugging and user trust.
  • A small compute cluster with good caching provides massive speedups for repeated queries.

Next steps

  • Add distributed backtesting support and cloud-based spot compute for burst workloads.