Data Eng · Web2026

Global Market Lab

A Bloomberg Terminal-inspired analytics platform covering 84 instruments across equities, bonds, forex, metals, energy and macro. Live FRED, ECB and EIA data sits behind a from-scratch forecasting engine, and every chart is native SVG with no charting library.

Instruments covered
84
Live data sources
3
Forecast models in the ensemble
3
Charting libraries used
0

The problem

Professional terminals are impressive and effectively unavailable. The price is institutional and the internals are opaque. What makes them useful is the live data plumbing, the forecasting mathematics, and an information density most modern web design has abandoned. I wanted to find out whether one person could rebuild those three things honestly.

The approach

Build every layer directly. Real connectors to FRED, ECB and EIA instead of a mock feed. A hand-written maths engine implementing ARIMA, Holt-Winters and GARCH instead of importing a stats package. Native SVG for every chart, which turned out to be the only way to get the density right.

I wanted to see whether one person could rebuild the parts of a professional trading terminal that matter most: the live data, the forecasting maths and the dense information design, without leaning on any UI or charting library.

Why I built it

Architecture

  1. 01

    Live connectors

    FRED, ECB and EIA feeds streaming data for 84 instruments across six asset classes.

  2. 02

    Forecast engine

    ARIMA, Holt-Winters and GARCH implemented from scratch and combined into an ensemble.

  3. 03

    Risk layer

    Monte Carlo VaR/CVaR simulation plus regime detection across the instrument set.

  4. 04

    Render layer

    Every chart is native SVG, including the correlation network graph. No charting or UI library anywhere.

Stack

Frontend
Next.js 16 · TypeScript · native SVG
Data
FRED · ECB · EIA
Models
ARIMA · Holt-Winters · GARCH · Monte Carlo
  • Next.js 16
  • TypeScript
  • FRED API
  • ECB
  • EIA
  • ARIMA
  • GARCH
  • Holt-Winters
  • Monte Carlo
  • SVG

Outcome

A dense, live analytics surface covering 84 instruments, with a forecasting ensemble and risk simulation behind it. The zero-library constraint was the most instructive part of the build. It forced an understanding of both the mathematics and the rendering that importing a package would have hidden.

NextML · Data Eng

Predicting equity impact from physical supply-chain shocks.