MarketSphere

Ditch the dreadful spreadsheets and endless article scrolling. Spin the globe instead — MarketSphere turns global markets into a living 3D world you explore, not scroll through.

Comment

Inspiration

We were inspired by the Bloomberg Terminal the industry standard, all-in-one command center that lets Wall Street traders see the entire world's markets and news at a glance. It's an incredible tool, but it's average price is $30,000/year and has a UI built for professionals who already know exactly what they're looking for. We wanted to bring that same Bloomberg-style, world-view terminal experience to the everyday retail investor instead.

That idea became MarketSphere: an interactive 3D globe as your world-view terminal, where spinning to any region surfaces its live markets and news. To help users actually navigate and interpret all of it, we paired the globe with a Gemini-powered LLM assistant grounded in live news and financial data. Ask it a question about what you're looking at and it answers in that context, or ask it, in natural language, to build you a dynamic panel of markets and news on the fly, completely controlled by you.

What it does

MarketSphere is a full-screen interactive 3D globe as your market dashboard. Spin it, click a continent, and the camera swoops in while a live panel slides out with real candlestick charts (yfinance OHLCV, rendered on our own chart) and real news headlines pulled from GDELT, the Guardian, and Alpha Vantage. A Gemini-powered chat assistant lives in in the browser, ask it a question and it gives you an answer grounded in live data, ask it to build something and it assembles a custom workspace on the fly, or ask it to analyze a company and it hands back a real price-vs-narrative analysis.

How we built it

Next.js/TypeScript frontend with react-globe.gl + three.js driving the globe, Framer Motion for the camera/panel choreography, and TradingView's open-source Lightweight Charts for the market panel. FastAPI + Postgres backend, with a news ingestion pipeline pulling from three sources and a Gemini-backed router that classifies each chat message into answer / build / analyze before responding.

Challenges we ran into

We're especially proud of the AI copilot experience. Rather than acting as a generic chatbot, here Gemini is grounded in whatever the user is actively exploring on the platform. They help interpret the signals, explain what the data means, surface relevant context, and answer questions in the specific market and news context the user is viewing.

What we learned

We improved our ability to scope projects by turning large goals into actionable steps, prioritizing tasks, and iterating through smaller milestones instead of trying to solve everything at once. This approach helped us stay organized, identify issues earlier, and build more efficiently as a team.

What's next for MarketSphere

Country/state-level scope selection, a real tone-vs-price overlay directly on the chart (the data model supports it, the UI doesn't yet), persisting chat-built workspaces, and widening news source coverage beyond the current three. Also, we hope AI copilot interaction can change UI is relation to what user asks.

Built With

Share this project:

Updates