Hackathon Track: AGI / Fintech / UX
Built By: Solo Dev (3-Hour Sprint)
Trading isn't about a lack of information; it's about cognitive overload. Every trader has an "edge"—whether it's Wyckoff, Elliott Wave, or obscure Goldbach mathematics—but humans get tired, emotional, and miss setups.
Edge.ai is a Multimodal AI Co-Pilot that speaks your specific technical language.
It ingests entire trading manuals (PDFs), "learns" the strategy instantly via In-Context Learning, and analyzes market data strictly through that lens. It doesn't give generic advice; it clones your discipline.
Upload a 200-page PDF on any trading strategy. The AI reconfigures its entire analysis engine in seconds to identifying patterns, terminology, and risk rules specific to that document.
Experience the "Fractal Reality" of markets. As you zoom in from a Yearly view to a 5-minute view, our engine dynamically renders nested support/resistance grids based on the "Power of 3" math defined in our demo strategy.
Prove the edge without risking capital:
- Spin the Wheel: Load a random slice of historical market data.
- Analyze: Watch the AI construct a narrative and predict the next move based on your uploaded PDF.
- Reveal: Instantly see the future price action to validate the AI's logic.
- Brain: Google Gemini 1.5 Pro (via API) for multimodal reasoning and massive context window ingestion.
- Frontend: React + Vite + Tailwind CSS (Glassmorphism UI).
- Charting: TradingView Lightweight Charts (High-performance rendering).
- Backend: Python (FastAPI) + Pandas (Data Slicing).
- Data: Statistical sampling of ES (S&P 500) Futures.
Prerequisites: Node.js, Python 3.9+, Google AI Studio Key.
git clone [https://github.com/yourusername/edge-ai.git](https://github.com/yourusername/edge-ai.git)
cd edge-ai