Black Swan Guardian

Anurag & team built Black Swan Guardian: a Gemini 3 engine for Indian stocks that detects extreme events and suggests hedging via a Node.js/React full-stack dashboard.

Comment

Inspiration​As a B.Tech student managing the demands of hostel life and a growing interest in the Indian stock market, I noticed a significant gap for retail investors. While tracking majors like SBI and L&T, I realized that the sheer volume of news and data is overwhelming. Most tools only show "what" is happening, not "why" or how much risk is involved.​I built Black Swan Guardian to act as a student-friendly financial co-pilot. This project is inspired by my passion for teaching others about market dynamics—moving beyond simple price tracking to provide professional-grade, AI-driven risk clarity for the everyday investor.​What it Does​Black Swan Guardian is an intelligent sentinel that monitors your portfolio in real-time:​Dynamic Risk Scoring: Calculates a comprehensive "Risk Score" by merging quantitative volatility data with qualitative AI sentiment.​Market Ratings: Uses Gemini 3 to categorize stocks into actionable labels: Bullish, Bearish, or Strong Buy.​Tiered Intelligence: To democratize access, I implemented a tiered search system—standard users get 2 deep-dives, while verified students get 10 searches to support their learning journey.​Live Market Insights: Provides summarized news feeds that explain the "why" behind price movements using the latest market data.​Polished Dashboard: A high-end UI featuring glassmorphism design and smooth animations to make complex financial data feel intuitive.​How I Built It​The project is built on a modern, high-performance stack centered around the Gemini 3 ecosystem: AI Engine (Gemini 3): I utilized Gemini 3 Pro for deep financial reasoning and Gemini 3 Flash for low-latency market alerts. The system uses Gemini 3 Deep Think to simulate “Black Swan” scenarios and stress-test portfolios. Backend: Developed with Python (FastAPI) to handle high-concurrency requests and financial computations. Frontend: Built using React.js and Tailwind CSS, with custom animation libraries to enhance the data visualization experience. Database: Supabase (PostgreSQL) handles user watchlists and search quota management. Quantitative Logic: I implemented a custom risk engine using the Standard Deviation (σ) of daily returns:Challenges I Faced API Quota Logic: Implementing the backend logic to strictly enforce the 2 vs. 10 search limit while maintaining a smooth user experience required complex state management and validation. Prompt Precision: Fine-tuning Gemini 3 to provide grounded financial ratings (like "Strong Buy") rather than generic advice took multiple iterations of few-shot prompting and system instruction refinement. Real-time Optimization: Balancing the latency of fetching live NSE/BSE data with the processing time of a "Deep Think" AI model required implementing an efficient caching layer and asynchronous API calls. What I Learned This project was a deep dive into the Agentic Future of finance. I learned how to move beyond static chatbots to build a functional tool where Gemini 3 acts as a reasoning engine. I improved my skills in vibe-coding for rapid UI iteration, full-stack security, and the practical application of risk management theory in a live software environment. What's Next for Black Swan Guardian "Talk to Your Portfolio": Using RAG (Retrieval-Augmented Generation) so users can have a natural conversation with their holdings. Predictive Simulations: Adding "What-If" modes to stress-test portfolios against historical crashes using Monte Carlo simulations. Guardian Alerts: Direct integration with WhatsApp and Telegram for instant push notifications when a sentiment shift is detected

Built With

Share this project:

Updates