EarningAI Coach

An AI-powered financial literacy and risk-aware trading assistant designed to help beginners and low-income users learn trading safely using simulation, explainable AI, and risk control mechanisms

Comment

Inspiration

In today's fast-moving financial markets, retail investors are often left behind by institutional players who have access to multi-million dollar terminals and real-time AI analysts. We were inspired to bridge this gap. We wanted to build a platform that doesn't just show "red and green numbers," but actually explains what is happening in plain English.

The goal was to create a "Bloomberg for the rest of us"—a tool that combines live market data, generative AI for earnings analysis, and interactive 3D visualizations to make sense of global geopolitical risks.

What it does

EarningAI Coach is an all-in-one financial intelligence suite designed for the modern investor. It features:

  • AI Analyst Chatbot: A conversational partner that reads through complex earnings data and provides instant summaries, buy/sell recommendations, and risk assessments.
  • Live Market Snapshot: A real-time dashboard showing the "pulse" of the market with live price updates and trend sparklines.
  • 3D Geopolitical Risk Map: An interactive, spinning globe that visualizes global economic health, allowing users to see how international events affect their investments.
  • Intelligent Portfolio Tracker: A smart notebook that automatically calculates your realized and unrealized Profit & Loss (P&L) using real-time market valuations.

How we built it

EarningAI Coach is built on a high-performance full-stack architecture:

  • Backend: Powered by FastAPI (Python) for rapid data processing and API serving.
  • Frontend: A modern React application built with Vite, focusing on a high-fidelity "Dark Terminal" aesthetic.
  • AI Engine: We leveraged the Groq API (Llama 3.1-8b) for ultra-low latency natural language processing.
  • Data Sourcing: yfinance serves as our backbone for live market quotes and historical financial statements.
  • Visualizations: We used Three.js and react-globe.gl for the interactive 3D elements, and the TradingView Widget for professional-grade charting.

Challenges we ran into

  1. The "Hallucination" Hurdle: Financial data requires 100% accuracy. We implemented a strict "Grounding" layer to ensure the AI only uses verified data from our APIs rather than its internal training data.
  2. State Management: Synchronizing a 3D globe, live price feeds, and a chatbot simultaneously required advanced React state management to prevent UI lag.
  3. Data Parsing: Extracting clean insights from raw, often messy, earnings reports from various global exchanges was a significant technical obstacle.

Accomplishments that we're proud of

  • Near-Instant AI: Thanks to Groq's LPU architecture, our AI Analyst responds in milliseconds, making the conversation feel truly natural.
  • Visual Narrative: Successfully mapping abstract geopolitical risk data onto an interactive 3D globe that users can actually "play" with.
  • Mathematical Precision: Building a robust portfolio engine that handles the math of multiple transactions accurately: $$ P&L_{unrealized} = \sum_{i=1}^{n} (Price_{current, i} - Price_{average_buy, i}) \times Quantity_{i} $$

What we learned

  • Real-time Data Orchestration: We mastered the flow of high-frequency data between the backend and frontend.
  • User-Centric AI Design: We learned how to "prompt engineer" specifically for financial contexts to ensure the AI remains objective and helpful.
  • Geospatial UI: Gained deep experience in using Three.js to create interactive data visualizations that simplify complex global information.

What's next for EarningAI Coach

Built With

Share this project:

Updates