StockPredictor-AI

Inspiration

The stock market is complex and often overwhelming, especially for beginners. Most tools either provide raw data or predictions, but very few explain why something is happening. We wanted to build a system that not only predicts stock prices but also explains trends, analyzes news, and helps users make informed decisions using AI.

What it does

StockPredictor-AI is a full-stack web application that:

• 📈 Predicts stock prices using LSTM and ARIMA models
• 🤖 Uses Google Gemini AI to generate human-readable insights
• 📰 Fetches real-time news and performs sentiment analysis
• ⚠️ Classifies risk levels based on trends and news
• 💬 Includes an AI chatbot for user queries
• 📊 Provides an interactive dashboard with charts and watchlist

It transforms raw financial data into clear, understandable insights.

How we built it

• Frontend: React.js for a responsive fintech dashboard
• Backend: FastAPI for handling APIs and ML integration

• Machine Learning:

  • LSTM for time-series prediction
  • ARIMA for statistical forecasting

• AI Integration: Google Gemini API for:

  • Insights generation
  • Prediction explanation
  • Sentiment analysis

• Data Sources:

  • Yahoo Finance (yfinance) for stock data
  • NewsAPI for real-time news

• Database: PostgreSQL for user and watchlist storage

Challenges we ran into

• Integrating real-time data with ML models efficiently
• Designing prompts for Gemini to return structured, useful insights
• Handling inconsistent news sentiment across multiple sources
• Balancing accuracy vs performance in prediction models
• Creating a UI that is both simple and information-rich

Accomplishments that we're proud of

• Successfully combining ML + AI + real-time news in one platform
• Building a system that focuses on explainability, not just prediction
• Implementing end-to-end functionality from login to insights
• Designing a clean, modern fintech-style dashboard
• Creating a project that is both technically strong and user-friendly

What we learned

• How to integrate machine learning models into real applications
• Effective use of prompt engineering with Gemini AI
• Handling API orchestration and data pipelines
• Importance of UX in technical projects
• Real-world limitations of stock prediction models

What's next for StockPredictor-AI

• 📡 Real-time stock updates using WebSockets
• 📊 Advanced indicators (RSI, MACD, Bollinger Bands)
• 🧠 Improved sentiment scoring with historical correlation
• 📱 Mobile app version
• 🌍 Support for more global markets and assets (crypto, forex)
• 🤖 Smarter AI assistant with personalized recommendations

Built With

• css-frameworks:-react.js-(frontend)
• docker
• fastapi-(backend)-machine-learning:-tensorflow-(lstm)
• github-(version-control)-tools:-git
• html
• insights)-apis:-yahoo-finance-(via-yfinance)
• javascript
• keras
• languages:-python
• newsapi-databases:-postgresql-authentication:-google-oauth-2.0-visualization:-chart.js-/-recharts-platforms:-replit-(development-and-deployment)
• scikit-learn
• sentiment-analysis
• statsmodels-(arima)-ai-services:-google-gemini-api-(text-generation