This is a submission for the Agent.ai Challenge: Full-Stack Agent (See Details)
What I Built
I developed a Stock Prediction Agent designed to empower traders with AI-driven insights. My agent, Stock Price Predictor, analyzes historical stock data, leverages technical indicators (Bollinger Bands, MACD, RSI), and predicts actionable trading signals (Buy/Sell/Hold) alongside next day price targets (Open, Close, Buy/Sell levels). Built with Flask and machine learning, it transforms complex market data into digestible, confidence-scored recommendations.
Demo
https://agent.ai/agent/stock-price-predictor
How It Works
Step 1: Enter a stock symbol (e.g., IRFC.NS).
Step 2: The Agent fetches historical data, calculates technical indicators, and runs ML models.
Output:
- Action: Buy/Sell/Hold recommendation with confidence score.
- Price Forecasts: Predicted Buy (Low), Sell (High), Open, and Close prices for the next trading session.
Challenge Experience
Building this agent with Flask, XGBoost, and Random Forest was a rewarding deep dive into Stock Price Prediction. Key learnings:
- Balancing imbalanced classes with SMOTE drastically improved prediction accuracy.
- Integrating Yahoo Finance data via yfinance streamlined real-time analysis.
- Deploying the model as an API with error handling ensured scalability.
- While hyperparameter tuning with GridSearchCV was computationally intense, the performance gains justified the effort.
Next step? Adding real-time news for sentiment analysis!
Disclaimer: Predictions are educational and not financial advice. Trade responsibly!
Agent.ai Experience
Agent.ai simplified building and deploying Stock Price Predictor, letting me focus on refining ML models instead of infrastructure. Its intuitive tools made API integration and scaling effortless.
Top comments (1)
I made one ML model to predict the prices of a stock for next trading session as explained above. This still requires some enhancements, Any suggestions for improving this model are greatly appreciated. Thank You