📈 Stock Price Forecasting for Top 10 Indian Stocks using ML Models

This project performs time-series forecasting on the Top 10 Indian stocks using three different models:

• Auto-ARIMA
• Random Forest Regressor
• LSTM (Deep Learning)

Each model is evaluated using RMSE, and the best-performing model is selected for every stock.

---

🚀 Project Overview

Financial time-series data is noisy, non-linear, and volatile. This project benchmarks statistical, machine learning, and deep learning models to identify the most effective approach for 30-day stock price forecasting.

---

🔹 Forecast Horizon

• 30 Trading Days

🔹 Top 10 Stocks Analyzed

• TATAMOTORS
• SBIN
• ICICIBANK
• VEDL
• ITC
• HINDALCO
• TATASTEEL
• RELIANCE
• ZEEL
• AXISBANK

---

📊 Forecast Visualizations

Legend

• 🔵 Blue Line → Historical Closing Prices
• 🔴 Red Dashed Line → 30-Day Forecast

---

🧠 Models Used

1️⃣ Auto-ARIMA

• Traditional statistical time-series model
• Captures trend and seasonality
• Struggles with high volatility

2️⃣ Random Forest Regressor

• Ensemble machine learning model
• Handles non-linearity and feature interactions well
• Strong performance across most stocks

3️⃣ LSTM (Long Short-Term Memory)

• Deep learning model for sequential data
• Effective for long-term dependencies
• Requires more data and tuning

---

📊 Model Performance Comparison (RMSE)

Index	Stock	Auto-ARIMA RMSE	Random Forest RMSE	LSTM RMSE	Best Model
0	TATAMOTORS	293.03	7.19	13.46	Random Forest
1	SBIN	57.82	8.23	17.55	Random Forest
2	ICICIBANK	145.15	10.56	17.50	Random Forest
3	VEDL	80.72	5.68	22.64	Random Forest
4	ITC	42.01	4.83	31.07	Random Forest
5	HINDALCO	57.54	20.90	11.85	LSTM
6	TATASTEEL	153.24	14.70	22.23	Random Forest
7	RELIANCE	489.57	43.00	80.18	Random Forest
8	ZEEL	1561.85	13.47	16.47	Random Forest
9	AXISBANK	175.61	15.19	32.59	Random Forest

---

🏆 Key Results & Insights

• Random Forest outperformed all models for 9 out of 10 stocks
• LSTM performed best only for HINDALCO
• Auto-ARIMA showed significantly higher error due to:
  • Market volatility
  • Non-stationary behavior
• Machine learning models handled:
  • Non-linear patterns
  • Sudden price movements
  • Feature interactions more effectively

---

📌 Final Conclusion

Random Forest is the most reliable and consistent model for short-term stock price forecasting across multiple Indian equities.

---

🛠️ Tech Stack

• Language: Python
• Libraries:
  • Pandas, NumPy
  • Scikit-learn
  • TensorFlow / Keras (LSTM)
  • Statsmodels / pmdarima (Auto-ARIMA)
  • Matplotlib, Seaborn
• Environment: Jupyter Notebook

---

⚠️ Disclaimer

This project is for educational and research purposes only. It does not constitute financial or investment advice.