📈 Stock Price Predictor

A complete end-to-end Machine Learning system that predicts whether a stock's closing price will go UP ↑ or DOWN ↓ the next trading day.

Status: ✅ Production Ready | Version: 2.0 + Phase 1 | Last Updated: May 2026

🎯 Quick Overview

This project combines real-world ML best practices with financial prediction to create an intelligent trading signal system:

10+ Production Modules: 5,600+ lines of production code
40+ Technical Indicators: Comprehensive market feature engineering
Multiple ML Models: Logistic Regression, Random Forest, XGBoost, LSTM + Ensemble
11 Streamlit UI Tabs: Interactive web application for analysis and visualization
Complete Backtesting: Historical performance validation with detailed metrics
Smart Decision Making: BUY/SELL/NO TRADE with confidence scoring (0-100%)
Multi-Channel Alerts: Email, SMS, console notifications
Professional Exports: CSV, Excel, JSON, HTML reporting
Market Intelligence: VIX awareness, earnings calendar, global market context

🚀 Quick Start

Installation (5 minutes)

# Clone/navigate to project
cd c:\Projects\Stocks_predication

# Create virtual environment
python -m venv venv
venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

Run Application

streamlit run app.py

The app opens at http://localhost:8501

📊 Features & Capabilities

Signal Generation

✅ Weighted ensemble model (XGBoost 50%, RF 20%, MLP 20%, LR 10%) ✅ Dynamic confidence scoring (0-100%) ✅ 23 optimized market-aware features ✅ VIX-adjusted for market conditions ✅ Earnings-aware to avoid high-risk dates

Decision Making

✅ Strong Buy (>75% confidence) ✅ Buy (60-75% confidence) ✅ No Trade (<60% confidence) ✅ Risk levels: LOW / MEDIUM / HIGH ✅ Position sizing recommendations

Insights & Transparency

✅ Complete explanations (why the signal occurred) ✅ Supporting indicators and evidence ✅ Beginner-friendly & advanced analysis ✅ Indicator tooltips and definitions

Validation & Backtesting

✅ Historical trade simulation ✅ Win rate, P&L, Sharpe ratio, max drawdown ✅ Trade-by-trade detailed analysis ✅ Performance comparison across time periods

Alerts & Notifications

✅ Real-time console alerts ✅ Email notifications (Gmail, custom SMTP) ✅ SMS alerts (Twilio-ready) ✅ Alert history tracking

Professional Exports

✅ CSV (all data and results) ✅ Excel workbooks (multi-sheet) ✅ JSON (API-friendly) ✅ HTML (client-ready reports)

📚 Project Structure

Stocks_predication/
├── README.md                          ← Start here!
├── PROJECT_INFORMATION.txt            ← Complete documentation
├── requirements.txt                   ← Python dependencies
├── main_pipeline.py                   ← Script-based interface
│
├── 🎨 USER INTERFACES
├── app.py                             ← Main Streamlit app (11 tabs)
├── streamlit_app.py                   ← Alternative UI
│
├── 📦 SOURCE CODE (src/)
│   ├── DATA PIPELINE
│   ├── data_fetching.py               ← Yahoo Finance integration
│   ├── preprocessing.py               ← Data cleaning
│   ├── technical_indicators.py        ← 40+ indicators
│   ├── candlestick_patterns.py        ← Pattern detection
│   ├── sentiment_analysis.py          ← FinBERT sentiment
│   │
│   ├── MACHINE LEARNING
│   ├── model.py                       ← Baseline models
│   ├── model_training_v2.py           ← v2.0 ensemble
│   ├── lstm_neural_network.py         ← Deep learning
│   ├── evaluation.py                  ← Metrics & evaluation
│   ├── ablation_study.py              ← Feature importance
│   │
│   ├── DECISION INTELLIGENCE
│   ├── feature_optimization.py        ← 23 optimized features
│   ├── decision_engine.py             ← Trading decisions
│   ├── market_analysis.py             ← Market context
│   ├── vix_analyzer.py                ← Volatility analysis
│   ├── explanation_engine.py          ← Signal explanations
│   │
│   ├── EXECUTION & VALIDATION
│   ├── backtester.py                  ← Historical simulation
│   ├── alerts_manager.py              ← Notifications
│   ├── export_utils.py                ← Data exports
│   └── visualization.py               ← Interactive charts
│
├── 📓 NOTEBOOKS
├── notebooks/stock_prediction_pipeline.ipynb
│
└── 📁 DATA & MODELS
    ├── data/                          ← Data storage
    ├── models/                        ← Saved models
    └── exports/                       ← Export outputs

📖 How to Use

Via Streamlit Web App

Configure (left sidebar):
- Enter stock symbol (e.g., AAPL, RELIANCE.NS, TCS.NS)
- Select date range (default: 3 years)
- Choose models to run
Analyze (click "🚀 Fetch & Analyze"):
- Fetches historical data
- Calculates indicators
- Trains ML models
- Generates predictions
Explore Results (11 tabs):
- 📊 Price Chart: Interactive candlestick
- 📈 Technical Indicators: RSI, MACD, Bollinger Bands
- 🎯 Predictions: UP/DOWN with confidence
- 📉 Model Performance: Accuracy metrics
- 🔍 Feature Analysis: Feature importance
- 💬 Sentiment Analysis: News sentiment
- 🕯️ Candlestick Patterns: Pattern detection
- 📥 Export Results: CSV/Excel/JSON
- 🔔 Alerts: Set up notifications
- 📅 Earnings Calendar: Risk awareness
- 📊 VIX Analysis: Market volatility

Via Python Script

from src.data_fetching import fetch_stock_data
from src.feature_engineering import add_technical_indicators, create_target
from src.model_training_v2 import WeightedEnsemble

# 1. Fetch data
df = fetch_stock_data('AAPL', '2023-01-01', '2024-01-01')

# 2. Add indicators
df = add_technical_indicators(df)

# 3. Create target
df = create_target(df)

# 4. Train model
model = WeightedEnsemble()
model.fit(X_train, y_train)

# 5. Predict
predictions = model.predict(X_test)
confidence = model.predict_proba(X_test)

🔬 Technical Details

Data Pipeline

Raw Stock Data
    ↓ (fetch from Yahoo Finance)
Historical OHLCV Data
    ↓ (clean, handle missing values)
Preprocessing
    ↓ (add 40+ technical indicators)
Feature Engineering
    ↓ (create UP/DOWN target)
Target Generation
    ↓ (train-test split, normalize)
ML Training
    ↓ (XGBoost, RF, LSTM, LR ensemble)
Predictions & Signals
    ↓ (confidence scoring, decision making)
Trading Signals (BUY/SELL/HOLD)

Technical Indicators (40+)

Trend Analysis:

Simple & Exponential Moving Averages (5, 10, 20 day)
MACD (Moving Average Convergence Divergence)
ADX (Average Directional Index)

Momentum:

RSI (Relative Strength Index)
Stochastic Oscillator
Daily Returns

Volatility:

Bollinger Bands (Upper, Lower, Width, Position)
ATR (Average True Range)
Standard Deviation

Volume:

Volume Change %
On-Balance Volume (OBV)
Chaikin Money Flow (CMF)

Price Structure:

Close-Open Difference
Price vs Moving Average Distance
High-Low Range

Machine Learning Models

Model	Type	Strengths	Use Case
Logistic Regression	Linear	Fast, interpretable, good baseline	Quick predictions, feature analysis
Random Forest	Ensemble	Non-linear, feature importance	General-purpose predictions
XGBoost	Boosting	State-of-the-art, handles complexity	High-accuracy predictions
LSTM	Deep Learning	Temporal dependencies	Complex pattern recognition
Ensemble (v2.0)	Combined	Best of all (weighted voting)	Final production predictions

v2.0 Ensemble Weighting:

XGBoost: 50% (primary model)
Random Forest: 20%
MLP/LSTM: 20%
Logistic Regression: 10%

Evaluation Metrics

Classification Metrics:

Accuracy, Precision, Recall, F1-Score
ROC-AUC (Receiver Operating Characteristic)
Confusion Matrix (TP, FP, TN, FN)

Business Metrics:

Win Rate: % of profitable trades
Sharpe Ratio: Risk-adjusted returns
Max Drawdown: Largest peak-to-trough decline
P&L: Total profit/loss

📋 Supported Stocks

Indian Stocks (NSE)

RELIANCE.NS, TCS.NS, INFY.NS, WIPRO.NS, HDFC.NS, ICICIBANK.NS, AXISBANK.NS, SBIN.NS, MARUTI.NS, BAJAJFINSV.NS

US Stocks (NASDAQ/NYSE)

AAPL, MSFT, GOOGL, AMZN, TSLA, META, NVIDIA, JPM, V, WMT

Note: Any stock supported by Yahoo Finance can be used

🎯 Performance (v2.0 Improvements)

Metric	Before (v1.0)	After (v2.0)	Improvement
Overall Accuracy	60%	60-65%	✅ +5%
Precision	54.8%	65-70%	✅ +10-15%
False Positive Rate	~50%	~30%	✅ -40%
Features	54 (noisy)	23 (optimized)	✅ Better quality
Confidence Threshold	0.5	0.6	✅ Quality filter

🛠️ Dependencies

Data: pandas>=2.2.0, numpy>=2.0.0, scipy>=1.12.0
ML: scikit-learn>=1.4.0, xgboost>=2.0.0, tensorflow>=2.14.0
Finance: yfinance>=0.2.35, pandas-ta>=0.3.14b0
NLP: transformers>=4.36.0 (FinBERT)
Viz: matplotlib>=3.8.4, plotly>=5.18.0
Web: streamlit>=1.31.0

See requirements.txt for complete list.

⚠️ Important Disclaimers

This is a PREDICTION tool, not financial advice
Historical performance ≠ Future results
Always use proper risk management (stop losses, position sizing)
Start with paper trading before real capital
Monitor model performance continuously
Be aware of earnings dates and major news events
Combine signals with fundamental analysis

📚 Documentation

PROJECT_INFORMATION.txt ← Complete documentation (start here!)
ARCHITECTURE.md ← System design and data flow
MODEL_AND_ALGORITHM_GUIDE.md ← Detailed ML explanations
API_REFERENCE.md ← Function documentation
QUICKSTART.md ← Quick start guide
TROUBLESHOOTING.md ← Common issues

🐛 Troubleshooting

"No module named 'src'"

import sys
sys.path.insert(0, 'src')

"yfinance: No data found"

Check symbol format (e.g., RELIANCE.NS for NSE stocks)

"Streamlit connection reset"

Clear browser cache or restart terminal

"Model accuracy is low"

Check if market is sideways (low ADX)
Verify feature scaling

See TROUBLESHOOTING.md for more solutions.

🗓️ Roadmap

Phase 2: Multi-stock portfolio optimization, options strategies Phase 3: Deep reinforcement learning, sector rotation, crypto support Phase 4: Federated learning, advanced NLP, FastAPI deployment

💡 Key Learnings

Time-Series Data: Never shuffle! Use TimeSeriesSplit for proper validation
Feature Engineering: Quality over quantity (54 → 23 features improved accuracy)
Ensemble Models: Different models excel in different market regimes
Confidence Scoring: Threshold reduces false positives significantly
Market Context: VIX integration and earnings awareness improve decisions

📞 Support

Issues: Check TROUBLESHOOTING.md
Questions: Review API_REFERENCE.md or ARCHITECTURE.md
Examples: See notebooks/stock_prediction_pipeline.ipynb

Name		Name	Last commit message	Last commit date
notebooks		notebooks
src		src
.gitignore		.gitignore
ARCHITECTURE.md		ARCHITECTURE.md
FILE_GUIDE.md		FILE_GUIDE.md
MODEL_AND_ALGORITHM_GUIDE.md		MODEL_AND_ALGORITHM_GUIDE.md
PROJECT_INFORMATION.txt		PROJECT_INFORMATION.txt
QUICKSTART.md		QUICKSTART.md
README.md		README.md
SENTIMENT_ANALYSIS_SETUP.md		SENTIMENT_ANALYSIS_SETUP.md
SETUP_GUIDE.md		SETUP_GUIDE.md
STREAMLIT_INTEGRATION_TEST.py		STREAMLIT_INTEGRATION_TEST.py
TROUBLESHOOTING.md		TROUBLESHOOTING.md
app.py		app.py
main_pipeline.py		main_pipeline.py
streamlit_app.py		streamlit_app.py

Latest commit

History

Folders and files

Repository files navigation