py-candlekit 1.0.1

A lightweight, pandas-friendly Python library for detecting candlestick patterns in financial time-series data, based on NISON book.

Py-CandleKit

A lightweight, pandas-friendly Python library for detecting candlestick patterns in financial time-series data.

✨ Features

• 20+ Classic Patterns: Detect Hammer, Engulfing, Morning Star, and more
• Pandas Integration: Works directly with pandas DataFrames (OHLC format)
• Clear Classification: Bullish/bearish/neutral signal identification
• Academic Foundation: Based on Nison's "Japanese Candlestick Charting Techniques"
• Production Ready: Comprehensive testing and type hints
• Easy Extension: Simple API for adding custom patterns

📊 Supported Patterns

Single-Candle Patterns

• Doji - Indecision (neutral)
• Hammer - Bullish reversal after downtrend
• Shooting Star - Bearish reversal after uptrend
• Bullish/Bearish Marubozu - Strong conviction with minimal shadows
• Bullish/Bearish Belt Hold - Strong opening at extreme

Two-Candle Patterns

• Bullish/Bearish Engulfing - Reversal patterns
• Piercing Line / Dark Cloud Cover - Counter-trend reversals
• Harami Cross - Reversal with doji
• Kicking - Reversal with marubozu gap

Three-Candle Patterns

• Morning/Evening Star - Classic reversal patterns
• Morning/Evening Doji Star - Reversal with doji confirmation
• Three White Soldiers - Strong bullish continuation
• Three Black Crows - Strong bearish continuation
• Three Inside/Outside Up - Bullish reversal variations

Five-Candle Patterns

• Rising/Falling Three Methods - Continuation patterns
• Mat Hold - Bullish continuation with pullback

📦 Installation

pip install py-candlekit

🚀 Quick Start

import pandas as pd
from candlekit import scan_symbol_df, CandlePatterns

# Your OHLC data (lowercase column names required)
data = {
    'open':  [100, 102, 101, 103, 105],
    'high':  [103, 104, 102, 106, 107],
    'low':   [99,  101, 100, 102, 104],
    'close': [102, 101, 102, 105, 106]
}
df = pd.DataFrame(data)

# Scan for all patterns
results_df = scan_symbol_df(df)
print(results_df)

Output:

   index        pattern_name signal_type  candles
0      1              Hammer     bullish        1
1      3   Bullish Engulfing     bullish        2

📚 Basic Usage

Detect Specific Pattern

from candlekit import detect_pattern_at_index

# Check if index 1 is a Hammer
is_hammer = detect_pattern_at_index(CandlePatterns.Hammer, df, index=1)
print(f"Hammer detected: {is_hammer}")  # True

Get Results as List

from candlekit import scan_symbol

results_list = scan_symbol(df)
# Returns: [(1, 'Hammer', 'bullish'), (3, 'BullishEngulfing', 'bullish')]

Scan Specific Patterns Only

# Scan only for reversal patterns
reversal_patterns = [
    CandlePatterns.Hammer,
    CandlePatterns.ShootingStar,
    CandlePatterns.BullishEngulfing,
    CandlePatterns.MorningStar
]

results = scan_symbol_df(df, patterns=reversal_patterns)

🔧 Working with Real Data

Using yfinance (Optional)

import yfinance as yf
from candlekit import scan_symbol_df

# Download data
ticker = yf.Ticker("AAPL")
df = ticker.history(period="1mo", interval="1d")

# Ensure correct column names
df.columns = [col.lower() for col in df.columns]

# Scan for patterns
patterns = scan_symbol_df(df)

# Filter recent bullish patterns
recent_bullish = patterns[
    (patterns['signal_type'] == 'bullish') & 
    (patterns['index'] > len(df) - 5)
]

Using CSV Data

import pandas as pd
from candlekit import scan_symbol_df

# Load from CSV (ensure column names match)
df = pd.read_csv('your_data.csv')
df.columns = ['open', 'high', 'low', 'close', 'volume']  # Rename if needed

# Scan for patterns
results = scan_symbol_df(df)

⚙️ Advanced Configuration

Custom Pattern Parameters

# Adjust sensitivity for Doji detection
detect_pattern_at_index(
    CandlePatterns.Doji, 
    df, 
    index=0, 
    max_body_ratio=0.15  # Default: 0.1
)

# Custom Hammer parameters
detect_pattern_at_index(
    CandlePatterns.Hammer,
    df,
    index=1,
    max_body_ratio=0.3,           # Default: 0.25
    min_lower_wick_to_body=1.5,   # Default: 2.0
    max_upper_wick_ratio=0.4      # Default: 0.33
)

Pattern-Specific Parameters

Each pattern has sensible defaults, but you can adjust:

• max_body_ratio: Maximum body-to-total-length ratio (for thin candles)
• min_body_ratio: Minimum body-to-total-length ratio (for thick candles)
• min_lower_wick_to_body: For Hammer pattern
• min_upper_wick_to_body: For Shooting Star pattern

🧠 Understanding CandleStick Objects

The library uses CandleStick objects internally. You can access their properties:

from candlekit.src.candlekit.entity import CandleStick

candle = CandleStick(df, index=0)
print(f"Body length: {candle.body_length}")
print(f"Top wick: {candle.top_wick}")
print(f"Bottom wick: {candle.bottom_wick}")
print(f"Body ratio: {candle.body_ratio:.2%}")
print(f"Is bullish: {candle.is_bullish}")

📊 Complete Pattern List

Access All Patterns

from candlekit import CandlePatterns

# List all available patterns
for pattern in CandlePatterns:
    print(f"{pattern.pattern_name}: {pattern.signal_type} ({pattern.candles} candle(s))")

Pattern Categories

# Single-candle patterns
single_candle = [p for p in CandlePatterns if p.candles == 1]

# Multi-candle patterns
multi_candle = [p for p in CandlePatterns if p.candles > 1]

# Bullish patterns
bullish = [p for p in CandlePatterns if p.signal_type == 'bullish']

# Bearish patterns
bearish = [p for p in CandlePatterns if p.signal_type == 'bearish']

🧪 Testing & Quality

Run Tests

# Install test dependencies
pip install pytest

# Run all tests
python -m pytest

# Run specific test file
python -m pytest tests/test_patterns.py -v

# Run with coverage
python -m pytest --cov=candlekit tests/

Example Test

import pytest
import pandas as pd
from candlekit import detect_pattern_at_index, CandlePatterns

def test_hammer_detection():
    """Test Hammer pattern detection"""
    df = pd.DataFrame({
        'open': [100],
        'high': [105],
        'low': [95],
        'close': [99]
    })
    
    # This is NOT a hammer (body too large relative to wick)
    result = detect_pattern_at_index(CandlePatterns.Hammer, df, index=0)
    assert result == False
    
    # This IS a hammer
    df = pd.DataFrame({
        'open': [95],
        'high': [98],
        'low': [80],
        'close': [96]
    })
    result = detect_pattern_at_index(CandlePatterns.Hammer, df, index=0)
    assert result == True

🛠️ Development

Setup Development Environment

# Clone and install
git clone https://github.com/zhirodadkhah/candlekit.git
cd candlekit
pip install -e ".[dev]"

# Run code quality tools
flake8 src/candlekit      # Linting
black src/candlekit       # Formatting
mypy src/candlekit        # Type checking

Adding New Patterns

1. Add detection function in patterns.py
2. Add to CandlePatterns enum in utils.py
3. Write comprehensive tests

Example:

# In patterns.py
def is_tweezers_top(candle1: CandleStick, candle2: CandleStick) -> bool:
    """Detect Tweezer Top pattern."""
    return (
        candle1.high == candle2.high and
        candle1.is_bullish and
        not candle2.is_bullish
    )

# In utils.py
CandlePatterns.TweezerTop = (2, is_tweezers_top, "bearish", "Tweezer Top")

📖 Pattern References & Methodology

All patterns are implemented according to Steve Nison's "Japanese Candlestick Charting Techniques, 2nd Edition". Each pattern function includes page references to the textbook.

Key Concepts

• Body Ratio: body_length / total_length - measures conviction
• Window Gap: Price gap between candles with no overlap
• Engulfment: Current candle's body completely contains previous
• Containment: Candle's body is within another's range

Academic Sources

• Nison, S. (2001). Japanese Candlestick Charting Techniques
• Bulkowski, T. N. (2005). Encyclopedia of Candlestick Charts
• Morris, G. L. (1995). Candlestick Charting Explained

🤝 Contributing

We welcome contributions! Please:

1. Fork the repository
2. Create a feature branch: git checkout -b feature/amazing-pattern
3. Add tests for new functionality
4. Ensure tests pass: pytest
5. Update documentation as needed
6. Submit a Pull Request

Contribution Ideas

• Add more candlestick patterns
• Improve performance for large datasets
• Add visualization helpers
• Create Jupyter notebook examples
• Add pattern combination analysis

🙏 Acknowledgments

• Steve Nison for bringing candlestick analysis to markets
• The pandas and numpy communities
• All contributors and users of this library
• Financial analysts and quants who provided feedback

⚠️ Disclaimer

IMPORTANT: This library is for educational and research purposes only.

• Trading involves substantial risk of loss
• Past performance is not indicative of future results
• Always conduct your own research and consult financial advisors
• The authors are not responsible for any trading losses

Use this tool as part of a comprehensive trading strategy, not as standalone advice.

---

Happy analyzing! 📈

Found a bug or have a feature request? Please open an issue on GitHub!