Pattern Detection Tools

Python packages, MT5 methods, and code examples for automated candlestick pattern recognition.

Python Libraries

MQL5 Examples

60+

Detectable Patterns

Auto

Detection

Python Libraries

TA-Lib

The most comprehensive technical analysis library with 60+ candlestick pattern recognition functions.

pip install TA-Lib61 patterns

✅ Pros

Industry standard
Fast C library
Comprehensive pattern coverage
Well documented

⚠️ Cons

•Complex installation on Windows
•Requires C compiler

Code Example

Python

import talib

import numpy as np

# Get OHLC data

open_prices = np.array([...])

high_prices = np.array([...])

low_prices = np.array([...])

close_prices = np.array([...])

# Detect Hammer pattern

hammer = talib.CDLHAMMER(open_prices, high_prices, low_prices, close_prices)

# Detect Engulfing pattern

engulfing = talib.CDLENGULFING(open_prices, high_prices, low_prices, close_prices)

# Detect Morning Star

morning_star = talib.CDLMORNINGSTAR(open_prices, high_prices, low_prices, close_prices)

# Value: 100 = bullish signal, -100 = bearish signal, 0 = no pattern

pandas-ta

Modern pandas-native technical analysis library with easy-to-use candlestick pattern recognition.

pip install pandas-ta30+ patterns

✅ Pros

Pure Python
Easy installation
Pandas integration
Active development

⚠️ Cons

•Fewer patterns than TA-Lib
•Slower than C-based libraries

Code Example

Python

import pandas as pd

import pandas_ta as ta

# Load your OHLC data into a DataFrame

df = pd.DataFrame({

'open': [...],

'high': [...],

'low': [...],

'close': [...]

})

# Detect Doji pattern

df['doji'] = ta.cdl_doji(df['open'], df['high'], df['low'], df['close'])

# Detect all patterns at once (returns DataFrame with all patterns)

df.ta.cdl_pattern(name="all")

# Specific pattern

df.ta.cdl_pattern(name="engulfing")

mplfinance

Matplotlib-based financial visualization library for creating candlestick charts with pattern annotations.

pip install mplfinanceVisualization only

✅ Pros

Beautiful charts
Easy to use
Great for analysis
Publication quality

⚠️ Cons

•Visualization only, no detection
•Requires matplotlib

Code Example

Python

import mplfinance as mpf

import pandas as pd

# Load OHLC data with DatetimeIndex

df = pd.read_csv('data.csv', index_col=0, parse_dates=True)

# Basic candlestick chart

mpf.plot(df, type='candle', style='charles', title='EURUSD')

# With volume and moving averages

mpf.plot(df, type='candle', volume=True,

mav=(10, 20, 50), style='yahoo')

# Mark patterns with annotations

mpf.plot(df, type='candle',

alines=dict(alines=[('2024-01-15', 1.0850)],

colors=['r'], linewidths=2))

vectorbt

High-performance backtesting library that can incorporate candlestick patterns into trading strategies.

pip install vectorbtVia TA-Lib integration

✅ Pros

Fast vectorized operations
Great backtesting
Pattern strategy testing
Visualization

⚠️ Cons

•Steep learning curve
•Requires TA-Lib for patterns

Code Example

Python

import vectorbt as vbt

import talib

# Get price data

data = vbt.YFData.download('EURUSD=X', period='1y')

ohlc = data.get()

# Detect patterns using TA-Lib

hammer = talib.CDLHAMMER(ohlc['Open'], ohlc['High'],

ohlc['Low'], ohlc['Close'])

# Create entry signals from patterns

entries = hammer == 100 # Bullish hammer

exits = hammer == -100 # Or use other exit logic

# Backtest the pattern strategy

pf = vbt.Portfolio.from_signals(ohlc['Close'], entries, exits)

print(pf.stats())

candlestick-patterns

Lightweight Python library specifically designed for candlestick pattern recognition.

pip install candlestick-patterns20+ patterns

✅ Pros

Lightweight
Easy to use
Pure Python
No dependencies

⚠️ Cons

•Limited patterns
•Less maintained

Code Example

Python

from candlestick import candlestick

# Detect patterns on DataFrame

# Returns DataFrame with boolean columns for each pattern

patterns = candlestick.hammer(df, target='signal')

patterns = candlestick.inverted_hammer(df, target='signal')

patterns = candlestick.doji(df, target='signal')

patterns = candlestick.engulfing(df, target='signal')

MT5 / MQL5 Methods

Built-in Pattern Functions

MQL5 provides functions and techniques to detect candlestick patterns programmatically.

MQL5

// MQL5 - Hammer Pattern Detection

bool IsHammer(int bar) {

double open = iOpen(_Symbol, _Period, bar);

double close = iClose(_Symbol, _Period, bar);

double high = iHigh(_Symbol, _Period, bar);

double low = iLow(_Symbol, _Period, bar);

double body = MathAbs(close - open);

double upperShadow = high - MathMax(open, close);

double lowerShadow = MathMin(open, close) - low;

double totalRange = high - low;

// Hammer criteria:

// 1. Small body (< 30% of total range)

// 2. Long lower shadow (> 2x body)

// 3. Little or no upper shadow (< 10% of range)

if(totalRange == 0) return false;

bool smallBody = (body / totalRange) < 0.30;

bool longLowerShadow = lowerShadow >= (body * 2);

bool noUpperShadow = upperShadow < (totalRange * 0.10);

return smallBody && longLowerShadow && noUpperShadow;

}

Engulfing Pattern Detection

Detect bullish and bearish engulfing patterns in MQL5.

MQL5

// MQL5 - Engulfing Pattern Detection

int DetectEngulfing(int bar) {

double open1 = iOpen(_Symbol, _Period, bar);

double close1 = iClose(_Symbol, _Period, bar);

double open2 = iOpen(_Symbol, _Period, bar + 1);

double close2 = iClose(_Symbol, _Period, bar + 1);

bool prevBearish = close2 < open2;

bool prevBullish = close2 > open2;

bool currBullish = close1 > open1;

bool currBearish = close1 < open1;

// Bullish Engulfing: Previous bearish, current bullish

// Current candle body completely covers previous

if(prevBearish && currBullish) {

if(open1 <= close2 && close1 >= open2) {

return 1; // Bullish Engulfing

}

// Bearish Engulfing

if(prevBullish && currBearish) {

if(open1 >= close2 && close1 <= open2) {

return -1; // Bearish Engulfing

}

return 0; // No pattern

}

Morning/Evening Star Detection

Detect three-candle reversal patterns.

MQL5

// MQL5 - Morning Star Detection

bool IsMorningStar(int bar) {

// First candle (oldest) - Long bearish

double o1 = iOpen(_Symbol, _Period, bar + 2);

double c1 = iClose(_Symbol, _Period, bar + 2);

// Second candle (star) - Small body

double o2 = iOpen(_Symbol, _Period, bar + 1);

double c2 = iClose(_Symbol, _Period, bar + 1);

double h2 = iHigh(_Symbol, _Period, bar + 1);

double l2 = iLow(_Symbol, _Period, bar + 1);

// Third candle - Long bullish

double o3 = iOpen(_Symbol, _Period, bar);

double c3 = iClose(_Symbol, _Period, bar);

double body1 = MathAbs(c1 - o1);

double body2 = MathAbs(c2 - o2);

double body3 = MathAbs(c3 - o3);

bool firstBearish = c1 < o1 && body1 > body2 * 2;

bool smallStar = body2 < body1 * 0.5;

bool thirdBullish = c3 > o3 && body3 > body2 * 2;

bool closesIntoFirst = c3 > (o1 + c1) / 2;

return firstBearish && smallStar && thirdBullish && closesIntoFirst;

}

Pattern Scanner EA

Complete Expert Advisor framework for scanning multiple patterns.

MQL5

// MQL5 - Pattern Scanner Framework

enum SIGNAL { NONE, BULLISH, BEARISH };

SIGNAL ScanPatterns(int bar) {

// Check reversal patterns

if(IsHammer(bar)) return BULLISH;

if(IsShootingStar(bar)) return BEARISH;

int engulf = DetectEngulfing(bar);

if(engulf == 1) return BULLISH;

if(engulf == -1) return BEARISH;

if(IsMorningStar(bar)) return BULLISH;

if(IsEveningStar(bar)) return BEARISH;

// Add more patterns...

return NONE;

}

void OnTick() {

// Scan for patterns on the last completed candle

SIGNAL signal = ScanPatterns(1);

if(signal == BULLISH && PositionsTotal() == 0) {

// Open buy position

double sl = iLow(_Symbol, _Period, 1) - 10 * _Point;

double tp = Ask + 2 * (Ask - sl);

trade.Buy(0.1, _Symbol, Ask, sl, tp);

}

📋 TA-Lib Pattern Functions Reference

Complete list of candlestick pattern functions available in TA-Lib:

Reversal Patterns

CDLHAMMER

CDLINVERTEDHAMMER

CDLSHOOTINGSTAR

CDLHANGINGMAN

CDLENGULFING

CDLMORNINGSTAR

CDLEVENINGSTAR

CDLPIERCING

CDLDARKCLOUDCOVER

CDLHARAMI

CDLHARAMICROSS

Continuation

CDLRISEFALL3METHODS

CDLTASUKIGAP

CDLSEPARATINGLINES

CDLMATHOLD

CDL3LINESTRIKE

Doji Patterns

CDLDOJI

CDLDOJISTAR

CDLDRAGONFLYDOJI

CDLGRAVESTONEDOJI

CDLLONGLEGGEDDOJI

Multi-Candle

CDL3WHITESOLDIERS

CDL3BLACKCROWS

CDL3INSIDE

CDL3OUTSIDE

CDLABANDONEDBABY

CDLKICKING

Other

CDLSPINNINGTOP

CDLHIGHWAVE

CDLBELTHOLD

CDLCOUNTERATTACK

💡 Pattern Detection Best Practices

For Python/Backtesting

1.Use TA-Lib for production - fastest and most reliable
2.Combine multiple patterns for stronger signals
3.Filter patterns by trend direction
4.Add support/resistance context
5.Backtest before live trading

For MT5 EAs

1.Always check bar 1 (completed candle), not bar 0
2.Use tolerances for pattern detection
3.Add trend filters (moving averages)
4.Consider volume confirmation
5.Test on Strategy Tester first